Citation
Dynamics of the US sugar and HFCS markets

Material Information

Title:
Dynamics of the US sugar and HFCS markets implications for the CARICOM sugar producers within the framework of the proposed FTAA
Creator:
Evans, Edward A
Publication Date:
Language:
English
Physical Description:
xi, 147 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Derived demand ( jstor )
Elasticity of demand ( jstor )
Liquids ( jstor )
Market prices ( jstor )
Mathematical variables ( jstor )
Prices ( jstor )
Sugars ( jstor )
Sweeteners ( jstor )
Table sugars ( jstor )
United States Sugar Program ( jstor )
Dissertations, Academic -- Food and Resource Economics -- UF ( lcsh )
Food and Resource Economics thesis, Ph. D ( lcsh )
Sugar trade -- Caribbean Area ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1999.
Bibliography:
Includes bibliographical references (leaves 140-146).
General Note:
Printout.
General Note:
Vita.
Statement of Responsibility:
by Edward Anthony Evans.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
021557106 ( ALEPH )
43688651 ( OCLC )

Downloads

This item has the following downloads:


Full Text










DYNAMICS OF THE US SUGAR AND HFCS MARKETS:
IMPLICATIONS FOR THE CARICOM SUGAR PRODUCERS WITHIN THE
FRAMEWORK OF THE PROPOSED FTAA
















By

EDWARD ANTHONY EVANS













A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

1999

























To Sabrina, Latoya, Akeem, Attli and ma dodo















ACKNOWLEDGMENTS

I would like to express my profound gratitude to Dr. Carlton Davis, chairman of

my supervisory committee, for all his help and encouragement throughout the course of

this study. Dr. Davis' patience, understanding, guidance, assistance and constructive

criticisms given to me during the preparation of this dissertation are greatly appreciated.

His desire for excellence, and steadfast refusal to lower the bar kept my shoulders to the

wheel and has contributed immensely to my education and self-development.

Drs. Max Langham and Ron Ward, members of my committee, deserve special

mention. I have benefited a great deal from Dr. Langham's criticisms and vast experience,

and even though we did not always agree, I am often reminded that the beauty of a choir

comes from the fact that not everyone sings the same melody. Dr. Ward has provided

useful suggestions and has welcomed me into his home on several occasions during the

night.

To the other members of my committee, Drs. R. Emerson, A. Schmitz and D.

Denslow I am eternally grateful. Without their criticisms and useful suggestions this

effort would not have been possible. Special thanks is also due to Drs C. Moss and R.

Kilmer who made worthwhile contributions in the initial stages of the dissertation.

I gratefully acknowledge the financial support I received from the Department of

Food and Resource Economics. In this regard I must again say thanks to Dr. Davis and

will always remember the words he spoke which jolted me into finally accepting his

offer--"there are more dogs than bones." I would also like to acknowledge the stipend I


iii








received from the Caribbean Agricultural Research and Development Institute (CARDI)

during my initial years. Special mention must be made of Mr. Hugh Saul whose faith in

me and timely intervention made it possible for me to receive the latter financial support

and take up the offer at the University of Florida.

Perhaps my greatest depth of gratitude however, is owed to my wife, Sabrina,

children, Latoya and Akeem, my mother and my godfather, Attli, for their encouragement

and for the sacrifice they have made in favor of this effort. To them this dissertation is

gratefully dedicated.

Finally, I wish to express my sincere gratitude to all those who stood in my corer

cheering my every move. Among them are, my dear friend Dr. Kenrick Jordan, my most

ardent supporter, Dr. Pauline Lawrence, Dr. Curtis McIntosh, Mr. Karl James, Greta and

Danny Roberts.


























iv















TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...........................................................................................iii

LIST OF TABLES .................................................................................................viii

LIST OF FIGURES .................................................................................................. ix

ABSTRACT.............................................................................................................. x

CHAPTERS

1 INTRODUCTION.................................... ............................................................

Statement of Problem................................... ....................................................... 3
Problematic Situation.....................................................................................3
Specific Problem............................................................................................7
Hypotheses ....... ........................................................................................... 10
Objectives......................................... ................................................................ 11
M ethodology ..................................................................................................... 12
Organization of the Study ................................................................................... 13

2 BACKGROUND................................................................................................. 14

Free Trade Area of the Americas (FTAA)........................................................ 14
Review of World and Western Hemisphere Sugar Markets.................................. 17
World Production and Market Trends--Sugar............................... .............. 17
Western Hemisphere Production and Market Trends--Sugar ............................21
Importance of the Sugar Industry to CARICOM Sugar Producers............................ 23
Socioeconomic Importance.......................................................................23
The EU/Lome ACP Agreement....................................................................25
The US Sugar Program and Policy...................................................................27
Review of World and Western Hemisphere HFCS Markets ...................................31
World Production and Market Trends--HFCS ....................................................33
Western Hemisphere Production and Market Trends--HFCS.............................. 34

3 LITERATURE REVIEW AND THEORETICAL BACKGROUND ......................... 39

Previous Empirical Studies.................................................................................... 39
Sugar and HFCS Related Studies ....................................................................39


v









Evaluation and Critique ................................................................................46
Time Series Analysis......................................................................................47
Serial Correlation.........................................................................................48
Stationarity .................................................................................................. 49
Cointegration and Error Correction Models.....................................................51
Time-Varying Parameters...............................................................................55
The State-Space Framework............................................................................... 57
The State-Space Model.............................................................................. 57
Kalman Filter...............................................................................................61
Advantages of State-Space Approach and the Kalman Filter............................ 64
Relevance of the State-Space Model and Kalman Filter to the Current Study.....66
The Concept of a Derived Demand ..................................................................... 67

4 CONCEPTUAL MODEL OF US SUGAR AND HFCS MARKETS....................... 71

The US Sugar and HFCS Subsectors ....................................................................72
The US Sugar Subsector ............................................................................ 73
The US HFCS Subsector............................................................................78
The Interrelationship between the US Sugar and HFCS Markets..................86
Trade Implications of Changes in the US Sugar Program within an FTAA ..............90
Free Trade Area.........................................................................................90
Scenario I- Partial Liberalization of the US Sugar Program................................92
Trade implications................................................................................92
Welfare implications............................................................................ 96
Scenario II- Complete Liberalization of the US Sugar Program..........................98
Trade implications..................................................................................98
Welfare implications...................................................................... 101

5 EMPIRICAL MODEL OF THE US DEMAND FOR HFCS ................................. 103

Em pirical M odel.................................................................................................... 103
Econometric and Statistical Issues ......................................................................... 106
Description of Data.......................................................................................... 110
Estimation Results ....................................................... .................................. 11
M odel Implications.......................................................................................... 113
Intercept Adjustment Over Time ()................................................................ 114
Own-Price Elasticity ( t)......................................................................... . ... . 116
Cross-Price Elasticity (p2t) ......................................................................... 118

6 SUMMARY, POLICY RECOMMENDATIONS AND SUGGESTIONS FOR
FURTHER RESEARCH ...................................................................................... 121

Summ ary and Conclusions..................................................................................... 121
Theoretical Findings ........................................................................................ 125
Em pirical Findings........................................................................................... 128
Policy Implications and Recommendations............................................................ 131
Limitations of the Study and Suggestions for Further Research............................ 135


vi









APPENDIX

STATE AND SM OOTEHED ESTIM ATES ............................................................... 138

REFERENCES............................................................................................................ 140

BIOGRAPHICAL SKETCH ....................................................................................... 147














































vii















LIST OF TABLES



Table page

Table 2.1. Ranking of Countries of the Western Hemisphere on Basis of Population,
Land Size and GDP.............. .................................. .................................... 16

Table 2.2. Leading World Sugar Producers and Consumers, 1997-98 ............................. 19

Table 2.3. Leading World Sugar Exporters and Importers, 1997-98 .................................20

Table 2.4. Leading Hemispheric Sugar Producers and Consumers, 1997-98 ....................22

Table 2.5. Leading Hemispheric Sugar Exporters and Importers, 1997-98.........................23

Table 2.6. Socioeconomic Importance of CARICOM Sugar Industries.............................. 24

Table 2.7. CARICOM/EU Market Real Earnings and Quota Rent (1992=100) .................26

Table 2.8. CARICOM/US Market Real Earnings and Quota Rent (1992=100) ................28

Table 2.9. World HFCS Production, 1989-97 ('000 metric tons dry basis) .......................34

Table 2.10. Changes in US Per Capita Sweetener Consumption and Relative Market
Shares, (1975 - 1997) Pounds, Dry Weight Basis...........................................36

Table 5.1. Estimation of Price Equation............................................................................. 112

Table 5.2 Final Estimates Using Kalman Filter Over the Period 1977-98...................... 112














viii









LIST OF FIGURES



Figure pge

Fig. 2.1. World and US refined and raw real sugar prices, 1985-98 (1992=100) ................ 18

Fig. 2.2. Changes in US per capita consumption of selected sweeteners, 1975-97.............36

Fig. 2.3. Real HFCS and refined sugar prices, 1982-98 (1992=100)...................................37

Fig. 4.1. Flowchart of the US HFCS and Sugar Subsectors............................ ............ 72

Fig. 4.2. Hypothetical US demand for crystallized sugar (CS) and liquid sugar (LS),
respectively...................................................................................................... 74

Fig. 4.3. Hypothetical US aggregate demand for sugar ..............................................76

Fig. 4.4. Hypothetical US supply and demand for sugar.............................................77

Fig. 4.5. Hypothetical US demand for HFCS.............................................. ............... 79

Fig. 4.6. Shift in hypothetical HFCS short-run marginal cost curve..................................82

Fig. 4.7. Hypothetical US HFCS short-run marginal cost and demand configuration.........83

Fig. 4.8. Hypothetical HFCS industry average costs, reflecting external economies...........86

Fig. 4.9. Hypothetical supply and demand for sugar, and hypothetical demand for HFCS,
respectively......................................................................................................88

Fig. 4.10. Trade implications under partial liberalization of US sugar program.................93

Fig. 4.11. Hypothetical HFCS industry long-run demand and average cost...................... 95

Fig. 4.12. Trade implications under completely liberalized US sugar program.................. 100

Fig. 5.1. Dynamic path of HFCS intercept over time, 1982-98................................. 115

Fig. 5.2. Dynamic path of adjustment in HFCS own-price elasticity, 1982-98.................. 117

Fig. 5.3. Dynamic path of adjustment of cross-price elasticity of HFCS with respect to
sugar, 1982-98 ....................................................................................................... 119







ix














Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

DYNAMICS OF THE US SUGAR AND HFCS MARKETS:
IMPLICATIONS FOR THE CARICOM SUGAR PRODUCERS WITHIN THE
FRAMEWORK OF THE PROPOSED FTAA

By

Edward Anthony Evans

December 1999

Chairman: Carlton G. Davis
Major Department: Food and Resource Economics

To cope with growing regionalism, many developing countries must choose

between alternative and competing integration strategies. The Caribbean Community

(CARICOM) is faced with such a dilemma. The United States and the European Union

have proposed the formation of separate free trade areas (FTAs) with the subregion with

a tentative implementation date of 2005. The consensus emerging in CARICOM favors

participation in the proposed Free Trade Area of the Americas (FTAA)--a potential

grouping of 34 disparate countries in the Western Hemisphere. While the anticipated

benefits of such a group are unquestionable, there will be losses at the sector and industry

levels.

The CARICOM sugar industry is of considerable importance to the subregion but

could be adversely affected by the FTAA. First, it is possible that CARICOM could

violate the Lome Agreement through FTAA membership, denying continued access to

the lucrative EU sugar market. Second, CARICOM sugar producers could be forced to


x








compete with larger and more efficient sugar-producing FTAA member countries--such

as Brazil, Colombia and Guatemala--for a share of the regional sugar market. Third, and

the focus of this dissertation, a substantial portion of the regional sugar market might be

eroded as a result of the successful penetration of high fructose corn syrup (HFCS)

produced in the United States.

A qualitative assessment of the potential economic impact of changes to the US

sugar program and an expanded HFCS industry on the CARICOM sugar industry was

conducted. First, a conceptual model of the US sugar and HFCS markets was developed,

and potential trade and economic impacts were discussed. Second, an empirical model of

the demand for HFCS in the United States was constructed. To capture the dynamics

within the industry a state-space representation utilizing the Kalman filter was employed.

The maturity of the HFCS industry was assessed, and implications for the various sugar

concerns were drawn.

The findings substantiated the view that CARICOM membership in an FTAA

could have potential negative economic impact on the CARICOM sugar industry and that

an FTAA could further depress, rather than increase, the world market sugar price.

Finally, it was concluded that the US HFCS industry is entering a mature phase, forcing

the need for corn refiners to actively seek external markets for their products.














xi














CHAPTER 1
INTRODUCTION

The global economic environment is in a state of rapid transition. The new

dispensation in both the developed and developing countries is the opening up of national

and regional economies to market forces. In the developed countries, the principal forces

leading to trade liberalization stemmed from: 1) the growing budgetary pressure of

having to maintain protective regimes; 2) growing regionalism; and 3) multilateral

negotiations such as the General Agreement on Tariff and Trade (GATT/WTO).

Developing countries, on the other hand, have tended to liberalize their trading regimes

either unilaterally or as a result of following the dictates of multilateral financial and

developmental institutions (Wint, 1997). Whatever the reason, the result has been a

dramatic reduction in trade barriers.

For developing countries, the wide spread process of trade liberalization and

globalization is only one element of the new global dispensation to which they must

respond. An important related element is the erosion of non-reciprocal preferential

treatment under the Generalized System of Preferences (GSP) or other preferential

regimes such as Lome', which in many cases has been a mainstay of several of the

economies. At the same time, LDCs excluded from regional blocs--as in the case for the


' Lome Convention is a contractual non-reciprocal trading arrangement between African, Caribbean and
Pacific (ACP) countries and The European Union (EU). The first Lome Convention (Lome 1) was
established in 1975 and the Convention is now in its fourth round (Lome 4). It expires in the year 2000 and
is currently being re-negotiated albeit, under new philosophical circumstances.




1






2

Caribbean and some of their Latin American neighbors vis-A-vis NAFTA--are likely to

face losses due to trade and investment diversions, higher non-tariff barriers (NTBs) and

possibly higher tariffs (Blanco, 1997).

As regionalism grows, many developing countries must choose among alternative

integration strategies. Accordingly, they are reassessing existing economic and trading

arrangements as well as considering the prospects of forming new relations and strategic

alliances so as to preserve, consolidate or establish stronger trading positions. The

Caribbean Community (CARICOM)2--a group of 14 countries in the Western

Hemisphere--is one such alliance of developing countries. With a population of

approximately 6.0 million and a total gross domestic product (GDP) at market prices of

only about US $14.0 billion, these countries represent a very small market by world

standards (Bryan, 1995). In addition to the Lome arrangement, these countries have

benefited a great deal from other marketing arrangements such as Caribbean and Canada

Agreement (CARIBCAN), Caribbean Basin Initiative (CBI), and GSP. Both the US and

the EU, however, have now separately proposed forming Free Trade Areas (FTAs) with

CARICOM with a tentative implementation date being the year 2005 or there about.

While details of such arrangements are still being refined, they will have far reaching

consequences for CARICOM's relationships with Europe, the other ACP countries, the

Americas and much of the developing world. Within CARICOM, these arrangements





2 Member countries of the CARICOM include: Antigua and Barbuda, The Bahamas, Barbados, Belize,
Dominica, Grenada, Guyana, Jamaica, Montserrat, St. Kitts/Nevis, St. Lucia, St. Vincent and the
Grenadines, Suriname, and Trinidad and Tobago. Haiti has formalized arrangements for full membership as
of August 1999 bringing the total membership to 15. However, in view of that country's recent accession,
this study does not include it in the analysis.






3

will change how the subregion conducts its business with the rest of the world and how it

earns its foreign exchange.

Faced with a set of perplexing choices, the CARICOM countries have begun to

systematically weigh the benefits and costs of embarking on one of the following three

strategies: 1) becoming a part of a Free Trade Area of the Americas (FTAA) with the US

as the central figure; 2) seeking to form a Free Trade Area with the EU as part of the new

Lome arrangement; and 3) remaining outside of the formation of any free trade area with

the US or the EU. The emerging consensus favors (1). Irrespective of which strategy is

chosen, there will be serious internal economic conflicts because of gains and losses at

the sectoral and industrial levels. Moreover, the decisions made could have irreversible

economic consequences.

The CARICOM sugar industry, a major contributor to the economic and social

stability of the subregion, is undoubtedly one of the more important industries that will be

affected as the new economic arrangements emerge. Within the framework of the

proposed FTAA, for example, the industry is likely to be faced with the additional

economic challenges associated with significant structural changes that are occurring

within the US--one of its major sugar markets. In particular, this challenge could come as

a result of the rapid development of the US High Fructose Corn Syrup (HFCS) Industry

and its expansion in a much freer trading environment.



Statement of Problem

Problematic Situation

The CARICOM countries are characterized as having certain features which,

when taken in combination, makes them extremely vulnerable. These features include






4

having: 1) small economies with a relatively high degree of openness as reflected in

trade-to-output ratios; 2) undiversified sectoral and general economic structures; 3)

export concentrations in one or two products; 4) small firms with limited opportunity to

take advantage of economies of scale; 5) a high degree of dependency on non-reciprocal

trading arrangements for historical reasons; and 6) a limited-resource based sugar

industry with technologies heavily oriented toward labor intensity and high cost

productions structures (West Indian Commission, 1993).

As indicated earlier, these countries have benefited appreciably from the

preferences they enjoyed in various marketing and trading arrangements such as the

EU/ACP Sugar Program and the US Sugar Program. A consequence has been a degree of

complacency and technical obsolescence on the part of regional producers that have led

in part to an appreciable reduction in the levels of production and marketing efficiencies3.

A recent study conducted by LMC International and reported by the Economic Research

Service (ERS) of the United States Department of Agriculture (USDA) indicated that

none of the CARICOM countries can be regarded as low cost sugar producers. Barbados

and Trinidad and Tobago were among the highest cost producers in the world. The

remaining four CARICOM producers (Belize, Guyana, Jamaica and St. Kitts/Nevis) were

classified as medium cost producers (Haley, 1998a). Production cost in Jamaica is

reported to be in the vicinity of 33 cents per pound, compared to the US and Brazil prices

of 22 and 14 cents, respectively (Caribbean Update, 1998; USDA, 1998). To complicate


3 This point has been conceded by representatives of the Sugar Association of the Caribbean (SAC)-- the
official producer association. They pointed out however, that while current initiatives are being taken to
modernize the industry and improve both field and factory levels efficiencies, efforts to attract much
needed capital investments are being hampered by the uncertainty which surrounds the future of the
subregion's industry. They have also indicated that levels of efficiencies that can be obtained will not match
some of those reported in parts of the US due to their economies of scale (James, 1999 Personal Interview).






5

matters further, a considerable amount of the land used in the subregion to produce sugar

cane is marginal with respect to sugar cane production but has few viable alternatives.

Also, from a wider socioeconomic point of view, labor and capital could become socially

disruptive in these small undiversified economies if other alternatives were pursued too

quickly (Haraksingh, 1997).

Notwithstanding these drawbacks, the industry remains of considerable import to

the subregion, provides much of the valuable foreign exchange earned, and is a major

contributor to the employment of labor. For example, the industry provides over one half

of the foreign exchange earnings in St. Kitts and Nevis and employs about 97 thousand

persons or about 5 % of the active working population within CARICOM (James, 1998).

Moreover, the socioeconomic importance of sugar production to these countries goes

much further than the above figures might suggest as many of these industries do play a

vital role in the maintenance and upgrading of rural infrastructure, health, recreational

and educational facilities.

With the general consensus that CARICOM should seek to become part of the

proposed FTAA, there has been a genuine concern with regards to the future of its sugar

industry (Group of Experts, 1998). Such a move toward integrating the economies brings

with it the challenge of competing in a liberalized trading environment without the

benefit of preferences to which the subregion has grown accustomed.

The following three interrelated issues with respect to the subregional sugar

industry are of importance. First, there is the possibility that should the subregion pursue

such a path it could be in violation of the terms and conditions of its Lome Agreement.

Specifically, article 174 section 2a of the Agreement forbid recipients to enter into any






6

economic arrangement with any developed countries which would result in such

countries receiving terms and conditions more favorable than those given to the EU

(Group of Experts, 1998). As indicated earlier, more than three-quarters of the

subregion's exports of sugar goes to the EU market where producers receive a high

enough price to cover the relatively high cost of production. Becoming part of the FTAA

could therefore mean that CARICOM might either have to sacrifice its preferential

market in the EU or at best offer the EU similar terms and conditions as those given to

the US and Canada, the developed members of the proposed FTAA. Resolving this issue

within the framework of an FTAA might not be as simple as it seems in view of the fact

that the US took a similar stance, within the framework of its CBI initiative, to that taken

by the EU in their Lome Agreement.

Second, an FTAA, which includes the subregion, would place the CARICOM

sugar industry into direct competition with the larger and more efficient producers in the

hemisphere. The study conducted by LMC, referred to above, notes that three of the

world most efficient sugar producers, Brazil, Columbia and Guatemala are in the

hemisphere. These low-cost producers could easily supply the sugar requirements of the

other members of the FTAA.

Third, is the fact that the US while being the largest sugar importing country

within the hemisphere is also a major producer of all the principal sweeteners--cane

sugar, beet sugar, corn sweeteners and several non-caloric sweeteners. Its sugar and

sweetener industry is reported to be the largest in the world, with annual consumption of

caloric sweeteners approaching 20 million short tons (Lord, 1995). Moreover, with the

technological advances in the corn sweetener industry, the United States has become the






7

world's lowest-cost producer of sweeteners. The world average cost of production is

20.91 cents per pound. It is 13.73 cents per pound in the most efficient sugar producing

countries. In comparison the US produces HFCS in 1994/95 at an average cost of only

10.60 cents per pound dry weight equivalence (Haley, 1998a). HFCS competes directly

with sugar in most industrial uses. In a situation of no foreign tariffs, duties, taxes,

quotas, and other border barriers against corn sweetener imports, the US could export its

technology and/or corn sweetener product to some of the higher-cost markets within the

FTAA. The US would also benefit as the world's largest and cheapest producer of corn--

the main input used in the production of HFCS. The potential of HFCS to displace sugar

has already been seen in the US and to a lesser degree in Canada. And, though it has not

yet adversely affected the other sugar producing and consuming countries, those in the

planned FTAA are likely to feel its impact.



Specific Problem

This study is concerned with the threat that an expanded HFCS industry could

pose to the CARICOM sugar industry within the framework of a liberalized trading

environment. The US sugar industry and its sugar program have been adequately dealt

with in the literature. The same is not true in the case of the HFCS industry. Not much

has been written on the likely impact that an expanded HFCS industry could have on

world and regional prices, production, consumption and trade patterns of sugar. Also, the

dynamics of the US sugar-HFCS interrelationships and the implications of such

relationships for trade in an extended market have not been fully explored. Thirdly, as

noted by Tanyeri-Abur (1990) there appears to be an absence of studies to form a basis

for studying the trade implications of an expanded HFCS industry. As a consequence,






8

studies of production and consumption of the commodity have been largely confined to

the domestic market. With the advancements in HFCS technology on both the supply and

demand sides and the reduction in trading barriers in the international arena, the prospects

of this commodity assuming a greater level of importance in trade have increased

considerably.

One aspect of the HFCS market is the dual nature of the product, whereby within

a given range of the sweetener demand curve it substitutes almost perfectly for sugar but

within another range it does not. Either of the sweeteners (HFCS or sugar) can be used in

soft drinks, but for some other uses that require a crystallized sweetener, HFCS is not a

viable substitute. This dual nature of the substitutability relationship has implications for

the extent to which the commodity will be traded internationally, the shape of the

sweetener demand curve and the econometric approach to be used to estimate HFCS

demand parameters.

Another related aspect is the issue of the extent to which the industry has reached

a level of maturity. Thomas (1985), pointed out that HFCS when considered by industrial

standard is regarded as a relatively new product. Among other things, he noted that this

implies that there will almost certainly be further technological improvements and that

the behavior of the industry can be expected to be different from that of a mature industry

such as sugar. If this is indeed the case, then there are implications with regard to the

econometric approach used to estimate the important economic parameters. A cursory

investigation into the history of the industry seems to support the point made by Thomas.

On both the supply and demand sides of the market advances in technology have resulted

in noticeable development of the product and product usage. For example, the price of






9

HFCS since its inception has been lower than that of sugar on a sweetness equivalency

basis. Nevertheless, the adoption of this product had been cautious. It took a decade for

the major soft drink industries to convert from liquid sugar to HFCS. On the demand side

the adoption of the product has been aided by the producers of HFCS working in concert

with the users, providing technical assistance and adapting the product to the special

requirements of the various clients. It is in this context that HFCS-55 was developed as a

more effective substitute for liquid sugar than HFCS-42 in the soft drink industry.

Likewise on the supply side there has been a constant effort aimed at improving the

technology involved in the production of the commodity, with a view to producing an

effective substitute sweetener at the lowest cost. The implication is that the industry is

evolving and as such the parameters of the industry might be changing. This time-varying

issue must be recognized in characterizing the industry in any analytical model.

Ward and Tilley (1980) pointed out that in markets, such as the one we are

concerned with where significant adjustments are occurring, the parameters are likely to

be time-varying. They noted that technological changes result in structural changes in

economic phenomenon, giving rise to varying parameters. They cautioned that a failure

to take into consideration such variations in the estimated parameters could cause serious

errors in projections and policies drawn from them.

This issue of the degree to which the HFCS industry in the US can be considered

mature is therefore important from an estimation point of view. It is also important in the

context of this study since the dynamic paths of adjustments of various economic

parameters have such implications for the CARICOM sugar industry such as: 1) the

extent of saturation of HFCS in the US market and by extension the size of the US






10

domestic sugar requirement; and 2) likely rent-seeking behavior on the part of the corn

refiners.

From the above it is clear that any assessment of the likely impact of an FTAA on

the CARICOM sugar industry must include a thorough understanding of developments

occurring within the US sweeteners industry. To focus only on sugar per se would be to

overlook an important component of the dynamics occurring within the region's

sweeteners market. Understanding the dynamics between HFCS and sugar and generating

reliable parameters for the HFCS industry, are required for a meaningful assessment of

the manner in which the HFCS industry might respond in a freer trading environment.



Hypotheses

1. The parameters of the supply and demand configuration of the HFCS industry

have been varying over time because of the relative "newness" of the HFCS

industry and the influence of both scale and technological factors.

2. The market for HFCS is approaching an asymptotic level, giving rise to the need

to seek external markets for the product, and a basis for a new round of rent

seeking by corn refiners.

3. If the FTAA becomes a reality, the US could become a major exporter of HFCS

in the near future with a sizable proportion of the product going to the countries in

the regional trading blocs.

4. If HFCS remains competitive and can be substituted for sucrose causing a

structural shift in the demand and supply of sugar, there will be an excess supply

of sugar in the FTAA resulting in depressed regional and world prices.






11

5. Given the above, membership of CARICOM countries in the FTAA will in the

short to medium term have potentially negative economic impacts on the sugar

industry of the subregion.



Objectives

The study has two general objectives. Firstly, to qualitatively assess the potential

economic impact that likely changes to the US sugar program, and an expanded High

Fructose Corn Syrup (HFCS) industry, might have on the CARICOM sugar industry

within the framework of the proposed Free Trade Area of the Americas (FTAA).

Secondly, to provide quantitative estimates for the HFCS demand configuration that can

be used in subsequent modeling exercises. Specific objectives are to:

1. Assess selected trends in the US sweeteners industry.

2. Develop a conceptual model of the US sugar and HFCS markets.

3. Qualitatively assess the trade and welfare implications of likely changes in the US

sugar program and an expanded HFCS industry on the sugar producing countries

of the FTAA, with specific reference to the CARICOM sugar producers.

4. Develop and estimate the derived demand for HFCS in the US sweeteners market.

5. Investigate the demand side dynamics within the HFCS industry, and

6. Identify critical policies regional planners might wish to take into consideration in

their ongoing integration negotiations, as they relate to the future of the

CARICOM sugar industry.






12

Methodology

The purpose of this study is to assess the potential economic impact that likely

changes to the US sugar program, and an expanded High Fructose Corn Syrup (HFCS)

industry, might have on the CARICOM sugar industry within the framework of the

proposed Free Trade Area of the Americas (FTAA). This will be accomplished in two

stages. First, a conceptual model of the US sugar and HFCS markets will be constructed.

Although constructed within a static framework, the model will attempt to highlight some

of the dynamics between sugar and HFCS. Based on the model, the implications of

possible changes in the US sugar program, and the impact of an expanded HFCS industry

on non-US sugar producers will be discussed.

The second stage of the exercise will be to empirically model the demand for

HFCS in the US with a view of assessing the maturity of the industry, and drawing

implications for the various sugar concerns. The focus will be on the dynamic paths of

adjustments of various parameters of the estimated demand function. To facilitate such

an investigation a state-space modeling approach and the Kalman filter will be employed.

Among the advantages of utilizing the state-space form and the Kalman filter are that: 1)

it allows for modeling of time-varying parameters, if the situation so warrants, and can

provide an estimate at any point within the sample and hence a valuable way of

examining how a trend might have evolved; 2) it can deal fairly easily with nonstationary

time series which traditional econometric methods have difficulty in handling; and 3) it

allows modeling to be done in level form (rather than differenced) in situation where the

data is nonstationary thus facilitating interpretation of results. The flexibility of the state-

space model thus makes it an appropriate tool for estimation in situations, such as the






13

demand for HFCS, in which the phenomenon being considered might be nonstationary

and still evolving.



Organization of the Study

This dissertation consists of six chapters. Following this introduction, chapter 2

provides background information on the trends and developments and the institutional

settings of the sugar and HFCS markets. It therefore places the study within the wider

context, highlighting the complex nature of the issue on hand and the varied interests.

Chapter 3 begins with a review of the literature, focusing on studies that have attempted

to include the HFCS industry in their modeling exercises. This is followed by a review of

the theoretical background to some of the issues, which are considered and used in the

subsequent chapters. In particular, a detailed overview of the state-space model and

Kalman filter is provided. In Chapter 4 a conceptual model is constructed of the US sugar

and HFCS subsector. The model is then used to examine a few scenarios within the

framework of the proposed FTAA. Implications of the results are drawn for the non-US

sugar producers, in particular the CARICOM producers. Chapter 5 takes the analysis a

step further and presents the two-stage estimation of the derived demand for HFCS in the

US. The first section of the chapter focuses on the specification of the model while the

latter section presents and discusses the implications of the results. The final chapter

summarizes the conclusions from the research and makes suggestions with regard to

some of the critical policies that political and regional planners might wish to take into

consideration in their ongoing integration negotiations as they relate to the future of the

CARICOM sugar industry.














CHAPTER 2
BACKGROUND

The purpose of this chapter is to provide background information on key players

and institutional arrangements involved with sugar and HFCS. The chapter contains a

brief description of the proposal to form a Free Trade Area of the Americas and

highlights some of the disparities among the potential members. This is followed by a

review of world and Western Hemispheric trends in the production and marketing of

sugar. Next is a discussion of the importance of the sugar industry to CARICOM. Key

marketing arrangements such as the EU/ACP Sugar Protocol and US Sugar Program are

examined. Implications of the GATT/WTO rules are also briefly examined as they relate

to these special marketing arrangements. The final section of the chapter focuses on the

HFCS and reviews world and Western Hemisphere production and marketing trends for

this commodity.



Free Trade Area of the Americas (FTAA)

A Free Trade Area of the Americas (FTAA) is one of several trade agreements

that the United States is encouraging and actively pursuing. The decision to establish an

FTAA and to conclude the negotiation for its implementation by the year 2005 was made

at the Summit of the Americas held in Miami, Florida in December 1994. Among other

things, the Plan of Action calls for the promotion of prosperity among the 34

participating countries through economic integration and free trade, the eradication of

poverty and discrimination, and the guaranteeing of sustainable development by the


14






15

conservation of natural environment for future generations. The process has five

elements: 1) liberalization of barriers to trade in goods and services; 2) elimination of

restraints on investments; 3) provisions of free labor movement for specialized workers;

4) harmonization of tax and monetary policies; and 5) establishment of supra regional

institutions to administer the arrangements and to engage in dispute resolution. Bryan

(1995) notes that the new integration process goes beyond the traditional concept of

countries simply extending reciprocal preferences to their trading partners. He cautions,

however, that while absolute reciprocity may be the ultimate goal of the hemispheric

trade liberalization, many of the smaller Caribbean economies simply cannot compete

with countries having more productive infrastructures and technologies. Stated

differently, it might not be possible for them to offer absolute reciprocity to industrial

countries at least in the short run, and there will be the need for some compromise on the

part of the more developed trade partners (Bryan, 1995).

To better appreciate some of the disparities among the countries, Table 2.1 shows

a relative ranking of the 34 countries that will comprise the FTAA with respect to the

selected indicators of population, land area and GDP. In general, the CARICOM

countries (underlined in Table) occupy the lower levels in all three cases. The subregion

accounts for only 0.67% of the Hemisphere's land area, 0.82% of its population and

approximately 0.2% of total GDP. The Table also reveals that the most populous and

highest income country is the United States and the least populous and lowest income

country is St. Kitts and Nevis. Barbados has the smallest land area while Canada has the

largest. What is not obvious from the Table however, is the fact that many






16


Table 2.1. Ranking of Countries of the Western Hemisphere on Basis of Population,
Land Size and GDP

Rank Population (Millio Land Size ('000 sq. GDP (US$ M
1995 ns) Km.) 1995 1990=100)

34 St. Kitts & Nevis* 0.04 Barbados 0.3 St. Kitts & Nevis 150
33 Antigua & Barbuda 0.06 Grenada 0.3 Dominica 178
32 Dominica 0.07 St. Kitts & Nevis 0.3 St. Vin. and Gren. 218
31 Grenada 0.09 Antigua & Barbuda 0.4 Grenada 229
30 St. Vin. and Gren. 0.11 St. Vin. and Gren. 0.4 Suriname 334
29 St. Lucia 0.14 St. Lucia 0.6 Antigua & Barbuda 366
28 Belize* 0.21 Dominica 0.7 Belize 491
27 Barbados* 0.26 Trinidad & Tobago 5.0 St. Lucia 509
26 Bahamas 0.27 Jamaica 11.0 Guyana 602
25 Suriname 0.42 Bahamas 14.0 Haiti 1,642
24 Guyana* 0.82 El Salvador 21.0 Barbados 1,717
23 Trinidad & Tobago* 1.26 Belize 23.0 Nicaragua 2,590
22 Jamaica* 2.53 Haiti 28.0 Bahamas 3,053
21 Panama* 2.63 Dom. Republic 49.0 Honduras 3,378
20 Uruguay* 3.19 Costa Rica 51.0 Jamaica 4,171
19 Costa Rica* 3.07 Panama 76.0 Trinidad & Tobago 5,707
18 Nicaragua* 4.54 Guatemala 109.0 Bolivia 6,496
17 Paraguay* 4.83 Honduras 112.0 Panama 6,570
16 El Salvador* 5.64 Nicaragua 130.0 El Salvador 6,674
15 Honduras* 5.95 Suriname 163.0 Costa Rica 7,027
14 Haiti* 7.18 Uruguay 177.0 Paraguay 7,177
13 Dom. Republic* 7.91 Guyana 215.0 Dom. Republic 7,341
12 Bolivia* 8.06 Ecuador 284.0 Guatemala 9,706
11 Guatemala* 10.62 Paraguay 407.0 Uruguay 11,431
10 Ecuador* 11.46 Chile 757.0 Ecuador 15,132
9 Chile 14.20 Venezuela 912.0 Peru 47,618
8 Venezuela 21.64 Bolivia 1,099.0 Chile 48,326
7 Peru* 23.53 Colombia 1,139.0 Colombia 56,379
6 Canada*t 29.61 Peru 1,285.0 Venezuela 64,980
5 Argentina*t 34.59 Mexico 1,958.0 Argentina 196,949
4 Colombia* 35.10 Argentina 2,767.0 Mexico 250,936
3 Mexico*t 94.78 Brazil 8,512.0 Brazil 432,433
2 Brazil* 159.22 United States 9,809.0 Canada 608,658
1 United States*t 263.06 Canada 9,976.0 United States 6,173,900
* Major sugar producing and exporting countries
t Major HFCS producers
Source: Adapted from Bemal (1998)

of these CARICOM countries are relatively small, open and undiversified economies

with over one-quarter of their total exports concentrated on one or two products. In the

case of countries such as St. Kitts and Nevis, Dominica and St. Vincent and the






17

Grenadines and St. Lucia, one primary exported commodity, either raw sugar or bananas,

accounts for more than 50% of exports.



Review of World and Western Hemisphere Sugar Markets

Sugar4 is one of the few commodities that can be produced from crops grown in

both temperate and tropical climates. A consequence of this is that widespread

governmental interventions have long been a feature of the world sugar market, making it

one of the most volatile of all primary commodity markets (Borrell and Duncan, 1993).

To gain some independence from the volatile world market, producers in most countries

have lobbied their governments to operate intervention schemes aimed at controlling

domestic prices, supply, and demand. Numerous pricing tiers, and trade and stockholding

mechanisms (policy wedges) have been devised and have resulted in distorted

production, trade, consumption, and world price. The trends in real world and US market

prices of raw and refined sugars are shown in Figure 2.1 for the period 1985-98. Among

other things, Figure 2.1 reveals the relative volatility of prices in the world market

compared with those in the US market.



World Production and Market Trends--Sugar

Global sugar output in 1998-99 is projected by the US Department of Agriculture

(USDA) to increase by about 1.14 million metric tons (MMT) (0.9 percent) to 126.5

MMT, whereas consumption is projected to increase by 0.4 percent to 127.5 MMT,

reflecting some degree of market tightening. The five leading world sugar producers and


4 In this study, the word sugar refers to sucrose obtained from sugarcane and sugar beets.






18


consumers in 1997-98 are shown in Table 2.2. The European Union, followed by Brazil,

continues to dominate production. The United States is listed as the fifth largest sugar

producer, reflecting growth in domestic sugar beet production. In terms of consumption,

the leading country is India, followed by the European Union and China. The United

States occupies the fourth position in spite of its massive consumption of HFCS. As seen

from the Table, both production and consumption are relatively concentrated, with



32


28




02-


0 16







41 1

Years

Fig. 2.1. World and US refined and raw real sugar prices, 1985-98 (1992=100)
Source: Compiled with USDA sugar data and GDP deflator


the top five producing countries (out of 120 sugar-producing countries) and top five

consuming countries (out of all countries) accounting for 53 percent and 45 percent of

global production and consumption, respectively. Finally, it should be observed that

developing countries occupy three of the top five positions in both cases, reinforcing the

importance of this commodity to these economies.






19


Table 2.2. Leading World Sugar Producers and Consumers, 1997-98
Producers Consumers

Country Quantity % of World Country Quantity % of World
(MMT) Production (MMT) Consumption
European Union 19.3 15.4 India 16.7 13.1
Brazil 16.6 13.2 European Union 14.4 11.3
India 14.5 11.6 China 9.0 7.0
China 8.6 6.8 United States 8.9 7.0
United States 7.3 5.8 Brazil 8.8 6.9
World 125.4 100.0 World 126.9 100.0
Source: USDA, Sugar and Sweeteners Situation and Outlook Yearbook.


Only a relatively small proportion of the large world production of sugar is traded

each year. Schmitz (1995, p 54) notes that it is common to divide the market for sugar

into three distinct markets: 1) the domestic market within the sugar-producing countries.

This market, he contends, is the largest and accounts for about 75% of all world sugar

production; 2) the market for the various international agreements between certain

importers and certain exporters. These agreements include the import quotas under the

US program, the bilateral agreements between Cuba and the former Soviet Republics and

the agreements of various countries and groups of countries such as the ACP with the

EU. This market makes up about 10%/ of world production; and 3) the market for the

residual "free market" in world sugar, which accounts for the remaining 15% of world

production.

For the fiscal year 1997-98, approximately 35.6 MMT (28.3 percent of world

production) of sugar was traded globally (Table 2.3). The five most important sugar

exporters--Brazil, European Union, Australia, Thailand, and Cuba--accounted for 65.1

percent of global exports. Imports, on the other hand, have been less concentrated, with

the share of the top five sugar-importing countries and regions in 1997-98--Russia,

United States, European Union, Japan, and Korea--amounting to only 30.1 percent.






20


Table 2.3. Leading World Sugar Exporters and Importers, 1997-98
Exporters Importers
Country Quantity % of World Country Quantity % of World
(MMT) Exports (MMT) Imports
Brazil 7.2 20.2 Russia 3.8 10.7
European Union 6.2 17.4 United States 2.0 5.6
Australia 4.6 12.9 European Union 1.8 5.1
Thailand 2.9 8.1 Japan 1.6 4.5
Cuba 2.3 6.5 Korea 1.5 4.2
World 35.6 100.0 World 35.6 100.0
Source: USDA, Sugar and Sweeteners Situation and Outlook Yearbook.


Owing to domestic government policies, significant changes have occurred in the

status of many of the major players in the market. For example, the European Union has

changed its status from being a net sugar importer through 1976 to becoming the world's

largest exporter. In the United States, sugar imports have declined considerably from a

share of 20 percent of world imports in the 1970s to the current 5.6 percent. As noted by

Hannah and Spence (1997), the United States--which until 1975 had been the largest

sugar importer by a wide margin--had slipped to second position during the 1975-80

period and to fifth position by 1994. Japan--once the world's second largest importer of

sugar--has reduced its imports considerably, owing mainly to domestic support policy.

Table 2.3 however, does not reflect the significant change that has occurred in the

structure of the global sugar import market. The market was previously dominated by

developed countries but is now dominated by developing countries. Indeed, Hannah and

Spence (1997) reported that in 1994 the developing countries accounted for

approximately 68 percent of net imports, compared with only 25% in 1975. These

authors noted that this change in the structure of global import demand has implications

for the extent of the price volatility since, unlike the developed countries, these countries






21


have relatively low income and high price elasticities. Consequently, the change has

resulted in greater stability of world market prices.



Western Hemisphere Production and Market Trends--Sugar

Of the thirty-four countries comprising the FTAA, twenty-five are major sugar

producers/exporters. Thus in the 97/98 production year the Hemisphere produced about

39.61 MMT or about 31.6% of world production, estimated at about 125.38 MMT.

Consumption of sugar within the Hemisphere was estimated at 31.38 MMT, which

represented 24.7% of world consumption. This implies that the region5 is a net surplus

producer of sugar of approximately 8.23 MMT. If production of sugar in Cuba6 were to

be added, output would increase by another 3.0 MMT bringing the regional surplus to

10.54 MMT. Table 2.4 shows the top six sugar producing countries plus CARICOM and

the top seven consuming countries in the region together with their respective market

shares. On the production side the largest producer is Brazil with a share of 39.6%,

followed by the US and Mexico with shares of 18.4% and 13.9%, respectively. Total

CARICOM production, comprising output from Barbados, Belize, Guyana, Jamaica, St.,

Kitts and Nevis, and Trinidad and Tobago, amounted to less than one million metric ton

in the 97/98 fiscal year7. This amount represents about 1.9% of regional output and less

than 1% of world output.



5 In the document the word "region" will be used to refer to the Western Hemisphere, and the "subregion"
will be used to refer to the CARICOM countries, unless otherwise defie.
6 Cuba is excluded from the prospective countries slated to form the FTAA.

7 These countries are referred to as the CARICOM sugar quota holders and comprise the Sugar Association of
the Caribbean (SAC).






22


On the consumption side the US and Brazil were the main consumers each

accounting for about 28% of regional consumption. Next in line were Mexico, Argentina,

Columbia and Canada, respectively. The fourteen CARICOM countries consumed 0.32

MMT or about 1% of regional consumption.


Table 2.4. Leading Hemispheric Sugar Producers and Consumers, 1997-98

Producers Consumers

Country Quantity % of Country Quantity % of
(MMT) Hemispheric (MMT) Hemispheric
Production Consumption
Brazil 15.70 39.6 US 8.91 28.4
US 7.28 18.4 Brazil 8.80 28.0
Mexico 5.49 13.9 Mexico 4.24 13.5
Columbia 2.15 5.4 Argentina 1.45 4.6
Guatemala 1.79 4.5 Colombia 1.37 4.4
Argentina 1.75 4.4 Canada 1.27 4.0
CARICOMa 0.75 1.9 Peru 0.91 2.9
a] Represents the total output of six CARICOM countries
Source: International Sugar Organization


With respect to the hemispheric trade, in the 97/98 period total exports amounted

to 12.77 MMT representing about 35.9% of world exports, estimated at 35.59 MMT. In

contrast, total imports amounted to just 4.46 MMT or about 12.5% of world imports.

Consequently, the region is a net exporter of 8.3 MMT (excluding net exports from Cuba

of 2.3 MMT). Table 2.5 shows the leading sugar exporters and importers in the region.

The Table also shows exports as a percentage of domestic production and imports as a

percentage of consumption. The leading sugar exporters within the region are Brazil,

Guatemala, Mexico and Columbia, respectively. Of interest however, is the fact that the

CARICOM exports, while accounting for only 5.1% of total regional exports, represent

more than 85% of their domestic production--an indication of the importance of this

commodity in trade for these countries.






23


Table 2.5. Leading Hemispheric Sugar Exporters and Importers, 1997-98
Exporters Importers

Country Share (%) Exports as a % Country Share (%) Imports as a %
of Hemispheric of Domestic of Hemispheric of Domestic
Exportsa Production Imports Consumption
Brazil 56.4 45.9 US 44.1 22.1
Guatemala 10.7 76.0 Canada 27.3 95.7
Mexico 10.0 23.3 Peru 5.8 28.2
Columbia 6.5 38.5 Venezuela 5.4 30.8
SACb 5.1 86.5 Chile 4.9 32.4
Dominican Rep. 2.3 51.2 Ecuador 2.9 32.5
El Salvador 1.6 45.4 Uruguay 2.1 86.4
a] Include re-exports
b] Total exports from six CARICOM countries
Source: International Sugar Organization

In terms of regional imports, the United States and Canada dominated the trade in

97/98. Together these two countries accounted for in excess of 70% of the region's sugar

imports. Peru and Venezuela were the other major importers, accounting for about

11.2% of the import market. The fourteen CARICOM countries import a total of about

0.1 million metric ton. This amount represents about 2.2% of the overall regional imports

and less than a third of the total CARICOM sugar requirements. The bulk of such imports

was from the world market and in some cases was done in order that the exporting

country would be able to satisfy its sugar export quota commitments.



Importance of the Sugar Industry to CARICOM Sugar Producers



Socioeconomic Importance

The sugar industry is of considerable importance to CARICOM countries. This is

particularly true for the six members that comprise SAC. Among other things, the

industry contributes substantially to valuable foreign exchange earned by these countries.

For example, in St. Kitts and Nevis the industry provides more than one-half of that






24


country's foreign exchange earnings. In Jamaica, sales from exports of sugar were in

excess of US $100 million in 1996 (Caribbean Update, 1998). In the subregion, the

industry directly employs about 97,000 persons, or about 5 percent of the active working

population, and contributes appreciably to the Gross Domestic Product (GDP) of the

respective countries (Table 2.6). Moreover, the socioeconomic importance of sugar

production to these countries is much more than these figures might suggest since these

industries play a vital role in the maintenance and upgrading of rural infrastructure,

health, recreational, and educational facilities (James, 1998).


Table 2.6. Socioeconomic Importance of CARICOM Sugar Industries
Country Revenue Revenue Foreign Earnings % of Active Hectares
as % of GDP As % of Agric. as % of Total Work Force Cultivated
Production
Barbados 1.7 33.2 17.0 3.6 11,700
Belize 10.5 61.1 32.8 45.3 24,000
Guyana 26.0 41.0 27.7 9.6 42,000
Jamaica 2.6 29.7 8.0 4.1 45,000
St. Kitts/Nevis 49.0 85.0 64.1 12.5 4,000
Trinidad & 1.4 67.0 1.6 3.8 19,000
Tobago
Source: James 1998


Historically, sugar in the Caribbean has been marketed under preferential trading

arrangements. Currently, these arrangements are the Lome Agreement (via the Sugar

Protocol) and the US Preferential Quota (via the Sugar Import Tariff Quota). In 1997-98,

of total SAC countries' sugar output of 0.75 MMT (raw sugar) produced, 0.66 MMT

(86.5 percent) were exported. Of this amount, 79.9 percent went to the European Union,

and 11.3 percent went to the United States. The remaining 8.8 percent of sugar exports

from CARICOM were sold on the subregional and world markets, with each accounting

for an equal share. Prices obtained in both the EU and US markets were significantly

higher than the world market price (James, 1998).






25


The EU/Lome ACP Agreement

The EU provides special support to ACP countries, which includes CARICOM,

through a series of Protocols annexed to the Lome Convention. These Protocols give free

access to EU for fixed quantities of bananas, sugar, beef, veal and rum. Producers usually

receive EU prices, which, by virtue of the Union's Common Agricultural Policy (CAP),

are usually two to three times the world market price. Of all the Protocols, the Sugar

Protocol is considered unique in that it is characterized by its "indefinite period," and its

claimed "autonomy." Signatory countries benefit from a formal contractual commitment

by the EU to buy specific quantities of cane sugar at guaranteed prices. The quota is fixed

at about 1.3 MMT (of which 0.43 MMT is allocated to CARICOM) and ACP countries

are permitted to import sugar for domestic purposes if their surplus in a year is

insufficient to fill the quota. Additional preferential access was granted to ACP sugar

producers on July 1995, for a six-year period, under the Special Preferences Sugar (SPS)

quota bringing the total allocation to the CARICOM subregion to 0.53 MMT (James,

1998). Table 2.7 gives an indication of the real earnings and the quota rents from sale of

sugar to the EU market. The values are calculated on the basis of only the main quota

(does not include shipment under the SPS) and assumes SAC countries fulfill their

quotas.

Trade liberalization puts at risk the benefits of the Protocol to the signatory parties

and by implication its very existence. One problem stems from the fact that the Lome

Convention was ruled in 1994 by a GATT panel to be inconsistent with GATT rules by


8 Quota rent is used here loosely to refer to the extra revenue (surplus) earned by the exporting country--in
excess of what the country would have earned selling at the world market price-from selling in a particular
market where the price is higher than world market price because of import restrictions.






26


virtue of its non-reciprocal nature (Blanco, 1997). In particular, the Arrangement was

ruled to be discriminatory vis-i-vis other LDCs and thus not covered by the enabling

clause permitting GSP schemes. This GATT ruling led to the EU seeking a waiver,

which was subsequently granted by the WTO. However, the non-reciprocity permitted by


Table 2.7. CARICOM/EU Market Real Earnings and Quota Rent (1992=100)
Year SAC UK Price Real UK Real World Total Real Quota Rent
QUOTA (Pounds Price raw Price raw Earnings (US $M)
(MMT) /MT) (US$/MT) (US$/MT) (US $M)
90-91 0.4281 346.10 628.35 204.27 269.00 181.55
91-92 0.4281 351.12 587.79 199.98 251.64 166.03
92-93 0.4281 403.00 629.91 215.66 269.67 177.35
93-94 0.4281 400.28 587.39 252.06 251.47 143.56
94-95 0.4281 420.56 608.52 273.07 260.51 143.61
95-96 0.4281 445.58 649.58 240.39 278.09 175.18
96-97 0.4281 414.44 577.45 227.90 247.21 149.64
97-98 0.4281 364.64 512.96 177.08 219.60 143.79
Source: SAC data


the GATT/WTO waiver is contrary to Lome principles and is only valid until the end of

the present Convention in the year 2000. What this means in practical terms for the ACP

countries and CARICOM in particular, is that the Sugar and Banana protocols will have

to stand alone as either viable WTO-compatible commercial arrangements between the

EU and the nations to which they apply, or they will disappear completely. The WTO

1999 negotiations will undoubtedly include strong calls for free trade in sugar and the

demise of the Protocol.

A second factor which stands to reduce the potency of the Sugar Protocol,

irrespective of the shape or form of any future EU-ACP Agreement, is the reductions in

domestic support measures by the EU as it seeks to implement its WTO liberalization

commitments. Firstly, as the EU implements its agreed tariff reductions, access by third






27


(non-ACP) countries, presently granted Most Favored Nation (MFN)9 status, will be

enhanced thereby reducing the preference margin for exports from ACP. Secondly, as

support is reduced, particularly subsidization of exports to domestic producers in the EU,

the domestic prices paid to EU producers will begin to fall. This will in turn, trigger a

reduction in ACP prices since the Sugar Protocol guarantees that ACP quota holders will

be paid the same prices as EU producers.

Other factors which will or have begun to erode the value of the EU/ACP Sugar

Protocol include: 1) the pre-EU membership arrangements with the countries of Central

and Eastern Europe some of which are sugar producers; 2) a surge in EU centered

regional trade arrangements, in particular, preferential agreements in the pipeline with

some of the sugar producing countries in Latin America; 3) the fact that markets have

evolved significantly since the first Lome Convention in 1975, with the result that Europe

has become a leading sugar exporter; and 4) that cane sugar can be bought on the world

market at much lower prices than those guaranteed by the Protocol which now cost the

EU several hundred million Euros'0 each year. A final issue that is unresolved is the

impact of the Euro, the EUs common currency, on the sugar prices that quota holders

receive.



The US Sugar Program and Policy

In spite of the less lucrative nature of the US market relative to the EU market, the

former nonetheless has provided CARICOM sugar producers with a much more viable


9 The Most Favored Nation (MFN) clause of a commercial treaty is a binding contract by the signatories to
confer upon each other all the most favored trade concessions that either may grant to any other nation
subsequent to the signing of the agreement.






28


alternative than selling sugar on the world market. It has also permitted the subregional

sugar industry to survive and, through inter-industry economic linkages, has fostered

growth in services and other sectors. Table 2.8 gives a rough indication of the real quota

rent earned by CARICOM as a result of the US program over the period 1990-98. The

data are based on the assumption that CARICOM fulfilled its quota commitment to the

US, which is approximately 4.8 percent of the total tariff rate quota (TRQ)". Also, no


Table 2.8. CARICOM/US Market Real Earnings and Quota Rent (1992=100)
Year TOTAL SAC Real US Real World Total Real Quota Rent
TRQ TRQ Price raw Price raw Earnings (US $M)
(MMT) (MMT) (US$/MT) (US$/MT) (US $M)
90-91 2.1027 0.1051 487.41 204.27 51.24 29.77
91-92 1.3863 0.0693 468.82 199.98 32.50 18.63
92-93 1.1347 0.0567 464.86 215.66 26.37 14.14
93-94 1.1347 0.0567 458.00 252.06 25.98 11.68
94-95 1.3241 0.0662 466.49 273.07 30.88 12.81
95-96 1.9702 0.0985 439.92 240.39 43.34 19.66
96-97 1.9065 0.0953 414.98 227.90 39.56 17.83
97-98 1.4545 0.0727 403.56 177.08 29.35 16.47
Source: Compiled from USDA data.


adjustments have been made to accommodate shipping expenses. As can be seen, the

amounts of quota rents in real terms are substantial and their loss would have serious

negative social repercussions. However, as indicated in Fig. 2.1, prices have trended

upward in the world market and downward in the US market during the period. An

obvious implication of this observation is that the real premium enjoyed by quota holders

and US producers has been declining. As will be seen later, this factor--coupled with a




o1 The Euro is currently valued at approximately 1 Euro = US $1.06.

" The Tariff Rate Quota (TRQ) system is a key component of the current US sugar program. This
component of the program is elaborated on below.






29


reduction in the quota levels--implies a considerable loss of real earnings for the quota

holders.

A stated objective of the US Sugar Program is to ensure a reliable supply of sugar to

American consumers at competitive prices, while providing some stability for American

sugar interests (growers and processors). Key components of the program include: 1)

loan rates; (2) an implicit market stabilization price (MSP); and 3) Tariff-Rate Quota

(TRQ). These components include:

1. Loan rates--a loan rate of US $0.18 per pound of raw sugar is made available to

processors. The loan rate of 22.9 cents per pound for refined beet sugar was

historically set in relation to raw sugar with a prescribed formula. Loans were

previously non-recourse in that, at any time, the processors could refuse to repay the

loan and the government would accept the sugar in lieu of cash. However, with the

implementation of the Federal Agricultural Improvement and Reform Act (FAIR) of

1996 current loan rate programs are to operate on a recourse basis under certain

conditions. Whenever the tariff-rate quota (TRQ, discussed below) is 1.36 MMT or

less, the USDA can demand repayment of the loan with interest at maturity,

regardless of the price of sugar. The loans are considered to be non-recourse only

when the TRQ is above the stipulated 1.36 MMT. In addition, loan rates were fixed at

the 1995 levels of 18.0 and 22.9 cents per pound for raw and refined sugar,

respectively.

2. Market stabilization price (MSP)--a major stipulation of the program in the past was

that it should run at no cost to the government. To ensure that such provision was

met, an MSP was explicitly established at a level to ensure that commercial market






30

channels were cleared rather than forfeited to the Commodity Credit Corporation

(CCC). The MSP was set at a level of $0.22 per pound of raw sugar. However, under

the FAIR program the no net cost provision was eliminated and the MSP is no longer

set explicitly although it is expected that supply would be managed in such a way as

to ensure the market price would be above the loan rate.

3. Tariff-rate quota (TRQ)--the TRQ for sugar and sugar-containing products is not

technically a part of the domestic sugar support legislation (Buzzanell, 1998).

Nonetheless, it is one of the principal tools used to support the loan program and to

ensure that the domestic price does not come within the forfeiture range. The US

Trade Representative (USTR) allocates the TRQ for raw sugar to 40 countries. The

total quota varies from year to year in keeping with the objective to stabilize domestic

raw-sugar prices. Although the quota can vary from year to year, it cannot fall below

1.14 MMT, which is the commitment given by the United States under the Uruguay

Round of the GATT (USDA, Situation and Outlook Series, 1998). Of the 40 countries

participating in the program, 23 are from the Western Hemisphere and include all the

SAC sugar exporters. Country allocations are based on historic trade with the United

States. Haiti, and St. Kitts and Nevis were included in the allocations as minimum

quota-holding countries, and the allocation to Mexico was increased to fulfill

obligation pursuant to North American Free Trade Agreement (NAFTA). About 66

percent of the raw cane sugar TRQ are allocated to the hemispheric countries. The

Dominican Republic, Brazil, CARICOM, and Argentina account for the bulk of the

overall regional quota with TRQ shares of 17 percent, 14 percent, 5 percent, and 4.6

percent, respectively. All countries in the Hemisphere, with the exception of Brazil,






31

receive preferential treatment (zero-tariff) in that they are able to sell their in-quota

sugar to the United States at the established US domestic price and retain the export

quota rent. In the case of Brazil, a small tariff of 0.625 cent/pound is charged on the

in-quota exports. Imports outside of the quota have been subject to a high tariff of

17.62 cents per pound since January 1, 1995. In accordance with GATT/WTO

commitments, this tariff is expected to decline to 15.36 cents per pound by the year

2000.

Since the quota is reserved for selected countries and thus can be considered

discriminatory towards the LDC countries, there is the view that the TRQ will be

seriously questioned in the 1999 WTO review round (Contrearas, 1998). In addition,

there is increasing pressure within the US to eliminate the current sugar program in

keeping with the Federal Agricultural Improvement and Reform (FAIR) Act of 1996,

also known as the "Freedom to Farm Act." FAIR seeks to eliminate or greatly reduce

government intervention into the agriculture sector as a means of making the sector more

competitive. While the issue of the elimination of the US sugar program is somewhat

debatable, there is the unconfirmed view that the once strong lobby, comprising the

growers of cane and beet sugar and the manufacturers of caloric and non-caloric

sweeteners, is losing its political clout, in part because manufacturers of HCFS no longer

need the protection accorded in order to make their operations viable.



Review of World and Western Hemisphere HFCS Markets

High Fructose Corn Syrup is a liquid caloric sweetener made from ordinary

cornstarch. It can be substituted for sugar (sucrose) in most liquid uses. As a consequence

of its relative cheapness in comparison to other forms of caloric sweeteners, it has been






32

used in a wide range of processed food products such as beverages, baked goods, dairy

products and jams and jellies. Although converting cornstarch into sweet substances was

discovered as early as 1811, it was not until the late nineteen-century that sweeteners

such as glucose corn syrup and dextrose were produced in corn wet milling. However,

these products were only about 70% as sweet as sugar and hence were not competitive

with sugar on a sweetness equivalency basis. The search to find a sweetener comparable

to sugar did continue and although promising results were obtained in the early 1960s

when scientists discovered an effective method of obtaining fructose from glucose --a

substance which is 110 to 170 percent sweeter than sucrose --the break through did not

occur until 1967 and was based on research carried out in Japan and the US. However,

commercialization of the product did not occur until 197212

HFCS comes in two strengths, HFCS-42 and HFCS-55. The numbers indicate the

strength of the fructose percentage. Thus, in the case of HFCS-42, the product contains

42% fructose, and about 50% and 8% of dextrose and other saccharides, respectively.

This product is approximately 90% as sweet as sugar. HFCS-55 contains 55% of fructose

and about 40% and 5% of dextrose and other saccharides, respectively. This product has

sweetness of about 110%. Commercial production of HFCS-42 began in 1972 while that

of HFCS-55 began in 1977. In 1985, through further processing of HFCS-55 a crystalline

form of the product was prepared for commercial use. However, certain technical and

economical blotches still hamper its manufacture and limit its widespread use as a direct

substitute for crystallized (tabletop) sugar. The main problems have to do with its still

relatively high cost of production compared with sugar and the fact that its sweetness


12 For details on the invention process of the product and facts about its diffusion see Zitt (1998).






33


appears to vary depending upon the particular use (Polopolus and Alvarez, 1991;

Thomas, 1985).

Although both syrups share certain favorable features such as stability, high

osmotic pressure or crystallization control, each offer special qualities to food

manufactures and consumers. Thus HFCS-42 is popular in canned fruits, condiments and

other processed foods which need mild sweetness that won't mask natural flavors. The

sweeter HFCS-55 is used mainly in the soft drinks industry, in ice cream and frozen

desserts. The main drawbacks to HFCS are that it is available mainly in liquid form, and

therefore restricted to only certain industrial uses. In addition, due to its high water

content it costs more to transport than an equivalent quantity of sugar, and it is difficult to

handle, since it must be maintained at 800 -1000 F temperature when stored or

transported (American Sugar Alliance, 1998).



World Production and Market Trends--HFCS

Table 2.9 shows world HFCS production for the period 1989-1997. It shows that

in 1997 world output of HFCS amounted to 10.41 MMT which is about 8.5% of the

combined world output of sugar and HFCS. The US dominates production with output of

7.71 MMT or 74% of total output in 1997. Japan's production of 0.8 MMT accounted for

another 7.6% of total output. Thus, together these two countries accounted for almost

82% of world output. It should be pointed out however, that while production of HFCS in

the US is unrestricted, production levels in both Japan and the EU (third largest producer)

are restricted.






34


Owing to the problems alluded to earlier with regards to storing the commodity,

very little is held in stock. Of particular interest is the fact that, between 1990 and 1997,

world consumption of HFCS increased at an annual rate of 4.0% while consumption of

sugar grew at a lackluster rate of 1.5% per annum. With the demand growth for sugar

failing to keep pace, HFCS has captured an increasing share of the combined sugar

market for these sweeteners, increasing from 7.2% in 1990 to an estimated 8.5% in 1997

(LMC International, 1998).


Table 2.9. World HFCS Production, 1989-97 ('000 metric tons dry basis)
Country 1989 1990 1991 1992 1993 1994 1995 1996 1997
US 5,345 5,677 5,852 6,038 6,459 6,813 7,143 7,408 7,711
Japan 744 783 794 747 727 806 789 792 800
EU 276 280 284 286 288 290 303 303 303
Korea 244 270 276 263 246 260 277 279 280
Canada 239 245 252 250 255 255 265 260 295
Argentina 146 161 184 184 195 207 190 198 210
Taiwan 51 67 110 125 150 171 182 195 210
Mexico 0 0 0 0 0 0 0 110 175
Others 156 210 197 229 258 308 354 388 423
World 7,201 7,693 7,950 8,122 8,579 9,109 9,493 9,933 10,407
Adapted from LMC International, (1998)


Currently, only a small amount of HFCS output enters into trade. As indicated

earlier, this is due to the technical difficulties and high costs associated with transporting

the liquid sweeteners. Consequently, trade is confined mainly to cross-border

transactions. In this regard, the only region where considerable quantities are traded

internationally is in North America.



Western Hemisphere Production and Market Trends--HFCS

With respect to the Western Hemisphere, total output of HFCS in 1997 amounted

to 8.43 MMT, representing 81.0% of world production. The main producing countries






35

within the region are the US (91.5%), Canada (3.5%), Argentina (2.5%) and Mexico

(2.1%). Production in Mexico started in 1996 by US interests and already output level is

close to that of Argentina and Taiwan. At the moment, the bulk of the corn used in the

production of this commodity in Mexico is imported from the US. The 1997 regional

consumption of HFCS was estimated at 8.32 MMT. Again the US was the main

consuming country accounting for 91.1% of total demand. This was followed by Mexico

(3.7%), Argentina (2.5%) and Canada (2.5%). Only a small quantity of HFCS is currently

consumed in CARICOM and none of the member states engage in the production of the

commodity.

The considerable success in HFCS as a substitute for sugar is best illustrated in

the case of the US. Table 2.10 shows the changes in the per capita consumption and the

relative market shares of caloric sweeteners consumed in the US over the period 1975 to

1997. And, Figure 2.2 shows a comparison of the trends in per capita consumption of

HFCS, total corn sweeteners and sugar for the same time period. In particular, the Figure

reveals that between 1975 and 1997, the per capita consumption of sugar fell from 89.2

lbs. to 67.1 lbs. after reaching a low of 60.8 lbs. in 1986, while the per capita

consumption of HFCS increased from 5 to 61.4 lbs. over the same period. This meant that

whereas in 1975 sugar accounted for about 76 % of the caloric market shares and HFCS

only 4%, by 1997 the market share of sugar had fallen to 43% while in the case of HFCS

it had risen to 40%. The rapid and considerable increase in the production and

consumption of HFCS in the US and the concomitant displacement of a portion of the US

sugar demand, has come largely at the expense of the sugar quota holders and by

implication the US sugar refiners. This is easily seen when one considers the fall in US






36


Table 2.10. Changes in US Per Capita Sweetener Consumption and Relative Market
Shares, (1975 - 1997) Pounds, Dry Weight Basis

Year Refined Relative HFCS Relative Total Corn Relative Honey & Edible Relative Total Caloric
Sugar Share (%) Share (%) Sweet'rs Share (%) Syrups Share (%) Sweet'rs
1975 89.2 76 5.0 4 27.5 23 1.4 1 118.1
1976 93.4 75 7.2 6 29.7 24 1.3 1 124.4
1977 94.2 74 9.5 7 31.2 25 1.4 1 126.8
1978 91.4 72 12.1 10 33.7 27 1.5 1 126.6
1979 89.3 70 14.9 12 36.4 29 1.4 1 127.1
1980 83.6 67 19.1 15 40.2 32 1.2 1 125.0
1981 79.4 63 23.2 19 44.5 36 1.2 1 125.1
1982 73.7 60 26.7 22 48.2 39 1.3 1 123.2
1983 71.1 57 30.7 25 52.2 42 1.3 1 124.6
1984 67.6 53 36.3 29 57.8 46 1.4 1 126.8
1985 63.2 49 44.6 35 63.9 50 1.5 1 128.6
1986 60.8 48 45.1 36 64.6 51 1.6 1 127.0
1987 63.1 48 47.1 36 66.8 51 1.7 1 131.6
1988 62.6 47 48.3 36 68.3 52 1.5 1 132.4
1989 62.8 47 47.5 36 68.0 51 1.6 1 132.4
1990 64.8 47 49.2 36 70.4 51 1.6 1 136.8
1991 64.4 47 50.0 36 72.0 52 1.6 1 138.0
1992 64.4 46 51.6 37 74.3 53 1.6 1 140.3
1993 64.6 45 54.4 38 77.8 54 1.6 1 144.0
1994 65.8 45 56.4 38 80.2 54 1.5 1 147.5
1995 66.2 44 58.4 39 82.6 55 1.5 1 150.3
1996 66.9 44 59.8 39 84.2 55 1.5 1 152.6
1997 67.1 43 61.4 40 86.3 56 1.5 1 154.9
Source:USDA, ERS, Commodity Economic Division, Sugar and Sweetener: Situation and Outlook Report,
various issues

100oo


80
2 70


650










Years
Fig. 2.2. Changes in US per capita consumption of selected sweeteners, 1975-97
Fig. 2.2. Changes in US per capita consumption of selected sweeteners, 1975-97






37


sugar imports from a level of 5.0 MMT in the seventies to current level of about 1.6

MMT (Hannah & Spence, 1997).

The growth of the industry in the US is attributed to several factors including: 1) a

marketing environment with no restriction on supply; 2) advancements in technology;

and 3) the relative cheapness of the product in relation to sugar on a sweetness

equivalency basis. Since its inception in the US the product has sold at a discount to

sugar. Figure 2.3 shows a comparison of the real price of HFCS-42 and the real world

and US prices of refined sugar. The evidence indicates that the pricing of HFCS is now

much closer to that of the world market price of refined sugar.


40

35






20




10

5




Years



Fig. 2.3. Real HFCS and refined sugar prices, 1982-98 (1992=100)
Source: USDA, ERS, Commodity Economic Division and Dept. of Commerce,
Bureau of Economic Analysis






38

As noted before, most of the international trade in HFCS occurs within North

America. Although the Canadian and the US-HFCS industries are difficult to separate,

from a custom point of view, Canada is a net exporter to US, which in turn is a net

exporter to Mexico. The growth in this trade is reportedly due to increases in the Mexican

demand (USDA, Situation and Outlook Report, 1998). However, owing to a dispute

concerning the legitimacy of the US-Mexican Side Agreement on trade in sugar under the

North American Free Trade Agreement (NAFTA), the Mexicans have recently (January

1998) placed a 100% tariff on imports of HFCS from the US. The issue is now being

addressed by both the WTO and the NAFTA dispute settlement bodies. The tariff has

retarded the growth in trade but the US has maintained current levels of exports to

Mexico. Total exports of HFCS from the US increased by 37 % in 1998 and now account

for as much as 6.2 % of domestic production compared with 3.7% and 4.7 % in 1996 and

1997, respectively.

Finally, even with the shipping problems of the liquid product, an increasing

amounts of HFCS are currently being shipped to markets such a Japan and Trinidad and

Tobago (USDA Situation and Outlook Report, 1997). Industry experts believe that

shipping problems are surmountable. Within CARICOM, Trinidad and Tobago is the

only country, which currently imports HFCS for use in its soft drink industry, however

there is the possibility that other countries might soon follow.














CHAPTER 3
LITERATURE REVIEW AND THEORETICAL BACKGROUND

Several studies have been done on modeling the US Sugar/Sweetener subsector

and drawing economic/welfare implications for various groups including domestic sugar

producers, processors, sweeteners' users, consumers and trading partners. Most of these

studies have either not taken into consideration the impact of HFCS in their analysis or

have only done so in a cursory manner. The first section of this chapter therefore

undertakes a critical review of those studies in which the HFCS industry was taken into

consideration. The review provides a basis for the conceptual model of the US sugar and

HFCS subsector developed in chapter 4. Because of the problem of time-varying

parameters in the HFCS industry and the need to provide more reliable parameters for the

HFCS industry, the second section of the chapter focuses on some critical time series

issues such as serial correlation, unit root and cointegration. The methodology of state-

space model and the Kalman filter, allows for estimation of time-varying parameters and

is employed later in estimating the derived demand for HFCS. Consequently, this

methodology is discussed in detail. The chapter ends with a discussion on the nature of a

derived demand and the general approaches used to estimates such functions.



Previous Empirical Studies

Sugar and HFCS Related Studies

One of the earliest studies to include HFCS in the analysis of the US sweetener's

market is a study conducted by Carman (1982). This study had as its objective the


39






40


projection of the rate of adoption of HFCS in the US and an assessment of its economic

implications for major sweetener industry participants. The study used a logistical trend

model with estimates of a ceiling market share of HFCS and total demand for caloric

sweeteners to carry out the projections. Among other things, the study concludes that the

total quantity of sugar demanded in the United States would decrease for several years as

HFCS demand increases. Also, that the impact of the HFCS on domestic sugar producers,

under the restrictive sugar policy, would be minimal since the bulk of the costs would be

borne by those countries exporting sugar to the US. Carman assumed that the saturation

of the HFCS sweetener market would occur between 25 -30% of the total caloric

sweetener demand. This assumption has proven to be too conservative since HFCS

currently accounts for as much as 40% of the caloric sweeteners' market (Table 2.10).

Lopez and Sepulveda (1985) developed and estimated a demand model for sugar

and on the basis of the model's results derived implications for sugar import policies. In

that model they addressed what they considered to be some shortcomings of earlier

studies. They: 1) estimated separate demand functions for industrial and nonindustrial

sectors; 2) compared the periods before and after the introduction of high fructose corn

syrup in industrial uses; and 3) utilized a partial adjustment framework in an attempt to

measure the speed of demand adjustment. Their empirical results indicated that: 1)

changes in consumer preferences and the availability of the cheaper HFCS in the food

processing segment were exerting downward pressure on sugar demand; and 2) the

demand for sugar had become less responsive to sugar price change after the introduction

of HFCS. They estimated that the short-run own price elasticity went from -0.15 to -

0.04 and the long-run (full adjustment) own price elasticity went for -0.31 to -0.06 after






41

the introduction of HFCS. They attributed the decreased in own price elasticity of sugar

to the substitution taking place in the more price elastic segment of the sweetener market

such as the soft drinks and canning industries. They concluded that as the US demand for

sugar decreases and the food industry adjusts faster to sweetener choices, the US

government would have to impose more restrictive import barriers to maintain prices to

domestic sugar and HFCS producers.

Leu, Schmitz, and Knutson (1987) used a general equilibrium model to analyze

the policy options for the US Sugar program and to empirically estimate the

substitutability between sugar and HFCS by including the price of HFCS as a

determinant of sugar demand. They used a composite price of corn sweeteners - glucose,

dextrose and HFCS--in order to generate a series dating back to 1955. Demand

elasticities were estimated for sugar and HFCS while supply elasticities were obtained

from previous studies. Their results indicated that the net social cost of the quota program

with an elasticity of excess supply of 2.37 was $203 million when substitution with

HFCS was allowed and $253 million with no HFCS substitution. Under the assumption

of a perfectly elastic excess supply curve the net social cost with HFCS substitution was

about $1 billion. They concluded that a deficiency payment program would be more

costly than an import quota. However, Marks (1993) stated that the study suffered since

price data on the HFCS were only available after 1975. He also questioned the composite

commodity assumption because glucose and dextrose are much less substitutable for

sugar than is HFCS.

Sudaryanto (1987) developed a trade model to analyze the effects of trade

liberalization by the US and EEC on international markets for sugar. In relation to the






42


US, he pointed out that analysts in previous sugar studies ignored the derived nature of

the demand for sugar by treating all sugars as if they were consumed directly. Utilizing

Fourier's flexible functional forms he separately estimated, a derived demand and direct

demand for sugar. He then computed an aggregate elasticity of demand for sugar using

the weighted average of elasticities obtained from the two models. While he did take into

consideration the impact of the possible substitution between HFCS and sugar in

estimating the various demand functions, he followed previous studies and utilized a

composite price index of HFCS, glucose and dextrose. The results of his analysis

indicated that in the United States, sugar for final consumption is far less elastic than

sugar used as intermediate input. Sudaryanto's results suggest that consumption for final

uses had reached the near saturation point and a change in sugar price had little effect on

consumption. In contrast, the demand for sugar in the food processing industry was more

elastic because of available substitutes. The result of his trade model showed that

unilateral trade liberalization in the US market would force an increase in US imports by

114% and would be accompanied by a decrease in both production and supply price by

8.3% and 11.2%, respectively. Complete removal of protection in the EEC was estimated

to cause a reduction in supply price and production in that region of 13.2% and 5.6%,

respectively.

Rendleman and Hertel (1989) examined the impact of policy changes on sugar

and HFCS within a general equilibrium framework. The results of their analysis showed

that any substantial increase in the production of corn sweeteners would cause an

increase in the production of byproducts, which would have the effect of lowering the

byproducts prices and thus raising the cost of production of corn sweeteners. They






43

estimated that free trade would result in the US prices for sugarcane, sugar beet and corn

sweeteners declining by 34%, 31% and 10%, respectively. Corn sweetener byproducts

would decline by 2% but prices would increase by 12 %. In terms of the changes to

output, they estimated that US sugarcane production would decline by 77%, sugar beet

production by 47%, and corn sweetener production by 2%. This study was an

improvement over many of the previous studies in that an attempt was made to seriously

analyze the interrelationship among the various segments of the sweetener market.

Tanyeri-Abur (1990) analyzed the implications of current and potential changes in

the US sugar policy with a price endogenous agricultural sector model. The sugar and

HFCS markets and the industries, which used these two sweeteners, were modeled.

Several policy runs were made under varying assumptions pertaining to the level of

imports, target prices and substitution relationship between sugar and HFCS. The results

were compared with the based year (1996) in which imports of raw sugar was restricted

to 1.6 million short tons. Her work indicated that: 1) the removal of US sugar quotas

would benefit US consumers by $657 million and increase net foreign surplus by $350

million; and 2) that US producers' surplus would decline by $288 million leaving a net

welfare gain of about $718 million. However, when the agricultural sector with all

commodities was considered, there was a reduction in net social welfare for the sector of

$1.2 billion dollars despite an increase in total consumer surplus of $1.4 billion. She

explained that this result was due mainly to the increase in government payments that

would result as the price of cotton and rice fall as land was shifted from sugarcane to

these crops.






44

She conceded that conclusions in relation to the HFCS subsector could be a

reflection of rigid assumptions that were made in relation to the supply and cost of

production of HFCS. For example, an unlikely result is that HFCS does not effectively

compete with sugar and is replaced by sugar when free substitution is allowed. In further

analysis in the said study, however, she showed that the above result was highly sensitive

to the assumptions made with respect to the HFCS supply elasticities. She concluded that

further analysis is needed into the behavior of HFCS, but that, by and large, the signs of

the welfare change and the distributional implications would remain the same.

Marks (1993) constructed a partial-equilibrium model in which he assessed the

long-run static welfare effects of alternative US sugar policies. He evaluated two policies.

The first focused on estimating the overall economic impact of the US sugar program for

crop years (1984/85 to 1988/89)--years in which the domestic-foreign price differential

varied widely. The analysis considered the long-run effect of the program on sugar prices

in the US and the rest of the world (ROW), the economic welfare of US and foreign

consumers and producers, and U. S. government revenues. His second policy analysis

examined the implication of a 10% real reduction in the US sugar loan rate, by which the

US provides support to its sugar producers. His approach to modeling the US sweetener

market differs from most of the previous studies in two ways. Firstly, he estimated the

total demand for refined sugar plus HFCS (US sweetener demand) as function of a

composite index of their prices. Secondly, the prices of refined sugar and HFCS in turn

were modeled as functions of the US raw sugar prices, which according to Marks is

targeted by the policy makers. In addition, he estimated equations for aggregate sugar

demand and supply for the rest of the world (ROW). From these equations he was able to






45


infer a price elasticity of ROW export supply. The price elasticity is important because it

gives an indication of the extent to which changes in the US imports would affect the

world market price.

Marks (1993) investigation reveals that in all years considered, the unilateral

liberalization of the US program would result in the world market price increasing, by an

amount which depends on the price differential between the US and world market price.

For example, the US program's depressive effect on the world market price in the year

1984/85, when the price differential was high, was approximately 16% whereas in the

year 1988/89, when world market price was relatively high, it was estimated at less than

9%. Among his other conclusions was the suggestion that the US sugar program

benefited the rest of the world in the aggregate over the period by an average of $0.331

billion. The program depressed the world raw sugar price, and cost foreign sugar

producers an annual average of $1.867 billion. However, these costs were outweighed by

$2.109 billion in gains to foreign consumers. The paradox he explained, was due to the

effects of the rent to foreign quota holders who profit from the price differential between

the US and the rest of the world. In his analysis of a marginal reform to the US sugar

program the result indicated that a 10%/ reduction in the real US loan rate would lead to

higher welfare in the aggregate in both the United States and the rest of the world.

Marks (1993) study although novel in some respects, suffered from two

weaknesses. First, the approach in estimating the US sweetener demand used the

historical annual market shares of refined sugar and HFCS as weights in computing the

composite sweetener price index. This method assumed a kind of fixed proportion good.

Second, in estimating the effects of future policy changes the erroneous assumption was






46

made that the technical substitution of HFCS for liquid sugar was complete and that the

existing market share would continue into the future.



Evaluation and Critique

A common feature of the studies reviewed here is that all utilized time series data

in their analyses to estimate parameters in supply and demand equations. With the

exception of the study by Lopez and Sepulvelda (1987), none considered the issue of

time-varying parameters. Secondly, although some attempt was made to take into

consideration the substitution effect between HFCS and sugar in the US sweeteners

market, none of the studies directly included HFCS in their trade analysis. Thirdly, all

studies inferred that if the US were to liberalize its sugar program unilaterally, it would

have a positive (upward) impact on the existing world market price. Fourthly, all the

trade-related studies were either spatial or nonspatial equilibrium models. According to

Thompson (1981) although spatial equilibrium models represent one of the most popular

approaches to agricultural trade modeling, especially for comparative statics analysis of

the effect of a change in policy, their adequacy for this purpose is questionable. He noted

that such models were not consistent with how many international agricultural markets

normally operate and hence do not do well in accounting for actual trade flows. Chief

among the reasons given for the real world inconsistency of the models was their faulty

assumption of a homogeneous traded product, ignoring the fact that many importers

differentiate among exporters on historic or political grounds. Consequently, changing

from one source to another is usually much more difficult than assumed by such models.

The second class of models used was the nonspatial equilibrium types in which

simulation rather than optimization procedures are used to give the net trade position of






47


each country. Such models are usually less efficient as a means of examining the effects

of changes in policy variables, such as tariffs and transport costs. The main advantage of

these models is that they are usually much easier to solve (Thompson, 1981). Both spatial

and non-spatial models are based on a competitive framework. However, as shown by

authors such as Krugman (1979), models with imperfect competition features might be

more realistic when modeling international trade. Although it is recognized by this author

that of necessity any model is an abstraction of the real world, an attempt will be made to

address a few of the shortcomings alluded to.



Time Series Analysis

Aoki (1990) states that the main aims of time series analysis are one or a

combination of the following: 1) to describe the data behavior succinctly; 2) to explain

the behavior of the time series in terms of exogenous variables; and 3) to forecast and in

some cases to control. Harvey (1987) makes a distinction between a pure time series

model and a structural time series model. The distinguishing feature of the two is that no

attempt is made in the former to formulate behavioral relationships among variables, in

contrast to the latter. Rather, the movements in the variable are 'explained' solely in terms

of its own past, or by its position in relation to time.

In utilizing time series data to carry out analysis several issues must be confronted

including a determination of whether the data are: 1) serially correlated; 2) stationary; 3)

possess a deterministic or stochastic trend; and 4) the stability of the parameters (time-

varying parameters). In cases where two or more time series are to be regressed on each

other there is the added concern of cointegration and guarding against spurious

regression. The following discussion provides brief insight into each of these issues.






48

Serial Correlation

Serial Correlation (autocorrelation) is the extent to which a time series variable

(usually the error term of the time series model), lagged one or more time periods, is

correlated with itself (Shim and Siegel, 1995). When such a situation exist it has

implication for the classical linear regression (OLS) model. The OLS estimator of the

parameters remains unbiased but no longer efficient and in fact is asymptotically

inefficient. Moreover, the standard formula used by the OLS to compute the estimate of

the variance of the errors is no longer unbiased and could either over- or underestimate

the true variance. The implication is that all tests of reliability, using the t- and F-

distribution, which rely on the estimated variance-covariance matrix of the estimated

parameters, become invalid.

Several tests such as the Durbin Watson (DW) statistic are available to detect the

presence of serial correlation and depending on the causes, various remedial actions can

be taken. In pure times series analysis, taking account of the pattern of serial correlation

usually allow for better forecast of future observations. Darnell (1994) notes that

detection of a pattern of first-order positive autocorrelation by means of the residuals can

result from several factors other than autocorrelated true errors (unobserved). These

factors include mispecification of functional and structural breaks in the relationship

(time-varying parameters). Consequently, he suggests that a respecification of the model

might be more appropriate and only when the evidence strongly suggest autocorrelated

true errors should the model be re-estimated by feasible generalized least squares or

maximum likelihood.






49


Stationarity

Stationarity is an important aspect of a time series. Formally, an infinite sequence

of random variables is said to be strictly stationary if the joint probability function of any

finite subset is identical to any other subset (Darnell, 1994). Less formal, when a series is

generated by stationary process it tends to fluctuate around a constant level and there is

no tendency for its spread to increase or decrease over time. In other words, the mean of

say Yt, its variance, and its covariance with other Y values, say Yt-k, are invariant to time

and the covariance between say Yt and Yt+, is a function ofs only, for all t. Hence an

immediate implication of a stationary time series is that it has a time independent mean

and is homoscedastic. Kennedy (1998) points out that although many scientific time

series data are stationery, most economic time series data are nonstationary since they

tend to exhibit some kind of trend over time (i.e., the mean changes over time). One

reason for this (Aoki, 1990) is that the circumstances facing optimizing economic agents

change with time. Another is that in several cases the dynamic structures generating the

data are mostly nonlinear.

Nonstationarity has implications for the standard regression model in

econometrics since the model usually makes assumptions regarding the stationarity of the

error term as well as that of the variables in the regression. It has been shown that running

regression on nonstationary data can give rise to misleading (or spurious) values of R2,

Durbin Watson, and t-statistics, which may lead to an erroneous understanding of the

relationships which exist among the regression variables. For instance, it has long been

recognized (Yule, 1926) that regressing one nonstationary (trended) variable on another

could result in spurious regression.






50

Nonstationary time series have been subjected to one of two detrending

procedures in order to make them stationary before subsequent analysis. The two

procedures are: 1) regressing the time series as a simple linear (or higher order) function

of time and then using the residuals as the detrended series; and 2) an approach which

involves using first (or higher) order differencing of the time series. The first method

works only if the series say Yt has a deterministic trend and is generated according to the

following equation:

Yt = a +pt + Et (3.1)

where et is white noise. If however, the time series is generated by say a random walk or

random walk with a drift as for example:

Yt = Yt- + p + vt (3.2)

where vt is white noise, the first approach will not work. More formally, equations (3.1)

and (3.2) are referred to as trend-stationary process (TSP) and difference stationary

process (DSP) with a trend, respectively.

The second approach has had some measured success even though as Kennedy

(1998) points out some analysts have adopted it without any justification based on

economic theory. Within this approach, a variable is said to be integrated of order d,

written as I(d) if it has to be differenced d times to be made stationary. A stationary

variable is integrated of order zero. Economic variables are seldom integrated of order

greater than two and if non-stationary they are usually I(1) (Kennedy, 1998). Equation

(3.2) above represents a series, which is integrated of order one. Taking Yt-i from both

sides of the equation gives equation (3.3):

AY, = P +vt (3.3)






51


which is a stationary process. If 1 equals zero then the series is said to be integrated

(random walk) without a drift and if P � 0 then it is said to be integrated of order one

with a drift.

Because of the serious econometric implications of modeling variables which are

nonstationary, as pointed out earlier, tests for stationarity and the degree of integration

have all but become routine to econometric analysis when working with time series data.

The most popular tests continue to be the unit root tests which are designed to test the

order of integration of the variable. Kennedy(1998) points out that although a wide

variety of unit root tests are available, none of them is very powerful. The most popular

unit root tests are the Dickey Fuller or the Phillips-Perron in which the null hypothesis is

that the variable is non-stationary. But as Kwaitowski et al (1992) have shown

conflicting results could come about from the use of these tests and the one they proposed

with the null hypothesis being that the variable is stationary.



Cointegration and Error Correction Models

A major problem stemming from the use of variables made stationary by

differencing is that valuable information pertaining to the long-run equilibrium properties

of the data would be lost. The solution to this problem appears to be in a concept known

as cointegration. Two times series Yt and Xt are said to be cointegrated if both series are

individually integrated of the same order say I(1) and there exists a linear combination of

the two variables which is not integrated. In such circumstances it will be possible to

carry out a regression on levels of the variables and the results are meaningful (not






52


spurious) and there is no loss of any valuable long-term information, which would result

if their differences were used instead.

Moreover, in such situations, it is possible to tie in the short-term behavior of a

variable obtained from differencing, with a long-term equilibrium relationship. This is

done with a special formulation which involves a mix of both level and difference terms.

Such models are referred to as error correction models (ECMs). The ECMs can therefore

be viewed as comprising the short-run transitory effects and the long-run relationship and

describes how the long-run solution is achieved by a combination of negative feedback

and error correction (Darnell, 1994). In this regard, it is seen as a way to model the short-

run dynamics without losing sight of the longer-term steady state as may be posited by

economic theory. In other word, the ECM specification provides a means by which the

short-run observed behavior of variables can be associated with their long-run

equilibrium growth paths. Kennedy(1998) notes that this is one of the reasons why

economists have shown such interest in the concept ofcointegration, since it provides a

formal framework for testing for and estimating long-run (equilibrium) relationships

among economic variables. He suggests that if a cointegration relationship can be found,

advantage should be taken of the ECM framework.

It should be pointed out however, that the use of a single representation, implicitly

assumes that all the explanatory variables are exogenous. If this is not the case, then it is

suggested in the literature that one needs to use a vector auto regression (VAR) approach.

Furthermore, as long as the equation contains more than two variables, there is the

possibility that there could exist more than one cointegrating relationship. In such cases,

the traditional estimation procedures become inappropriate and again there is need to






53


switch to the VAR approach in which each variable is modeled in terms of lagged values

of all the other variables (Kennedy ,1998). Within such a general framework, testing can

be done to determine the number of cointegration relationships and the exogeneity of the

variables. The most common method is that of Johansen (1988). Manipulation of the

single equation VAR can produce a vector error correction model (VECM) in which the

vector terms are expressed as lagged differenced vector terms plus a lagged level term

which represents the error correction mechanism. As in the case of the ECM discussed

earlier, the intent is to exploit the steady state conditions resulting from cointegration

relationship and the short-term effects from the differenced variable in a single

framework.

A survey of the literature reveals, however, that there are unresolved issues with

the concept of cointegration and the use of ECMs and VECMs. For one thing, although

ECMs and VECMs are used widely in the literature, several analysts (Harvey, 1997; Hall

et al, 1992; Aoki, 1990) have expressed serious reservations about the mixing of level

and difference variables in a given equation. Secondly, Monte Carlo studies have shown

that the estimates of the cointegration regression have considerable small sample bias in

spite of excellent large sample properties ( "superconsistency"). In this connection,

Cheung and Lai (1993) point to several finite-sample shortcomings in the Johansen

method, and have attempted to address some of these shortcomings by adjusting upwards

the critical values of the Johansen test. However, an implication of this is that the

likelihood of finding cointegration with finite sample correction is much more difficult.

Thirdly, in situations in which there are more than two variables in the regression

equation there is the likelihood that there could be more than one cointegration






54


relationship. In such situations, the standard OLS procedure does not produce estimates

that are consistent (Kennedy, 1998). Also, in these circumstances when there are more

than one cointegrating relationship there is usually a certain degree of ambiguity

surrounding the interpretation of the estimated cointegrating vectors (Johnston and

Dinardo, 1997; Kennedy, 1998). Kennedy (1998) further points out that the interpretation

of multiple cointegrating vectors can be frustrating. He refers to a practice among some

researchers, in dealing with this issue, of ignoring those cointegrating vectors that seems

not to make good economic sense as tantamount to imposing slightly false restrictions to

improve mean square error.

Fourthly, according to Johnston and Dinardo (1997) VARs have serious

limitations as a tool for the analysis of economic systems and are susceptible to the

problem of vanishing degrees of freedom, since the number of unknown coefficients can

rapidly approach the available sample size. An implication of this is that researchers

wishing to use this technique, but having limited data (observations), are forced to be

quite parsimonious in their model specification. However, this could be disastrous since

the failure to include an important variable could result in absence of any detection of

cointegrating relationship among the variables (Kennedy, 1998). On the other hand, as

pointed out by Johnston and Dinardo (1997), as more variables are added to the VARs,

problems arise in testing the number ofcointegrating relationships. This comes about

because the test statistics are nonstandard distribution and require simulation. At present

the available tables can only account for eleven variables.






55


Time-Varying Parameters

Another potential problem with time series regression models is that the estimated

parameters may change. A review of the literature indicates that econometricians have

been aware of the problems of structural change since the late sixties as evidence by the

work on random coefficients (Hildreth and Houck, 1968; Langham and Mara, 1973).

However, it was not until the work of Rosenberg (1973) and Cooley and Prescott (1973)

that much more attention was focused on this particular phenomenon. Cooley and

Prescott (p. 463, 1973), in making a case for such analysis wrote:

"In recent years economic theory has increasingly abandoned the relative security of
static equilibrium and perfect certainty...[as] it has become increasingly clear that to
assume behavioral and technological relationships are stable over time is in many cases
not only heroic, but completely untenable on the basis of economic theory."

Since then there has been a proliferation of articles in the literature dealing with these

issues (see for example survey by Nichols and Pagan, 1983). Ward and Myers (1979) for

example, used a distributed lagged advertising model with coefficients that had random

and systematic adjustments to demonstrate the dynamic effects of advertising on

consumer demand. Ward and Tilley (1980), using time-varying parameters with random

component, illustrated quite clearly the problems that can occur when parameters that

have changed are, in fact, ignored in the specification of the econometric model. Two

other interesting examples include the work of Engle and Watson (1987) in forecasting

electricity sales with time-varying parameters, and the work of Harvey et al (1986) on

stochastic trends in dynamic regression models, focusing on the employment--output

equation. In all such models one or more of the parameters is allowed to evolve over

time, and estimation techniques (ranging from simple models of varying parameters to






56


more complex ones involving state-space framework and Kalman filter (discussed

below)) are used for both the fixed operators and the varying parameters.

Time-varying parameters is regarded as a problem because standard regression

analysis of economic phenomenon assumes that the relationship being studied is stable

within the sample of estimation. For example, in the simple regression model of Y = XP

+ ut it is presumed that the P vector, which represents the effect of changes in the

explanatory variables upon the dependent variable, is constant. However, there are good

reasons why this might not be so and in fact should be allowed to vary. Ward and Myers

(1979) grouped the possible sources of parameter variation under three broad categories:

1) structural changes in the economic phenomenon being studied; 2)model

mispecification; and 3) aggregation. They noted that technology and institutional changes

always cause structural changes in economic phenomena. The Lucas (1976) critique of

economic analysis provides a good example. Lucas' critique is to the effect that if the

policy regime is changed and economic agents takes account of the general policy

environment into their decision-making, then agents will adjust their behavior and the

coefficients that held in the previous regime will change to new values.

Sources of misspecification are varied but include omission of independent

variables, wrong functional forms and the use of proxy variables. For example, if the true

relation is:

Yt = al + a2Xt + a3X2t + Ut (3.4)

and the analyst considers the linear relationship give by

Yt = P1 + P2Xt + vt (3.5)






57


then it is clear that the effect of B2 which is equal to (X2 + 2a3Xt is not a constant. In the

case of aggregation as source of parameter variation, Ward and Myers (1979) cites as an

example the fact that over time, the relative importance of microeconomic agencies will

change and that such changes might not get reflected in the aggregate weight. In addition,

they noted that aggregated(macro) variables are discrete (Le., indexed as discrete time

points) whereas the underlying microvariables may be continuous, providing another

reason why parameter variation are likely to occur in macromodels.

In the literature, parameter variations are classified under the two broad headings

of stochastic and nonstochastic parameter variation. Each of these classifications is

further broken down into sub-categories. Hence, nonstochastic parameter variations,

which are caused by structural change in the economic phenomenon being studied, can be

divided into: 1) discrete variation or switching regression; and 2) systematic variation.

Likewise, stochastic models are divided into: 1) those which are stationary and are

referred to as random coefficient models; and 2) those which are nonstationary and are

termed sequential (Markovian) models3.



The State-Space Framework



The State-Space Model

Kennedy (1998) states that the state-space model can be regarded as a

generalization of the linear regression model and provides a unifying framework for all

dynamic linear models used in econometrics. A similar view is expressed by Harvey


13 For additional information on the various aspects of these types of models see Ward and Myers, 1979
and Nichols and Pagan, 1983.






58

(1993, 1997) that the state-space model is fundamental to dynamic modeling, and that

since most times series models can be put in this framework they provide a more robust

alternative to the modeling of time series analysis. The state-space framework is a

concept borrowed from engineering and is based on the original work of Kalman (1960)

and Kalman and Bucy (1961). The concept centers around the possibility of tracking the

economy in a similar manner as engineers would track the "state" of a system, such as the

location of a satellite or a tanker, using noisy measurements. The Kalman filter

(discussed below) was used to derive optimal estimate of the state, given knowledge of

certain parameters. Engle and Watson (1987) note however, that there is a slight

difference between the approach used by engineers and that used by economists. In the

main, engineers usually have qualitative theories that describe the equation of motion of

physical systems and were primarily interested in the "state" of the system obtain from

noisy measurements. Economists on the other hand, were less fortunate in not having

such laws of motion of the economy at their disposal and consequently were much more

interested in discovering them from noisy data, rather than in merely estimating the state

of the economy. Consequently, economists had to overcome the initial problem of not

having the parameters of process in estimating the state. The focus for economists

therefore shifted slightly to finding a suitable methodology by which such parameters

could be estimated. The solution to the problem came in the mid-1970s when economists,

with the use of the Kalman filter were able to evaluate the likelihood function in

complex cases. Engle and Watson (1987) note that this development allowed the

parameters to be estimated using maximum likelihood methods as in standard

econometrics. It therefore opened up the way for the widespread use of the technique in






59


applied economics and the model became a natural generalization of latent variable

models to a full dynamic framework A brief taxonomy of empirical applications of such

modeling in economics can be found in Engel and Watson (1980).

To illustrate the state-space model, first consider the following standard single

equation regression:


y, = a +t+ JCix,_, +, (3.6)
i=0

where yt is a scalar observable data series, xt, is the single observable explanatory

variable with the structure of the lagged coefficients being 60...6m, t is the trend variable

with level and slope coefficients a and P, respectively, and Et is the stochastic disturbance

term distributed such that &, - NID(0,e2 ). Harvey (1987) asserts that while equation

(3.6) is typical of many regression equations and that while many economic time series

show a trend, it is equally apparent that unless the time period is fairly short the trends

cannot be adequately captured by a deterministic (straight line) representation. He

observed that a great deal of applied economic work fall prey to the above error, since

they start off by detrending the data by regressing on time, thus making all that follows

invalid. He further points out that there is no reason in principle, why the parameters of

the explanatory variables of time series should not also be allowed to vary over time. He

suggests that a better approach would be to formulate the model in a state-space

framework. Such a representation could account for any unobservable components as

well as time-varying parameters.

Following along the lines suggested by Harvey (1987), reparametizing equation

(3.6) in terms of a simplified version of the state-space model, appropriate for time-

varying parameters yields the following:






60

m
y, =, + -iXt,- +,t (3.7a)
i=0

P, = ,,_ + f-, + (3.7b)

A = A-., + , (3.7c)

it, = .it- + v, (3.7d)

where it is assumed that st ~ NID(0,o02 ), T),- NID(0,o,2 ), ,~ NID(0,0g2 ), and v,-

NID(0,Ov2 ). The component At is the trend, whereas Pt is the slope of the trend and 8it are

the parameters of the explanatory variables, which are assumed to vary over time. The

introduction of the stochastic disturbance terms tit and �t allows the level and slope of the

trend to change slowly over time. Likewise the disturbance term vt , allows 6it to be

generated by a random walk. More formally, the above state-space representation with

the simplifying assumption of no lags is:

Y, = (1 0 )at + 6t (3.8)

where

P, t ' 1 1qt 0 ,.. ,
at = , = 0 1 0 l ,- + ,1 (3.9)
9t, 0 0 o1 A- J , Y,

In terms of state-space description, such models usually consist of two parts: 1)

the measurement equation which describes how the data actually observed is generated

from the state variables (equation (3.7a) or equation (3.8)); and 2) the transition

(dynamic) equation which describes the evolution of a set of state variables (equations

(3.7b to 3.7d) or equation (3.9)). The vector at comprising the elements .t, it, and 6t is

regarded as the state vector to be estimated and the measurements that are used to






61


estimate this state are the y,. . The statistical algorithm used to derive these state

parameters is the Kalman filter. Yt is the measurement equation. The parameters on the

right hand side of the transition equations (3.7a to 3.7c) or more precisely the system

matrix in equation 3.9, are described as the hyperparameters. They are regarded as being

fixed and known although they may change over the period in question. Engle and

Watson (1987) points out that on the basis of empirical experience and theoretical

arguments, in most cases these transition equations should to have unit root.

Concerning equations 3.7b to 3.7c, it should be noted that in the special case

where o,2, o2 , and ov2 equal zero, the hyperparameters remain constant and equation

(3.7) collapses to the standard regression equation (3.6) with the deterministic trend and

fixed regression parameters. Harvey (1997) also showed that ifo, 2 = oQ2 = 2 = 0 and if

o,2 > 0, then the first difference formulation is obtained as:

Ayt = p + iAXt-i + it (3.10)



Kalman Filter

The Kalman filter is simply a statistical algorithm, which allows certain

computation to be carried out for a model cast in the state-space form. Following the

approach of Harvey (1987) the typical model for which the Kalman filter is applied takes

the form of:

Yt = xtt + ct (3.11)

St = Gtt.A + ot (3.11a)

where as previously noted, equations (3.11) and (3.1 la) comprise the state-space system

in which the former is defined as the measurement equation and the latter as the dynamic






62


equation. Also, it is assumed that Ct - NID (0, o2 ), and o,~ NID (0, O,2 ). Although xt,

Gt, ,o,2 and 0,2 may ultimately depend on a set of unknown parameters, they are for the

purpose of the Kalman filter regarded as being fixed and known. Gt, oE2 and O,2 are

defined as the hyperparameters and are normally supplied by the researcher. In cases

where such information is not provided each defaults to the identity matrix. The

parameter of interest in the model is the state variable 6t, which evolves over time.

The Kalman filter uses a recursive estimation procedure to update the state as new

observations become available. In order to begin the procedure one must specify values

for ,o2 02, x2, x, Gt and the starting value for 6o. However, since such values are not

usually available the usual procedure is to construct starting values from the observations

themselves. Thus, in the case of a classical linear regression model with k explanatory

variables the starting values can be obtained by applying ordinary least squares (OLS) to

the first k observations with G = I and a,2 = 0.

The operation of the Kalman filter is easier understood when interpreted from a

Bayesian point of view. In this regard the goal of estimating 6t can be though of as being

carried out in two stages (Shen et al, 1999). In the first stage the 8t is estimated prior to

observing yt through equation 3.1 la. This can be represented as:

5t, = G, A, (3.11 b)

where, , is regarded as the best guess of the true value based on all information

available up to time t-1. In the second stage when the actual value ofyt becomes

available, 8t is updated as follows:


,t = G,,, + A,e (3.11c)






63


where, Ae, is considered to be the correction term. This term comprise two components.

The first component, the prediction error et, is obtained as the difference between the

actual yt and the estimated or forecast value, y,, which is based on the prior (un-

updated) estimate of 8t i.e. et = yt - y, and k, = xts, + ,i. The second term, A,, can be

considered as the regression coefficient of the regression of 8 on et (Shen et al, 1999).

The entire process can be viewed as one in which the Kalman filter takes the

starting estimate of 6, and corrects it as new information becomes available, through a

one step at a time process, so that the estimator at any time t is a better predictor of the

actual observation. As mentioned earlier, from a practical standpoint, the starting values

are usually obtained from the computation of an OLS regression. Consequently, if the

equation contains k variables then k observations will be required to construct the starting

values. Thus, the first k estimates are usually disregarded. Having obtained these starting

values the procedure continues in a recursive manner until the final set of information at

time T becomes available, producing T- k one-step-ahead prediction errors.

At the stage when all information has been accounted for, a better estimator can

be obtained. The techniques for computing such an estimator are known as smoothing.

There are three basic smoothing algorithms; fixed point, fixed lag, and fixed interval

(Anderson and Moore, 1979). The one most commonly used is the fixed interval

smoother, which starts with the final Kalman filter and works backwards. It thus ensures

that an optimal estimator is available at all point in time based on all information up to

and including the final observation, YT. This information as pointed out by Harvey (1987)

can be quite valuable for examining the way in which a component such as trend or

elasticity evolved in the past. Finally, it should be noted that when the model is linear and






64


time invariant the Kalman filter will usually converge to a steady state, in that the

covariance matrix becomes time invariant and the estimates of the parameters are the

same as those obtained from fixed coefficient regression (Harvey, 1987).



Advantages of State-Space Approach and the Kalman Filter

Several authors (Harvey 1987,1997; Engle and Watson 1987; and Aoki 1990;

Hall et al, 1992) have alluded to the advantages of the state-space model over the

traditional parametrization of time series models such as ARMA, ARIMA, VARs and

VECMs. They noted that from a strictly theoretical point of view both types of models

are equivalent, because generic models in one representation can be transformed into the

generic ones in the other. However, when judged on other grounds such as numerical

stability, sensitivity with respect to small specification errors, statistical properties of

parameter estimators, or simply ease of dealing with nonstationary series, the state-space

models are superior. Moreover, they point out that in order to avail oneself of the

theoretical results and computational algorithms that have been developed over the years

in system literature, time series must be put in this newer state-space representation.

Hall et al (1992) in motivating the discussion for the use of the state-space

approach and the Kalman filter notes that although it is widely used in certain branches of

engineering and by applied statisticians, it is only just emerging as a possible useful tool

of the applied economists. He notes that the Kalman filter can be interpreted in terms of

agents forming expectations. He points out that while the paradigm for modeling

expectation is the rational expectation hypothesis (REH) where agents act as if they know

the true model of the economy up to a set of white noise errors, the consideration of the

information availability assumption has led critics to label it unrealistic. He notes that the






65


Kalman filter can be used to address the concerns of critics of the REH, such as Friedman

(1979), with regard to the information-available assumption. This he states can be done

by extending Friedman's framework in which he advocates that given the true model yt =

xtP + ut (ut is white noise) agents may sequentially update their estimate of the fixed true

parameter vector p as more information on (yt, xJ) becomes available (e.g. time-varying

parameter models). Hall et al (1992) suggested that this could be done by: 1) assuming

that agents have some prior information about p (at time t = 0); and 2) by allowing P to

vary stochastically. Thus, agents are not assumed to know instantaneously the true model

but they do use the information optimally (efficiently), hence making the information

available assumption of the REH much more plausible. They further point out that in

certain models, the Kalman filter can be viewed as mimicking a learning process by

agents as in the case of adaptive expectation where the adjustment parameter is updated

each period, based on new information. This he notes formalizes Flemming's (1976) idea

of a 'change in gear' when forming expectations. He makes the point that while the

adaptive expectation has been shown to be optimal (in the sense of producing unbiased

forecasts) only when the data generation process is of the form of an integrated moving

average, IMA (1,1), or ARIMA (1,1,1), the Kalman filter is optimal under more general

conditions, and in fact produces minimum mean square estimators (MMSE) under the

normality assumption. Harvey (1987,1997) has pointed out several other advantages of

the state-space framework and the Kalman filter. These are summarized below14.


14 The approach allows unobserved components to be incorporated in a model and the Kalman filter
provides the means of estimating them by way of a likelihood function. Estimates of the unobserved
components can then be obtained by smoothing. It therefore allows the entire model to be set up in terms of
components, which have direct interpretation.It provides a framework in which time-varying parameters
can be estimated. For example, a stochastic trend can be estimated in which the level and slope are allowed
to evolve over time. Moreover, the framework is flexible enough to facilitate estimation of those






66


Relevance of the State-Space Model and Kalman Filter to the Current Study

The state-space framework was used in this study to estimate the derived demand

for HFCS. First, as pointed out elsewhere, the industry is relatively new which limits the

number of available observations (approximately 22). Also as discussed earlier, the VAR

based cointegration approach is biased with finite samples. In addition, one runs the risk

of encountering the vanishing degree of freedom problem referred to by Johnston and

Dinardo (1997). Second, there is strong evidence (Chapter 2) to suggest that the

parameters of the demand equation have been varying overtime. For instance, the

growth rate of the demand for the product has varied considerably over the investigated

period. This suggests some kind of a stochastic trend rather than a deterministic trend.

Also, as pointed out elsewhere, the adoption of the product was gradual, implying that



parameters which do not vary over time, since the hyperparameters, which allow the parameter to vary, in
these cases would be set to zero, resulting in the estimation of fixed coefficients. These models in
forecasting put more weight on the most recent observations; the faster the level and slope change, the
more past observations are discounted.The state-space approach allows one to work in level data, rather
than differencing, even in cases where the data are nonstationary. Working in level form makes
interpretation of results much easier. He notes that traditional time series analysis, in situations where the
data is nonstationary, stresses the role of differences. Within the state-space framework it is not usually
necessary to difference in order to specify a suitable model. A similar view is expressed by Aoki (1990)
that while economic time series are usually nonstationary because circumstances facing optimizing
economic agents change with time and do not remain the same, the Kalman filter can deal more effectively
with nonstationary time series than can the traditional time series methods. Within the state-space
framework it is not necessary to test for unit root, as concerns over the degree of integration of a series are
not crucial. The state-space frame work is flexible enough that if the slope parameter is deterministic, the
hyperparameters which allow it to change over time will be estimated as zero or close to zero, hence little is
lost if indeed the series is not integrated The situations where VECMS can be usefully employed are quite
limited and one should have reservations about using them to provide a general vehicle for modeling
economic time series. These reservations stem from the fact that the VAR-based cointegration methods are
based on auto regressive models and like the unit root tests can have very poor statistical properties. In
addition, it is very difficult to fit autoregressive models to data with slowly changing seasonality or trend.
The strongest objection, centers not on the fact that when there are two or more cointegrating relationships
they can only be identified by drawing on economic knowledge, but rather that the VECM does not provide
"a sensible vehicle for modeling the short-run...since it confounds long-run and short-run effects." As an
alternative to the VECMs Harvey proposes the use of multivariate structural time series model within the
state-space framework noting that such modeling can be formulated to incorporate long-run components of
cointegration directly by means of common trends.An added advantage of the state-space framework and
the Kalman filter is that the researcher need not understand the workings of the filter to apply the technique
and as such is free to concentrate on the selection of a suitable model and its interpretation.






67


time was needed for firms to learn about the product, determine its compatibility with the

manufacturing process, to formulate new recipes as well as make adjustment to existing

ones and to assess new product attributes. Also, with regard to some of the other

parameters such as the own price and cross price elasticity, for similar reasons mentioned

above, there is a strong suspicion that these might have undergone changes over the time

period, in view of the considerable changes which have taken place in the development of

the product making it far more attractive to the end users and consumers.

Third, is the fact that the objective of the study is not so much concerned with

whether there exists a long term relationship among the chosen explanatory variables, as

it is to identify the dynamic path of adjustments which might have occurred in the

industry and to draw inferences. As a consequence, the issues ofcointegration and unit

root, though important, are of lesser concern within the present context. As Engel and

Watson (1987) observed that in most of the studies where time-varying parameters have

been used the predominant consideration seems to be on the stability of the regression

equation, only in a few cases have the variations been interpreted as economically

important. The current study intends to add to those few studies.

Finally, the use of the state-space framework within this study is an attempt to

make use of some of the theoretical and practical advances that have been made in the

area of systems literature and to provide yet another case where the techniques have been

used in economics.



The Concept of a Derived Demand

The term derived demand is normally used to denote the demand for inputs that

are used to produce the final product. According to Shim and Siegel (1995) the term






68


refers to the demand for a factor of production or product that is derived from the demand

for other goods. Neoclassical theory suggests that a profit maximizing firm will employ a

combination of inputs up to the point at which the additional cost of employing one more

unit of that input is equal to the additional revenue generated from the presence of that

that input. Given such a framework, a firm's input demand functions (derived demands)

can be obtained in one of several ways. First, by solving the firm's profit maximization

objective function with respect to the given input. Alternatively, a system of derived

demands can be obtained by application of duality theory. This theory suggests that given

either the firm's cost function or its profit function and assuming that such functions

satisfy a set of regularity conditions, the system of derived demands can be obtained by

applying Shephard's lemma and Hotelling's lemma to the given cost and profit function,

respectively (McFadden, 1978). For example, given a simple firm which produces one

output using three inputs--the third of which is fixed in the short run--the profit value

(optimized) function can be represented as follows:

7t = n* (P,v,w,Z) =Max [Pq -vxl - wx2 : f(x, x2,Z) > q]

where 7* represents the maximum profit that the firm may obtain by allowing a subset of

inputs and output to vary while another subset of input is held fixed; P is the price of the

output; v and w are the prices of inputs xi and x2 , respectively; x, and x2 are the levels of

the variable inputs used in the production process, while Z is the fixed input; q is the

quantity of output and f represents the production function. Since the profit function as

specify above is the result of a maximization process, then by applying Hotelling's

lemma to the function the derived demands as well as the output supply can be obtained

as follows:






69


R/ = -x, (P, v, w, Z) < 0 derived demand for input one

n5 / = q(P, v, w, Z) > 0 supply function

The main difference between the derived demand obtained by way of the profit function

and that obtained from the cost function is that in the former output is allowed to vary

whereas in the latter it is held constant.

Once the derived demand is obtained, the own price elasticity of demand for the

input as well as the cross price elasticity of demand for an input can be computed. In the

case of the former, the elasticity will always be negative once the input is not considered

to be inferior. This is because in production theory, unlike the consumption theory, the

substitution and output effects always work in unison. In the case of the cross price

elasticity the sign is ambiguous. The ambiguity is due largely to the fact that in this case

the substitution and the output effects may not always move in the same direction. The

magnitude of both the own-price and cross-price elasticities is less predictable and

depends on factors such as the elasticity of substitution, the importance of the input in

terms of its share of total cost of production of the final good, and the price elasticity of

demand for the good being produced. According to Nicholson (1995) in general, the

own- or cross-price elasticity of demand for any input will be greater (in absolute value):

1) the larger the elasticity of substitution of that input for other inputs; 2) the larger is the

share of total cost represented by expenditures on that input; and 3) the larger is the price

elasticity of demand for the good being produced.

Thus, it is possible that given the profit function of the industries utilizing HFCS

as an input, and the assumption that such industries are operating within a perfect

competitive framework with a profit maximization objective, the derived demand curve






70


for HFCS can be obtained. However, in light of the difficulties surrounding the

estimation of a reliable profit function, such as obtaining cost sensitive information and

quantities used, a more straight forward approach can be taken in which the derived

demand is estimated directly on the basis of the explanatory variables suggested by the

above theory. These explanatory variables include the price(s) of the final

commodity(ies) and the respective prices of the various inputs used in the production

process. This latter approach will be used in this current study.














CHAPTER 4
CONCEPTUAL MODEL OF US SUGAR AND HFCS MARKETS

As the process of development takes place, sweeteners in general are increasingly

being consumed as raw materials and inputs of industrial production and to a lesser extent

by households as essential foodstuffs'5. Thus, in developed countries and regions such as

those of the EU, US, Canada and Japan, between 60%-70% of the sweeteners consumed

are in such goods as cakes, soft drinks and ice creams. However, in the developing

countries the situation is the reverse where only about 30/%- 40% of the consumption of

sweeteners is in the form of sugar containing products (Atlas, 1996). This has obvious

implications for the patterns of demand, the direction of technical progress and the

underlying scientific investigation. The purpose of this chapter therefore is two fold. First

we construct a conceptual model of the US HFCS and sugar markets focusing on the

derived demand nature of the demand for sweeteners. While the model is essentially

static, some attempts are made to examine the dynamics of the relationship between

HFCS and sugar. Second, the conceptual model is used to assess some of the likely trade

and welfare implications resulting from changes to the US sugar program and an

expanded HFCS industry within the framework of the proposed FTAA. In carrying out

this assessment two scenarios are examined. The first focuses on the likely impact

resulting from a partial liberalizing of the US domestic sugar program, while the second

considers the more unlikely situation of the complete abandonment of the program.


15 Industrial (manufacturing ) uses include those for food (bakeries and cereals, confectioneries, soft drinks
and canned and frozen dairy products) and non-food items such as ethanol.


71






72


The US Sugar and HFCS Subsectors

Figure 4.1, presents a flowchart of the US sugar and HFCS sweeteners'

subsectors. As stated earlier, HFCS is available mainly in a liquid form and is an almost

perfect substitute for liquid sugar, hence it is used in its entirety in the industrial sector.




Sugarcane Sugar beet Corn



SRaw Sugar J I | Wet Milling

g . .... ..... ...... .. .. .
By By " By
product
i, product s \ product product
Im pt...... ......... ...

Refined Sugar HFCS

Expt. Expt.
Industrial Demand




Consumer Demand




Fig. 4.1. Flowchart of the US HFCS and Sugar Subsectors



As evident from the Figure, useful byproducts are produced in the production of both

sweeteners. However, the revenue generated from the sales of the byproducts as a

proportion of the cost of the main input-- sugarcane, sugarbeet and corn, respectively--are

more substantial in the case of the production of HFCS.






73


The US Sugar Subsector

Focusing first on sugar, it should be noted that several studies have attempted to

model the US sugar market (Haley 1998b; Atlas, 1996; Bemirschka, Koo and Lou, 1996;

Marks 1993; Schmitz and Christian, 1993 ; Barros, 1992; Sudaryanto, 1987; Leu et al,

1987; Zietz and Valdez, 1986; Gemmill, 1976). Sudaryanto pointed out that many of the

studies fail to distinguish between direct and indirect consumption. They omit variables

explaining the derived demand characteristics of the consumption pattern. This study

recognizes the point made by Sudaryanto and others (Haley, 1998b; Barros, 1992) on the

need to disaggregate the market for sugars into derived and direct demand components.

However, a slightly different approach was taken in separating the market demand

components.

For convenience, a distinction is made between crystallized sugar (CS) and liquid

sugar (LS) . In this study, crystallized (tabletop) sugar include sugars used both by

households and those segments of the industrial sugar users, in which it is currently

technically impossible to substitute sugar with liquid HFCS. The importance of this

distinction will become evident later in the analysis. However, for the time being there is

some evidence suggesting that at this juncture, it is in respect of the demand for liquid

sugar that a structural shift in demand has taken place. Only the liquid HFCS competes

with sugar. As pointed out earlier, liquid sugar and liquid HFCS are regarded as almost

perfect substitutes (Marks, 1993; Barros, 1992). Further supporting evidence for this


16 Refined sugar is available in three main product categories: white granulated sugar, liquid sugar and
specialty sugars. Granulated sugar is the most common form of sugar used in the households. Liquid sugar
is a mixture of water and sugar that is preferred by some manufacturers and specialty sugars include icing
sugar, brown or yellow sugars. Chemically, there is no difference between liquid sugar and crystallized
(granulated and specialty).






74

assertion exists when consideration is given to the nature of the penetration of HFCS in

the sugar consuming industries. For example, Haley (1998b) points out that in the

beverage industry, where a relatively large proportion of liquid sugar was used, HFCS

has almost totally replaced the higher-priced sugar. Indeed, all major sweet drink

manufacturers in the US use HFCS as their main caloric sweetener.

Heuristically, one can therefore envisage the US sugar market as exhibiting two

distinct demand configurations in the early stages of the HFCS development--one

inelastic and the other being relatively elastic. The inelastic curve represents the demand

for crystallized sugar (CS) as shown in Figure 4.2a. The curve is assumed to be inelastic

due to the absence of any close substitute. It should be noted however, that the exclusion

of the non-caloric sweeteners from this model implicitly assumes no formidable






Price Price
a c



pk c






b L2

Fig. 4.2a Quantity Fig. 4.2b Quantity


Fig. 4.2. Hypothetical US demand for crystallized sugar (CS) and liquid sugar (LS),
respectively






75


challenge by that industry to this segment of the market. The second demand curve

represents the demand for liquid sugar (LS). As argued earlier, it is liquid sugar for which

HFCS is almost a perfect substitute (Figure 4.2b). The presence of this substitute (HFCS)

causes the demand curve for LS to become relatively elastic (rotating to the right as

shown in Fig. 4.2b) reflecting the fact that manufacturers are much more apt than

households to switch to a lower cost substitute because of the need to remain
17
competitive .

This characterization of the demand for sugar although novel, is nonetheless

anchored by evidence from the literature. For instance, Sudaryanto (1987) in partitioning

the US sweeteners market into a direct and derived demand for sugar found that the

elasticity of the derived demand was far greater than that for direct consumption. His

explanation was that since the consumption of final uses (direct) had approached

saturation, the responsiveness to a change in the price of sugar would be minimal

(inelastic). On the other hand, the demand for sugar in the food and beverage industry

was more elastic since many more substitution possibilities were available. The

Sudaryanto explanation of the differences in sugar demand configuration is somewhat

similar to that advanced by Thomas (1985). Specifically, Thomas noted that the US

market favored the development of the HFCS industry, since there existed high price

elasticity of demand for sugar among industrial sweetener users. These users, he

suggests, are far more ready than households to substitute one sweetener for another,



7 Economic theory holds that the price elasticity of demand for any good depends on many economic,
social, and psychological factors. Included among the economic factors are the availability of close
substitutes and necessity of a good. In general, necessities tend to have inelastic demand whereas goods
with close substitutes tend to have a more elastic demand (Mankiw, 1997)






76


given favorable price changes, and the technical characteristics of the individual

sweeteners.

Horizontally summing the two demand curves, CS and LS, in Figures 4.2a and

4.2b, gives us Figure 4.3, which represents the total market demand for refined sugar in

the US. Of significance in our model is the kinked nature of the resulting demand curve

ake. Point k on the curve, which represents the kink, can be interpreted as follows: At

prices above pk, there is no demand for liquid sugar--i.e. it is completely substituted for

by HFCS. Below the kink however, there is both a demand for crystallized sugar (CS)

and liquid Sugar (LS), the latter in competition with HFCS.





Price a





pk


\ e

\b
Quantity
Fig. 4.3. Hypothetical US aggregate demand for sugar



In Figure 4.4, the US domestic sugar supply curve (Ss) is introduced and the

established US domestic support (loan) price, PT, set at approximately US 23 cents per

pound refined sugar. As drawn, the Figure indicates a situation in which there still






77


remains a market demand for liquid sugar. Thus, X, and X3 indicate the quantities of

sugar produced and consumed, respectively. As such X3 - X1 represents the US tariff-rate

quota (TRQ), which cannot fall below 1.14 MMT. X2 represents the quantity of CS

demanded and X3 - X2 is the quantity of LS demanded. It should be noted that the

quantity of LS demanded does not represent the total demand for liquid sweetener that

would include the amount being satisfied by HFCS. Finally, Pw gives an indication of the



Price







PT


Pw

\ e
I\b
X1 X2 X3 Quantity


Fig. 4.4. Hypothetical US supply and demand for sugar



world market price in relation to the support price. Thus, the US sugar market might be

represented by the following equations:

Demand for CS: Dcs = f(Ps, PF, Z) (4.1)

Demand for LS: DLs = f(Ps, P(.),QH(.),PF,Z) (4.2)






78


Demand for Sugar: Ds = Dcs + Dis (4.3)

Supply of Sugar: Ss = Sc(.)+ S(.) (4.4)

Import Demand: Io = Ds - Ss > 1.256tonnes (4.5)

Equilibrium Condition: Ds = Ss + Io (4.6)

where:

Ps = domestic price of sugar in the US

PF = index of sugar and HFCS containing products

PH = price of HFCS

QH = quantity of HFCS supplied

Sc & SB = supply of cane and beet sugar, respectively

Z = a vector of other factors



The US HFCS Subsector

Turning attention to the HFCS subsector, the demand curve for HFCS is

conceptualized, as shown in Figure 4.5. Such a demand configuration is based on the

following grounds: 1) the observation that although HFCS is almost a perfect substitute

for liquid sugar (LS) it has always been sold at a discount to LS on a sweetness

equivalency basis (Fig. 2.3); and 2) the dual nature for the commodity, as discussed in

chapter one, which suggests that HFCS can only be substituted for sugar within a certain

range of the sugar demand function. It should be pointed out that above the controlled

price for sugar, PT, the particular shape assumed by the demand curve is very ambiguous.

This is due to the fact that the price of HFCS has never exceeded that of sugar or even

come close to it. If one is willing to assume the conclusion reached by Tanyeri-Abur






79


(1990) that if the price of HFCS should ever exceed that of sugar the industries would

revert to using sugar completely, then this would suggest that the demand curve kinks at

or just above the controlled price and become highly or even infinitely elastic. However,

if one were to accept the arguments put forward by Marks (1993, p. 80) that "a reversal

of the time path of the real US sugar prices would hardly likely bring about the same

substitution pattern in reverse" then one might conceive of the portion of the demand

curve above the controlled sugar price, tapers off as shown in Figure 4.5.



Price











HFCS demand


DH
Quantity

Fig. 4.5. Hypothetical US demand for HFCS



Several factors could account for a shift in demand for HFCS. As discussed in Chapter 3

in the section on derived demand, the demand for an input depends considerably on the

demand for the final good, hence any change in the demand for the latter will cause the

derived demand for the input to also shift. Thus, a shift in the demand for HFCS could

come about owing to any or a combination of the following: 1) changes in the prices of
come about owing to any or a combination of the following: 1) changes in the prices of






80


the other inputs used in producing the final good, especially the price of sugar; 2) an

increase in the number of products using HFCS; 3) seasonality ; 4) changes in the price of

the final commodity; and 5) an improvement in technology which improves the

substitutability of HFCS for sugar. Consequently, the derived demand for HFCS can be

represented by the following general specification:

Demand for HFCS: DH = f(PH, Ps, PF, T, Z) (4.7)
where:

Du = demand for HFCS

PH = price of HFCS

Ps = domestic price of sugar in the US

PF = price index of sugar and HFCS containing products

T = technology (trend variable)

Z = a vector of other factors

With respect to the supply, it is important to note that the consensus of industry

experts is that the HFCS industry operates in an oligopolistic framework (Polopolus and

Alvarez, 1991, Thomas, 1985). A consequence of this is that even if the exact profit

function of such an industry were available it would not be possible to obtain the input

supply function of the industry by way of the duality theory, since the underlying

assumption of the theory is the existence of a perfectly competitive framework.

Moreover, in light of the prevailing circumstances under which the HFCS is produced,

notably a kind of captive market resulting from the restrictive policies of the US sugar

program and the implicit floor price of sugar, it is argued that there is no supply function

for the industry but rather a reaction function. This reaction function can be viewed in the

context--although not identical--of the Stackelberg model in which one producer of the






81

substitutes has market intelligence and knows the reaction function of his competitor but

the other is naive. Consequently, the producer with the market intelligence take

advantage of his competitor by supplying an amount at a price which will maximize his

profit at the expense of his competitor. Here, the HFCS producers could be regarded as

the sophisticated suppliers, while those supplying sugar are regarded as naive. Taking this

analogy a step further, as in the case of the Stackelberg model in which it is possible for

one competitor to force another out of the market by sufficiently differentiating the

product, in a similar manner such an opportunity exist for HFCS to force liquid sugar out

of the market.

More specifically, it is assumed that the profit maximizing objective function of

the HFCS suppliers can be represented as follows:

7 = (PH - CH) *DH(.) (4.8)

where i represents the maximum industry profit; PH is the price of HFCS; CH is the cost

of producing HFCS and DH is the demand for HFCS as defined in equation 4.7.

Assuming other factors remain constant, the reaction function of the HFCS suppliers is

obtained by differentiating the profit function with respect to the price of HFCS, setting it

equal to zero and solving for the price of HFCS. The reaction function will therefore

reflect the best price to charge for HFCS given the price of sugar and the other exogenous

factors that affect profit of the HFCS industry.

Within the framework of the above conceptualization, it is assumed that: 1) the

demanders of HFCS are price takers; and 2) that the marginal cost of producing the

HFCS remains constant over the entire range of output. The first assumption seems

plausible since in comparison to the number of firms producing HFCS there are large






82


number of firms that use HFCS as input into their production process. With respect to

the second assumption, this again seems plausible given that, the industry is highly

capital intensive and that there currently exist a considerable amount of excess capacity

within the HFCS industry. In addition, the amount of corn used in the industry is less

than 10% of the overall US corn production, and consequently a moderate increase in

demand for HFCS is not expected to have any major impact on the market price of corn.

Finally, the industry is highly capital intensive and as such one might expect that the cost

of an additional unit energy will remain fairly constant. On this basis, and in the absence

of concrete cost of production data, the short-run marginal cost curve of the industry (the

sum of the marginal cost over the variable cost) is conceptualized as shown in Figure 4.6.

The curve is drawn showing a section which is nearly perfectly elastic (constant marginal

cost) over some quantity range and becoming vertical as output capacity is reached.


Price




SMCHI


MCH2
P2 -

Quantity


Fig. 4.6. Shift in hypothetical HFCS short-run marginal cost curve



In Figure 4.6, SMCHI represents the industry short-run marginal cost function in

period one and SMCH is the short-run marginal cost curve in period two. As drawn the

shift in the marginal cost curve represents an overall reduction in the cost of production.






83


The shut down price is represented by P. The main factors which appear to influence the

marginal cost of producing HFCS are: 1) the net price of corn (price of corn less the price

of the byproducts); 2) the price of enzymes used in the conversion process; 3) the cost of

energy and capital; 4) labor cost; and 5) improvement in technology.

The hypothetical US HFCS market, as depicted in Figure 4.7, is obtained by

combining Figures 4.5 and 4.6. This representation is consistent with the view of industry

experts that the producers operate within a basically oligopolistic market structure,

Price




IHFCS demand

PT

PHI
Pm

HFCS SMC

DH

QHI QI Quantity
(sugar equiv. basis)

Fig. 4.7. Hypothetical US HFCS short-run marginal cost and demand configuration



implying among other things that demand and supply of the product do not determine the

pricing of the commodity. Moreover, because of problem with storage, production

closely matches that of the quantity demanded at the established price (See Chapter 2). In

Figure 4.7, it is assumed that the general pricing policy followed is that discussed earlier,

whereby the producers of HFCS having determined the best price for the commodity,

based on their reaction function and knowledge of the sugar subsector, supply a quantity






84

of the product which closely matches that demanded. This pricing strategy implies,

among other things, that the quantity supplied and demanded does not occur at the

intersection of the industry short-run marginal cost curve and the demand curve, but

rather at the point where the demand curve intersect with the established price. The

significance of this as far as estimation is concerned is that price and quantity

observations for the HFCS industry traces out the demand for HFCS. To the extent that

the prices received by the producers of HFCS is above the average cost of production,

they will be receiving economic rent and are in a position to adjust price of HFCS within

a certain range, in order to stave off the challenge from sugar in order to maintain or

increase market shares.

Based on the above line of argument, it is therefore theoretically possible to

estimate the derived demand function for HFCS using shifters and the observed price and

quantity data. This is done in the next chapter.

With respect to the long-term prospect for the supply of HFCS, there is

consensus that the industry is expected to benefit much more in the future than the sugar

industry owing to the fact that there exist both internal as well as substantial external

economies of scale (Marks, 1993; Thomas, 1985). Economic theory holds that in cases

where there are internal economies, one can expect a reduction in the average cost of the

particular firm as it increases its output. It also suggests that in the case of external

economies of scale the average cost of the typical firm declines as the output of the

industry expands (Lindert and Pugel ,1996). These authors noted that external scale

economies can come about in a variety of ways, including: 1) specialized services for the

industry; 2) the spill over effects of new knowledge about the product and production






85

technology; and 3) specialized machinery. They further suggest that if scale economies

are substantial over a large range of output, then a likely result is that a few firms will

expand their operation in order to reap the scale economies, giving rise to oligopoly

market structure. As it is well known, firms in an oligopoly can influence prices, can earn

economic profit and can vary the level of competition. More importantly, the existence of

scale economies has been shown by authors such as Krugman (1979) to provide as solid a

basis for trade as is comparative advantage. This line of argument forms the basis of the

New International Trade Theories (NITTs).

Much of the above is true for the HFCS industry, which has benefited

substantially from path breaking gains in biotechnology and processes such as

fractionation technology. Indeed, there is the view that the HFCS industry represents the

successful marriage of two industries: the wet corn processors and the enzymes industry.

In light of the above and the fact that the industry is still relatively young and continues

to benefit from technological improvements, the study assumed a long-run HFCS

industry average cost reflecting external economies as depicted in Figure 4.8 .

Noticeable is the shape of the cost curve (LRAC) implying possible gains from

increasing return to scale. The figure also shows two hypothetical demand curves Dus

and DFTA, reflecting the possible increase in the demand for HFCS as a result of the

formation of a free trade area, emphasizing the possible gains to be made from the

lowering of the average cost. The assumption of the long run shape of the HFCS industry

cost function forms a crucial part of the subsequent analysis of the trade implications

within the framework of the proposed FTAA.






86




Price









LRAC

Dus DFrTA

Quantity

Fig. 4.8. Hypothetical HFCS industry average costs, reflecting external economies





The Interrelationship between the US Sugar and HFCS Markets

As indicated earlier there is a unique relationship between the US sugar and

HFCS markets. Sugar can substitute in all cases for HFCS but HFCS can only replace

sugar in a selected number of uses--albeit in such cases it has proven to be quite an

effective substitute (Barros, 1992). The implication is that it is only one segment of the

demand curve for sugar that is affected by changes in the demand and supply conditions

within the HFCS subsector. As indicated earlier in relation to Figure 4.3, the kink in the

sugar demand curve is of importance since it indicates the price above which there would

be no demand for liquid sugar. Several factors affect the positioning of the kink. This

point is not stationary but shifts in relation to changes to these factors. The primary

factors determining the positioning of the kink are: 1) the availability of other substitutes;

2) the price of HFCS which itself is influenced by several factors including the price of






87


corn, the prices of the by-products, the price of domestic sugar; and 3) the supply

behavior of HFCS. These three factors are discussed below in highlighting some of the

dynamics of the sweetener market.

First, assuming other factors remain constant, it can be expected that an increase

in the number of available substitutes and the closer the substitutes on a per unit

sweetness equivalency basis the lower will be the positioning of the kink. This follows on

the assumption that there does not exist any switching cost for the manufacturers (user of

sweeteners) in utilizing one input over another. This latter assumption however is not

entirely accurate since such costs do exist in one form or the other. However, the general

principle applies in cases where such costs are negligible compared with the anticipated

increase in profits resulting from the change.

Second, and of greater importance is the relative price movement. For instance a

decrease in the price of HFCS is expected to cause the kink in the sugar demand curve to

shift downward, from ke to k'e', implying a lower level trigger price that sugar would

have to be sold in order for it to become an effective HFCS substitute (Figures 4.9a and

4.9b). The converse situation would also hold, that is, an increase in the price of HFCS

relative to the price of sugar shifts the kink, ke, upwards, implying an increase in the

trigger price at which sugar becomes an effective substitute for HFCS. These movements

postulated are consistent with economic theory which holds that a reduction in the price

of one commodity that is a close substitute for another commodity will cause a downward

shift in the demand for the substitute commodity. Thus, the change in the relative price of

HFCS only caused that segment of the sugar demand, (below the point of the kink in

Figure 4.9a, which is ke) to shift, since it is this segment which represents the demand for






88


liquid sugar, the HFCS substitute. It should also be noted that, other factors remaining

constant, the movement of the segment of the curve ke occurs along the dotted portion of

the demand curve akb as shown in Figure 4.9a for the case of a shift from ake to ak'e'.

The nature of this movement has important trade implications. For one thing, it implies

that lowering the US support price for sugar (PT) would induce only a slight increase in

the quantity of crystallized sugar demanded if the final support price is still above the

kink. This postulated movement is contrary to popular view of a substantial increase in

sugar demand associated with sugar price reduction in the US (Atlas, 1996; Schmitz and

Christian, 1993; Sudaryanto, 1987). This is unlikely, unless the price of sugar is lowered



Price Price


Ds s




k



S . PH2
--^J-_[^_____PI_.__, __



\ \DLs
I DH
\b e'
XI X2 X3 Quantity YiY2 Quantity
(Sugar equivalency)
Fig. 4.9a Fig. 4.9b



Fig. 4.9. Hypothetical supply and demand for sugar, and hypothetical demand for HFCS,
respectively






89


substantially so that it falls well below the point of any kink in the demand curve. And, as

noted earlier, the location of the point of the kink is a function of the dynamics of the

rapid economic and technological changes occurring within the HFCS industry.

The third factor mentioned above which influences the positioning of the kink is

the supply behavior of HFCS. This factor is closely related to the previous two factors

discussed. Marks (1993) in alluding to the pattern of the substitution between HFCS and

sugar asserts that it is a function of both the demand for sweeteners and the supply

behavior of HFCS. Elaborating on the latter factor, he postulated that a reversal of the

time path of the real US sugar prices would hardly likely bring about the same

substitution pattern in reverse. His claim for the likely failure of the reverse pattern of

substitution to occur is based on the view that "the HFCS industry is in place with

hundreds of millions of dollars of sunk costs to protect and the advantage of learning-by-

doing and other scale economies" (p. 80). The implication of his assertion is that due to

the advancement made in the technology of producing HFCS, the benefit of product

development and the need to protect investment, the shifts in the sugar demand could be

structural. Moreover, on the demand side there also exist switching costs associated with

changing from one form of sweetener to the other. For example, in switching from liquid

sugar to HFCS certain costs would have been incurred both in modifying recipes and in

putting in place certain physical infrastructure such as storage facilities. While these

factors do not guarantee complacency on the part of the HFCS supplier, they imply a

lowering of the kink and that the price of sugar would have to fall substantially to bring

about a reversal of the current trend of sweetener usage.




Full Text
Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
DYNAMICS OF THE US SUGAR AND HFCS MARKETS:
IMPLICATIONS FOR THE CARICOM SUGAR PRODUCERS WITHIN THE
FRAMEWORK OF THE PROPOSED FTAA
By
Edward Anthony Evans
December 1999
Chairman: Carlton G. Davis
Major Department: Food and Resource Economics
To cope with growing regionalism, many developing countries must choose
between alternative and competing integration strategies. The Caribbean Community
(CARICOM) is faced with such a dilemma. The United States and the European Union
have proposed the formation of separate free trade areas (FTAs) with the subregion with
a tentative implementation date of 2005. The consensus emerging in CARICOM favors
participation in the proposed Free Trade Area of the Americas (FTAA)-a potential
grouping of 34 disparate countries in the Western Hemisphere. While the anticipated

benefits of such a group are unquestionable, there will be losses at the sector and industry
levels.
The CARICOM sugar industry is of considerable importance to the subregion but
could be adversely affected by the FTAA. First, it is possible that CARICOM could
violate the Lome Agreement through FTAA membership, denying continued access to
the lucrative EU sugar market. Second, CARICOM sugar producers could be forced to
x


Elasticities
119
little or no impact on the quantity HFCS being used. It now seems plausible that a
considerable reduction in the price of sugar would be needed to impact on the quantity of
HFCS currently being demanded
Fig. 5.3. Dynamic path of adjustment of cross-price elasticity of HFCS with respect to
sugar, 1982-98
This latter view is also consistent with the dynamic path observed in relation to
the cross-price elasticity as shown in Figure 5.3. Figure 5.3 shows clearly that the
adjustments over time has been in a manner such that the impact that a change in the
price of sugar was having on the demand for HFCS has been declining. If so, the extent
of the substitution of HFCS for sugar has been largely completed and the use of sugar


9
HFCS since its inception has been lower than that of sugar on a sweetness equivalency
basis. Nevertheless, the adoption of this product had been cautious. It took a decade for
the major soft drink industries to convert from liquid sugar to HFCS. On the demand side
the adoption of the product has been aided by the producers of HFCS working in concert
with the users, providing technical assistance and adapting the product to the special
requirements of the various clients. It is in this context that HFCS-55 was developed as a
more effective substitute for liquid sugar than HFCS-42 in the soft drink industry.
Likewise on the supply side there has been a constant effort aimed at improving the
technology involved in the production of the commodity, with a view to producing an
effective substitute sweetener at the lowest cost. The implication is that the industry is
evolving and as such the parameters of the industry might be changing. This time-varying
issue must be recognized in characterizing the industry in any analytical model.
Ward and Tilley (1980) pointed out that in markets, such as the one we are
concerned with where significant adjustments are occurring, the parameters are likely to
be time-varying. They noted that technological changes result in structural changes in
economic phenomenon, giving rise to varying parameters. They cautioned that a failure
to take into consideration such variations in the estimated parameters could cause serious
errors in projections and policies drawn from them.
This issue of the degree to which the HFCS industry in the US can be considered
mature is therefore important from an estimation point of view. It is also important in the
context of this study since the dynamic paths of adjustments of various economic
parameters have such implications for the CARICOM sugar industry such as: 1) the
extent of saturation of HFCS in the US market and by extension the size of the US


136
worthwhile approach in the future as more data become available. Consideration might
also be given to developing a dynamic partial equilibrium net trade sugar and HFCS
simulation model for the FTAA countries and the rest of the world (ROW). This study
focused on only the sweetener sector. The inclusion of other sectors in an analysis could
serve to better improve the understanding of the issues confronted. Such an approach
requires additional data and time not available within the confines of this study.
Only one aspect of the theoretical model was investigated empirically. Although
this is in keeping with the specific objectives identified in chapter 1, a more complete
model which facilitates testing of some of the theoretical findings is needed. This study
did not estimate the kink in the sugar demand curve that played an important role in
modeling the impact of the likely changes of the US sugar program. This too needs to be
studied. The location of this kink is a function of the dynamics of the rapid economic and
technological changes occurring with the sweetener market and should be studied in a
dynamic framework. A game theoretical analysis of the behavior of the HFCS suppliers,
may provide a clearer understanding of how suppliers might respond to a freer trading
environment.
A third shortcoming had to do with the exclusion of artificial sweeteners
noncaloric sweeteners such as saccharin, aspartame, acesulfame-K. The rationale given
for their exclusion was that there is a certain amount of health related issues surrounding
the use of these products. While to some extent this is true, within recent years various
health authorities have cleared most of the products. Such sweeteners are increasingly
being used in the so-called diet foods and soft drinks and are usually 200 to 300 times
sweeter than sugar. While these sweeteners have not yet made significant inroads into the


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosopl
Carlton G. Da
Distinguished Professor of Food and
Resource Economics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
ax R. Langham
Professor of Food and Resource
Economics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Ronald W. Ward
Professor of Food and Resource
Economics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation andjsjiilly-a^equate, in scppeand quality,
as a dissertation for the degree of Doctor o
2rt D. Emerson
Professor of Food and Resource
Economics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Andrew Schmitz
Eminent Scholar in Food and Resource
Economics


90
Related to the issue of the supply behavior of HFCS and the kink is the structure
of the HFCS industry. Specifically, the degree of market power which exists in the
industry can influence the positioning of the kink. In general, other factors remaining
constant, it can be assumed that the less the intensity of competition among the HFCS
producers (the higher the degree of market power among the producers) the greater will
be their influence on the positioning of the kink. This is evident since it implies a greater
probability that the producers of HFCS will be able to influence the price of HFCS
covertly, which in turn will influence the positioning of the kink.
Trade Implications of Changes in the US Sugar Program within an FTAA
Given the conceptual model developed for the US caloric sweeteners industry the
question now addressed is "What are the implications of changes in the US sugar
program within the context of an FTAA, for the rest of the sugar producing members, in
particular those of the CARICOM?" This question attains some primacy based on the
objectives of this study. In this section, an attempt to address this matter using the
conceptual model developed. The analysis starts with a brief review of the concept of a
free trade area (FTA) as background to the subsequent analysis.
Free Trade Area
In the literature, a free trade area in the strictest sense is defined as an association
of countries that have agreed to remove all trade barriers among themselves, but retain
their individual national barriers against trade with third countries. In such an area it
would still be necessary to have custom officials at the borders to ensure that goods from
third countries do not enter through lower barrier members and trade within the higher


TABLE OF CONTENTS
page
ACKNOWLEDGMENTS iii
LIST OF TABLES viii
LIST OF FIGURES ix
ABSTRACT x
CHAPTERS
1 INTRODUCTION 1
Statement of Problem 3
Problematic Situation 3
Specific Problem 7
Hypotheses 10
Objectives 11
Methodology 12
Organization of the Study 13
2 BACKGROUND 14
Free Trade Area of the Americas (FTAA) 14
Review of World and Western Hemisphere Sugar Markets 17
World Production and Market TrendsSugar 17
Western Hemisphere Production and Market TrendsSugar 21
Importance of the Sugar Industry to CARICOM Sugar Producers 23
Socioeconomic Importance 23
The EU/Lome ACP Agreement 25
The US Sugar Program and Policy 27
Review of World and Western Hemisphere HFCS Markets 31
World Production and Market TrendsHFCS 33
Western Hemisphere Production and Market TrendsHFCS 34
3 LITERATURE REVIEW AND THEORETICAL BACKGROUND 39
Previous Empirical Studies 39
Sugar and HFCS Related Studies 39
v


7
world's lowest-cost producer of sweeteners. The world average cost of production is
20.91 cents per pound. It is 13.73 cents per pound in the most efficient sugar producing
countries. In comparison the US produces HFCS in 1994/95 at an average cost of only
10.60 cents per pound dry weight equivalence (Haley, 1998a). HFCS competes directly
with sugar in most industrial uses. In a situation of no foreign tariffs, duties, taxes,
quotas, and other border barriers against com sweetener imports, the US could export its
technology and/or com sweetener product to some of the higher-cost markets within the
FTAA. The US would also benefit as the world's largest and cheapest producer of com-
the main input used in the production of HFCS. The potential of HFCS to displace sugar
has already been seen in the US and to a lesser degree in Canada. And, though it has not
yet adversely affected the other sugar producing and consuming countries, those in the
planned FTAA are likely to feel its impact.
Specific Problem
This study is concerned with the threat that an expanded HFCS industry could
pose to the CARICOM sugar industry within the framework of a liberalized trading
environment. The US sugar industry and its sugar program have been adequately dealt
with in the literature. The same is not true in the case of the HFCS industry. Not much
has been written on the likely impact that an expanded HFCS industry could have on
world and regional prices, production, consumption and trade patterns of sugar. Also, the
dynamics of the US sugar-HFCS interrelationships and the implications of such
relationships for trade in an extended market have not been fully explored. Thirdly, as
noted by Tanyeri-Abur (1990) there appears to be an absence of studies to form a basis
for studying the trade implications of an expanded HFCS industry. As a consequence,


54
relationship. In such situations, the standard OLS procedure does not produce estimates
that are consistent (Kennedy, 1998). Also, in these circumstances when there are more
than one cointegrating relationship there is usually a certain degree of ambiguity
surrounding the interpretation of the estimated cointegrating vectors (Johnston and
Dinardo, 1997; Kennedy, 1998). Kennedy (1998) further points out that the interpretation
of multiple cointegrating vectors can be frustrating. He refers to a practice among some
researchers, in dealing with this issue, of ignoring those cointegrating vectors that seems
not to make good economic sense as tantamount to imposing slightly false restrictions to
improve mean square error.
Fourthly, according to Johnston and Dinardo (1997) VARs have serious
limitations as a tool for the analysis of economic systems and are susceptible to the
problem of vanishing degrees of freedom, since the number of unknown coefficients can
rapidly approach the available sample size. An implication of this is that researchers
wishing to use this technique, but having limited data (observations), are forced to be
quite parsimonious in their model specification. However, this could be disastrous since
the failure to include an important variable could result in absence of any detection of
cointegrating relationship among the variables (Kennedy, 1998). On the other hand, as
pointed out by Johnston and Dinardo (1997), as more variables are added to the VARs,
problems arise in testing the number of cointegrating relationships. This comes about
because the test statistics are nonstandard distribution and require simulation. At present
the available tables can only account for eleven variables.


YEAR
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
APPENDIX
STATE AND SMOOTEHED ESTIMATES
STATE ESTIMATES
INTERCEPT
HFCS
SUGAR
FLOUR
SOFT DRINKS
2.35060
-1.15039
0.54531
0.36625
4.37817
2.35060
-1.15039
0.54531
0.36625
4.37817
2.35060
-1.15039
0.54531
0.36625
4.37817
2.35060
-1.15039
0.54531
0.36625
4.37817
2.35060
-1.15039
0.54531
0.36625
4.37817
1.77204
-0.75027
0.31236
1.01283
3.21563
2.19728
-1.04408
0.47241
0.69941
3.84027
2.69823
-1.58493
0.87955
-0.54212
5.23126
2.93150
-1.76100
1.05752
-2.34914
6.68421
2.71913
-1.89913
1.26641
-2.91761
6.98483
2.71488
-1.90004
1.26872
-2.92442
6.98753
3.39063
-2.30996
1.45474
-2.84288
6.89725
3.54696
-3.14583
2.15802
-2.66301
6.80810
0.40229
-2.37257
2.45879
-4.31041
6.26375
0.43293
-2.36084
2.43532
-4.19048
6.18130
1.41812
-2.33124
2.12672
-4.10239
6.47785
1.94168
-1.76743
1.48854
-4.29419
6.43354
2.58921
-1.66569
1.21797
-4.34319
6.19456
4.72430
-1.40368
0.35185
-3.49925
5.94294
7.56107
-2.34393
0.32677
-1.66712
5.41541
7.50306
-1.93869
-0.01940
-1.48805
5.35799
7.97624
-1.63964
-0.43357
-0.98789
5.34171
138


57
then it is clear that the effect of B2 which is equal to (X2 + 2ct3Xt is not a constant. In the
case of aggregation as source of parameter variation, Ward and Myers (1979) cites as an
example the fact that over time, the relative importance of microeconomic agencies will
change and that such changes might not get reflected in the aggregate weight. In addition,
they noted that aggregated(macro) variables are discrete (Le., indexed as discrete time
points) whereas the underlying microvariables may be continuous, providing another
reason why parameter variation are likely to occur in macromodels.
In the literature, parameter variations are classified under the two broad headings
of stochastic and nonstochastic parameter variation. Each of these classifications is
further broken down into sub-categories. Hence, nonstochastic parameter variations,
which are caused by structural change in the economic phenomenon being studied, can be
divided into: 1) discrete variation or switching regression; and 2) systematic variation.
Likewise, stochastic models are divided into: 1) those which are stationary and are
referred to as random coefficient models; and 2) those which are nonstationary and are
termed sequential (Markovian) models13.
The State-Space Framework
The State-Space Model
Kennedy (1998) states that the state-space model can be regarded as a
generalization of the linear regression model and provides a unifying framework for all
dynamic linear models used in econometrics. A similar view is expressed by Harvey
13 For additional information 1 the various aspects of these types of models see Ward and Myers, 1979
and Nichols and Pagan, 1983.


142
Haley, S. US and World Sugar and HFCS Production Costs. Sugar and Sweetener
Situation and Outlook Report, USD A, Economic Research Service, SSS 223
May, 1998a.
Haley, S. Modeling the US Sweetener Sector: An Application to the Analysis of Policy
Reform. International Agricultural Trade Research Consortium (IATRC),
Working Paper, # 98-5, 1998b.
Hall, S., K. Cuthbertson, and M. Taylor. Applied Econometric Techniques. Hemel
Hempstead, UK: Philip Allan, 1992.
Hannah, A. and D. Spence. The International Sugar Trade. New York: John Wiley &
Sons, 1997.
Haraksingh, K. The Future of Preferential Trade and Sugar. Caricom Perspective.
67(June 1997): 10-13.
Harvey, A. Application of the Kalman Filter in Econometrics, in T.F. Bewley, (ed.),
Advances in Econometrics: Fifth World Congress. Cambridge: Cambridge
University Press, 1(1987): 285-313.
Harvey, A. Time Series Models (2nd ed.), Cambridge, Massachusets: The MIT Press,
1993.
Harvey, A. Trends, Cycles and Autoregressions. Economic Journal 107(1997): 192-
201.
Harvey, A., S. Henry, S. Peters, and S. Wren-Lewis. Stochastic Trends in Dynamic
Regression Models: An Application to the Employment-Output Equation. The
Economic Journal, 96(1986): 975-85.
Helpman, E. "Politics and Trade Policy." NBER Working Paper No. 5309, Cambridge,
MA, 1995.
Hildreth, C. and J.P. Houck, Some Estimators for a Linear Model with Random
Coefficients, Journal of the American Statistical Association, 63(1968): 584-
595.
International Sugar Organization (ISO). Quarterly Review. London, Various Issues
James, K. Prospects for the Caribbean Sugar Agro-Industry. Paper presented at the
UWI Ag.-50 Workshop on Agriculture in the Caribbean Issues and Challenges,
Trinidad and Tobago 16-23 August 1998.
Johansen, S. Statistical Analysis of Cointegrating Vectors. Journal of Economic
Dyaamics and Control, 13(1988): .231-254.


LIST OF FIGURES
Figure page
Fig. 2.1. World and US refined and raw real sugar prices, 1985-98 (1992=100) 18
Fig. 2.2. Changes in US per capita consumption of selected sweeteners, 1975-97 36
Fig. 2.3. Real HFCS and refined sugar prices, 1982-98 (1992=100) 37
Fig. 4.1. Flowchart of the US HFCS and Sugar Subsectors 72
Fig. 4.2. Hypothetical US demand for crystallized sugar (CS) and liquid sugar (LS),
respectively 74
Fig. 4.3. Hypothetical US aggregate demand for sugar 76
Fig. 4.4. Hypothetical US supply and demand for sugar 77
Fig. 4.5. Hypothetical US demand for HFCS 79
Fig. 4.6. Shift in hypothetical HFCS short-run marginal cost curve 82
Fig. 4.7. Hypothetical US HFCS short-run marginal cost and demand configuration 83
Fig. 4.8. Hypothetical HFCS industry average costs, reflecting external economies 86
Fig. 4.9. Hypothetical supply and demand for sugar, and hypothetical demand for HFCS,
respectively 88
Fig. 4.10. Trade implications under partial liberalization of US sugar program 93
Fig. 4.11. Hypothetical HFCS industry long-run demand and average cost 95
Fig. 4.12. Trade implications under completely liberalized US sugar program 100
Fig. 5.1. Dynamic path of HFCS intercept over time, 1982-98 115
Fig. 5.2. Dynamic path of adjustment in HFCS own-price elasticity, 1982-98 117
Fig. 5.3. Dynamic path of adjustment of cross-price elasticity of HFCS with respect to
sugar, 1982-98 119
IX


Elasticities
117
superior input to sugar in some applications such as in the coloring of certain baked
products. Moreover, as pointed out by Marks (1993) many of the users, having been
convinced of the attributes of HFCS and its favorable long-term pricing
Years
Fig. 5.2. Dynamic path of adjustment in HFCS own-price elasticity, 1982-98
prospects relative to sugar, have made investments in the necessary machinery and
equipment for using and storing HFCS. Consequently, the responsiveness to a change in
the price of the product is expected to fall since the importance of the product has
increased and the degree of liquid-sugar as a substitute should have declined. Finally, the
decline in the elasticity is consistent with our earlier suggestion of a possible maturity
with regard to substitution possibilities. This is so since if one were to accept the


61
estimate this state are the y,.. The statistical algorithm used to derive these state
parameters is the Kalman filter. Yt is the measurement equation. The parameters on the
right hand side of the transition equations (3.7a to 3.7c) or more precisely the system
matrix in equation 3.9, are described as the hyperparameters. They are regarded as being
fixed and known although they may change over the period in question. Engle and
Watson (1987) points out that on the basis of empirical experience and theoretical
arguments, in most cases these transition equations should to have unit root.
Concerning equations 3.7b to 3.7c, it should be noted that in the special case
where o^ o^2, and ov2 equal zero, the hyperparameters remain constant and equation
(3.7) collapses to the standard regression equation (3.6) with the deterministic trend and
fixed regression parameters. Harvey (1997) also showed that if oe2 = o^2 = ov2 = 0 and if
o^2 > 0, then the first diiference formulation is obtained as:
Ayt p + XS,Ax,_j + rj, (3.10)
Kalman Filter
The Kalman filter is simply a statistical algorithm, which allows certain
computation to be carried out for a model cast in the state-space form. Following the
approach of Harvey (1987) the typical model for which the Kalman filter is applied takes
the form of:
Y, = xA + s, (3.11)
5t = Gt5t-i + where as previously noted, equations (3.11) and (3.1 la) comprise the state-space system
in which the former is defined as the measurement equation and the latter as the dynamic


143
Johnston, J. and J. Dinardo. Econometric Methods, 4th edn. New York: Mcgraw-Hill,
1997.
Kalman, R.E. A New Approach to Linear Filtering and Prediction Problems,
Transactions ASME, Journal of Basic Engineering, 82(1960): 35-45.
Kalman, R.E. and R. S. Buey, New Results in Linear Filtering and Prediction Theory,
Transactions ASME, Journal of Basic Engineering, 83(1961): 95-103.
Kennedy, P. A Guide to Econometrics. 4th edn. Cambridge, Mass.: The MIT Press, 1998.
Kennedy, P. L., R. Harrison, M. Piedra. "Analyzing Agribusiness Competitivenes: The
Case of the United States Sugar Industry" International Food and Agribusiness
Management Review, l(1998):245-57.
Krugman, P. R. Increasing Returns, Monopolistic Competition and International Trade.
Journal of International Economics, 9 (1979)469-79.
Kwaitowski, O., P. Phillips, P. Schmidt, and J. Shin. Testing the Null Hypothesis of
Stationarity Against the Alternative of a Unit Root Journal of Econometics,
54(1992):. 159-78.
Landell Mills Commodities (LMC) Annual Review of the World HFCS Industry, in
Sweetener Analysis, New York, NY: LMC International, December, 1998.
Langham, M. and M. Mara. More Realistic Single Equation Models Through
Specification of Random Coefficients. Southern Journal of Agricultural
Economics, 5(1973)161-66.
Leu, G., A. Schmitz, and R. Knutson. "Gains and Losses of Sugar Program Policy
Options." American Journal of Agricultural Economics, 69(1987):591-602.
Lindert, P. H. and T. A. Pugel. International Economics. 19th edn., Chicago: Erwin, 1996.
Lopez, R. A., and J. L. Sepulveda. Changes in the US Demand of Sugar and
Implications for Import Policies. Northeast Journal of Agricultural Resource and
Economics. 14 (1985): 177 82.
Lord, R. Sugar: Background for 1995 Farm Legislation US Department of
Agriculture, Economic Research Service, Commodity Economics Division.
Report #711 June, 1995.
Lucas, R. Econometric Policy Evaluation: A Critique," in The Phillips Curve and Labor
Markets, K. Brunner and A Meltzer, eds., Carnegie Rochester Conference Series
on Public Policy, Amsterdam: North-Holland, 1 (1976): 19-46.
Mankiw, N. G. Principles of Economics. New York: The Dryden Press, 1997.


105
where LQm is the log of the annual quantity of HFCS used in period t; LRPhi is the log of
the real price of HFCS; LRPst is the log of the real price of refined sugar; LRPd, is the log
of the real price of soft drinks; LRPFt is the log of the real price of flour; and et is the
error term.
Of major concern is the dynamic path of adjustments of the coefficients over time.
As discussed in Chapter 3, equation (5.1) can be expressed in the state-space framework,
which then allows for estimation, prediction and signal extraction to be carried out using
Kalman filters and a smoothing algorithm. In the state-space framework and allowing for
a stochastic trend and time-varying parameters equation (5.1) can be represented as
follows:
LQhi = Pt + PuLRPh, + P2tLRPst + P3tLRPot + P^LRPfi + et (5.2a)
Pt = Pt-i + Yt-i + fit (5.2b)
Yt = Yt-i + St (5.2c)
Pit = P>(t-i) + con for i = 1,2,3,4 (5.2d)
where, pt is the stochastic trend, and et, fit, St and co¡( are all disturbance terms assumed
to be distributed normally and independently of each other, with zero mean and constant
variance. Equation (5.2a) is the measurement equation and equations (5.2b 5.2d) are the
transition equations. In matrix notation the above set of equations could also be
represented as:
Y, = (1 0 Z,')a, + 8,
(5.3a)


42
US, he pointed out that analysts in previous sugar studies ignored the derived nature of
the demand for sugar by treating all sugars as if they were consumed directly. Utilizing
Fouriers flexible functional forms he separately estimated, a derived demand and direct
demand for sugar. He then computed an aggregate elasticity of demand for sugar using
the weighted average of elasticities obtained from the two models. While he did take into
consideration the impact of the possible substitution between HFCS and sugar in
estimating the various demand functions, he followed previous studies and utilized a
composite price index of HFCS, glucose and dextrose. The results of his analysis
indicated that in the United States, sugar for final consumption is far less elastic than
sugar used as intermediate input. Sudaryanto's results suggest that consumption for final
uses had reached the near saturation point and a change in sugar price had little effect on
consumption. In contrast, the demand for sugar in the food processing industry was more
elastic because of available substitutes. The result of his trade model showed that
unilateral trade liberalization in the US market would force an increase in US imports by
114% and would be accompanied by a decrease in both production and supply price by
8.3% and 11.2%, respectively. Complete removal of protection in the EEC was estimated
to cause a reduction in supply price and production in that region of 13.2% and 5.6%,
respectively.
Rendleman and Hertel (1989) examined the impact of policy changes on sugar
and HFCS within a general equilibrium framework. The results of their analysis showed
that any substantial increase in the production of com sweeteners would cause an
increase in the production of byproducts, which would have the effect of lowering the
byproducts prices and thus raising the cost of production of com sweeteners. They


CHAPTER 6
SUMMARY, POLICY RECOMMENDATIONS AND SUGGESTIONS FOR
FURTHER RESEARCH
Summary and Conclusions
This dissertation had two general objectives. The first objective was to
qualitatively assess the potential economic impact that likely changes to the US sugar
program, and an expanded High Fructose Com Syrup (HFCS) industry, might have on
the CARICOM sugar industry within the framework of the proposed Free Trade Area of
the Americas (FTAA). The second objective was to provide quantitative estimates for the
US HFCS demand configuration that can be used in subsequent modeling exercises.
In chapter 1 the problem to be addressed as well as the hypotheses and specific
objectives were elaborated. It was noted that the issues were complex and involved many
players and institutions operating within a changing global policy environment. Only a
subset of the problem was addressed in this study. The approach to the study was
outlined. It involved developing a conceptual model of the US sugar and HFCS markets
and assessing on the basis of this model the trade and welfare implications resulting from
likely changes to the US sugar program and an expanded HFCS industry within the
context of the proposed FTAA. Subsequently, the derived demand for HFCS in the US
was estimated and used to explain likely concerns of the CARICOM sugar producers.
The analysis began in Chapter 2 with the development of a detailed background
on some of the key players and institutions involved. The complex nature of the problem
121


46
made that the technical substitution of HFCS for liquid sugar was complete and that the
existing market share would continue into the future.
Evaluation and Critique
A common feature of the studies reviewed here is that all utilized time series data
in their analyses to estimate parameters in supply and demand equations. With the
exception of the study by Lopez and Sepulvelda (1987), none considered the issue of
time-varying parameters. Secondly, although some attempt was made to take into
consideration the substitution effect between HFCS and sugar in the US sweeteners
market, none of the studies directly included HFCS in their trade analysis. Thirdly, all
studies inferred that if the US were to liberalize its sugar program unilaterally, it would
have a positive (upward) impact on the existing world market price. Fourthly, all the
trade-related studies were either spatial or nonspatial equilibrium models. According to
Thompson (1981) although spatial equilibrium models represent one of the most popular
approaches to agricultural trade modeling, especially for comparative statics analysis of
the effect of a change in policy, their adequacy for this purpose is questionable. He noted
that such models were not consistent with how many international agricultural markets
normally operate and hence do not do well in accounting for actual trade flows. Chief
among the reasons given for the real world inconsistency of the models was their faulty
assumption of a homogeneous traded product, ignoring the fact that many importers
differentiate among exporters on historic or political grounds. Consequently, changing
from one source to another is usually much more difficult than assumed by such models.
The second class of models used was the nonspatial equilibrium types in which
simulation rather than optimization procedures are used to give the net trade position of


56
more complex ones involving state-space framework and Kalman filter (discussed
below)) are used for both the fixed operators and the varying parameters.
Time-varying parameters is regarded as a problem because standard regression
analysis of economic phenomenon assumes that the relationship being studied is stable
within the sample of estimation. For example, in the simple regression model of Y, = Xp
+ ut it is presumed that the p vector, which represents the effect of changes in the
explanatory variables upon the dependent variable, is constant. However, there are good
reasons why this might not be so and in fact should be allowed to vary. Ward and Myers
(1979) grouped the possible sources of parameter variation under three broad categories:
1) structural changes in the economic phenomenon being studied; 2)model
mispecification; and 3) aggregation. They noted that technology and institutional changes
always cause structural changes in economic phenomena. The Lucas (1976) critique of
economic analysis provides a good example. Lucas critique is to the effect that if the
policy regime is changed and economic agents takes account of the general policy
environment into their decision-making, then agents will adjust their behavior and the
coefficients that held in the previous regime will change to new values.
Sources of misspecification are varied but include omission of independent
variables, wrong functional forms and the use of proxy variables. For example, if the true
relation is:
Yt = ai + ct2Xt + 0C3X2t + ut (3.4)
and the analyst considers the linear relationship give by
Yt Pi + p2Xt + vt
(3.5)


99
possibility of a slight decrease. The actual analysis leading up to this result is as follows:
With the opening of the US sugar market under an FTAA the world market price of sugar
rises from the pre-liberalization price of Pwl to Pw2 ? the free trade market price (Fig.
4.12a). At this juncture, sugar now competes directly with HFCS with the potential to
regain some of the sweeteners market share. This is shown in Figure 4.12a as the
potential for the US consumption of sugar to increase from X2 to X3, which would imply
that both crystallized and liquid sugar are consumed. However, the increased demand for
HFCS in the non-US countries (ROW) and the realization of industry external economies
of scale cause the price of HFCS to fall (Fig. 4.11). This, as discussed earlier, results in a
downward shift in the US sugar demand curve from Dsi to Ds2, with the accompanying
shift in the excess sugar demand curve, EDi to ED2, in Figure 4.12b. At the same time,
the increase in demand for HFCS displaces some of the sugar demand in the non-US
countries (ROW) causing their demand function to slightly rotate counterclockwise while
shifting to the left, from RWDsi to RWDs2, as previously discussed.
The increased opportunity to export sugar to the US also causes some of the more
efficient sugar producers in the FTAA to increase their supply of sugar, resulting in the
ROW sugar supply curve shifting to the right, from RWSl to RWS2. The combination of
the leftward shift in the demand curve and the rightward shift in the supply curve in the
ROW result in a substantial rightward shift in the excess sugar supply curve, from ESi to
ES2 in Figure 4.12b. A new free trade price is established resulting from the intersection
of ED2 and ES2. As drawn in Figure 4.12b this new price Pw3 is the same as Pwi, the pre
liberalized price. While this might not be the exact outcome, the point to be made is that
the Agreement states that in order to be eligible for the benefits developing countries are obligated to
provide no less favorable treatment to the European Union than they provide to any developed country.


112
Table 5.1. Estimation of Price Equation
Description
Equation 5.4a-OLS
Estimates
Equati
Lik
cm 5.4b~Maximum
ihood Estimates
Parameters
Variables
Coef.
Std. Error
p- value
Coef.
Std. Error
p- value
4>o
Intercept
0.7395
1.4951
0.628
0.2797
0.9267
0.763
.
Price of Com
(LRPq)
0.2873
0.4150
0.499
0.6347
0.3310
0.055
§2
Price of
Electricity(LRPEt)
-2.0969
0.9381
0.041
-1.0939
0.8579
0.202
3
Interest Rate
(LRI,)
-0.4680
0.5829
0.435
-0.2425
0.4541
0.593
Price of Sugar
(LRPs.)
0.5301
0.4065
0.212
0.5772
0.2522
0.022
5
Price of Soft
Drink (LRP[>)
-1.2976
1.0847
0.250
-1.1964
0.8114
0.140
6
Price of Flour
(LRPfO
6.1545
2.3772
0.021
3.5046
1.8791
0.062
Adjusted R2
0.62
0.79
Durbin-Watson Statistic
1.09
1.73
rho
0.70
prices of the soft drinks (final goods) were significant at the 10% level. The signs on the
trend and HFCS price variable were consistent with expectations. In the case of the other
three variables no a priori assigning of the signs was possible given the nature of the
derived demand curve and the fact that the output and substitution effects do not
able 5.2 Final Estimates Using Kalman Filter Over the Period 1977-98
Parameter
Variable
Coefficient
Standard Error
p-value
Absolute Beta
Coefficient
Mi
Intercept
7.97624
4.47635
0.075
Pi.
Price of HFCS
(LRPh,*)
-1.63964
0.88958
0.065
2.26979
Pa
Price of Sugar
(LRPSt)
-0.43357
1.30173
0.739
0.87828
Pa
Price of Flour
(LRPf.)
-0.98789
1.82528
0.588
2.80602
P4
Price of Soft
Drink (LRPd.)
5.34171
2.77963
0.055
23.10574
view of the dynamic nature of the HFCS industry this is highly unlikely. Consequently, caution should be
exercise with interpreting the final estimates in Table 5.2 as representative of the derived demand curve.


129
The empirical results clearly showed that structural drift has been an important
component of the derived demand for HFCS over time. The intercept coefficient was
significant, and indicated an asymptotic growth pattern which supported the view of
some industry experts that the industry is now in a mature stage and has reached the limit
of its technological applicability to various uses as far as the US market is concerned. The
implications for the sugar interest and the CARICOM sugar producers suggest that
producers of the product, barring any significant break-through in the crystallized market,
now face a problem of actively seeking external markets for their product. This becomes
even more crucial with the additional gains to be had from the realization of external
I
economies within such an expanded market. All this could spark a iurther round of rent
seeking on the part of the com refiners as they actively seek to expand their market, with
fewer barriers to trade. It also implies a iurther weakening of the US sugar-corn coalition
lobby as the com processors appear no longer interested in the protection offered by the
sugar program at the expense of penetrating overseas markets.
For the sugar producers in the US, the findings represent somewhat of a comfort
to the extent that there might not be any further erosion of the market shares in the US
sweeteners within the confines of the current technology. In this regard, it allows one to
get an indication of the likely size of the sugar market in the US and by extension the size
of the sugar import market. However, the tabletop (crystallized sugar) market still
represents a substantial portion of the sweeteners market and could result in more
aggressive research on the part of the HFCS producers to perfect their effort to produce a
competitive crystallized HFCS product. With respect to the non-US sugar producers,
such as CARICOM, the main concern of the findings and its implications will be the


93
that the price ratio of HFCS to sugar remains constant. This assumption is made to
prevent the shifting of the kinked portion of the sugar demand curve as discussed earlier.
In this case it can be seen that such an outcome would not affect the US demand for
liquid sugar and hence the demand for HFCS. In other words, any changes in the US
sugar program which result in the new price still being at or above the original kink will
not affect the US demand for liquid sugar, as long as the price for HFCS remains below
the domestic price of sugar. This is so because the price decline, although it increases the
Price Price Price
Fig. 4.10a. US sugar market Fig. 4.10b. Trade diagram Fig. 4.10c. ROW
Fig. 4.10. Trade implications under partial liberalization of US sugar program
HFCS would be no more than 15% and would be reduced to zero over a 10-year period (American Sugar
Alliance, 1998).


145
Shim, J., and J. Siegel. Dictionary of Economics. New York :John Willey and Sons,
1995.
Sudaryanto, T. The Potential Impact of Liberalized Trade Policies in the United States
and in the European Economic Community on International Markets for Sugar.
Ph.D. Dissertation, North Carolina State University, 1987.
Tanyeri-Abur, A. An Agricultural Sector Analysis of the United States Sugar Import
Policy. Ph.D. Dissertation, Texas A&M University, 1990.
Thomas, C. Y. Sugar : Threat or Challenge? Ottawa: International Development
Research Center, 1985.
Thompson, R. A Survey of Recent US Development in International Agricultural Trade
Models. Bibliographies and Literature of Agriculture, No. 21, Economic Research
Service, USDA, 1981.
US Department of Agriculture (USDA). Economic Research Service, Commodity
Economic Division, Sugar and Sweeteners Situation and Outlook Report.
Washington, DC, Various Issues.
US Department of Agriculture (USDA), Economic Research Service, Commodity
Economic Division, Sugar and Sweeteners Situation and Outlook Yearbook.
Washington, DC, Various Issues.
US Department of Agriculture. Economic Research Service (USDA), Commodity
Economic Division, Free Trade in the Americas. Situation and Outlook Series,
WRS-98-1, Washington, DC, November 1998.
Ward R. and L. Myers. Advertising Effectiveness and Coefficient Variation Over
Time. Agricultural Economics Research, 31(1979): 1-11.
Ward R. and D. Tilley. Time-varying Parameters With Random Components: The
Orange Juice Industry. Southern Journal of Agricultural Economics 12(1980):
5-13.
West Indian Commission. Time for Action: Report of the West Indian Commission
(Second Edition). Kingston : University of West Indies Press, 1993.
Wint, A. South-South Integration. Paper presented a Seminar on New Global Trading
Arrangement, organized by the UWI and Sponsored by the European Union.
Kingston, Jamaica November 19, 1997.
Yule, G. U. Why Do W e Sometimes Get Nonsense Correlation Between Time Series?
Journal of the Royal Statistical Society, 89(1926): 1-64


28
alternative than selling sugar on the world market. It has also permitted the subregional
sugar industry to survive and, through inter-industry economic linkages, has fostered
growth in services and other sectors. Table 2.8 gives a rough indication of the real quota
rent earned by CARICOM as a result of the US program over the period 1990-98. The
data are based on the assumption that CARICOM fulfilled its quota commitment to the
US, which is approximately 4.8 percent of the total tariff rate quota (TRQ)11. Also, no
Table 2.8. CARI COM/US Market Real Earnings and Quota Rent (1992=100)
Year
TOTAL
TRQ
(MMT)
SAC
TRQ
(MMT)
Real US
Price raw
(US$/MT)
Real World
Price raw
(US$/MT)
Total Real
Earnings
(US $M)
Quota Rent
(US $M)
90-91
2.1027
0.1051
487.41
204.27
51.24
29.77
91-92
1.3863
0.0693
468.82
199.98
32.50
18.63
92-93
1.1347
0.0567
464.86
215.66
26.37
14.14
93-94
1.1347
0.0567
458.00
252.06
25.98
11.68
94-95
1.3241
0.0662
466.49
273.07
30.88
12.81
95-96
1.9702
0.0985
439.92
240.39
43.34
19.66
96-97
1.9065
0.0953
414.98
227.90
39.56
17.83
97-98
1.4545
0.0727
403.56
177.08
29.35
16.47
Source: Compiled from USDA data.
adjustments have been made to accommodate shipping expenses. As can be seen, the
amounts of quota rents in real terms are substantial and their loss would have serious
negative social repercussions. However, as indicated in Fig. 2.1, prices have trended
upward in the world market and downward in the US market during the period. An
obvious implication of this observation is that the real premium enjoyed by quota holders
and US producers has been declining. As will be seen later, this factorcoupled with a
10 The Euro is currently valued at approximately 1 Euro = US $1.06.
11 The Tariff Rate Quota (TRQ) system is a key component of the current US sugar program. This
component of the program is elaborated on below.


41
the introduction of HFCS. They attributed the decreased in own price elasticity of sugar
to the substitution taking place in the more price elastic segment of the sweetener market
such as the soft drinks and canning industries. They concluded that as the US demand for
sugar decreases and the food industry adjusts faster to sweetener choices, the US
government would have to impose more restrictive import barriers to maintain prices to
domestic sugar and HFCS producers.
Leu, Schmitz, and Knutson (1987) used a general equilibrium model to analyze
the policy options for the US Sugar program and to empirically estimate the
substitutability between sugar and HFCS by including the price of HFCS as a
determinant of sugar demand. They used a composite price of com sweeteners glucose,
dextrose and HFCSin order to generate a series dating back to 1955. Demand
elasticities were estimated for sugar and HFCS while supply elasticities were obtained
from previous studies. Their results indicated that the net social cost of the quota program
with an elasticity of excess supply of 2.37 was $203 million when substitution with
HFCS was allowed and $253 million with no HFCS substitution. Under the assumption
of a perfectly elastic excess supply curve the net social cost with HFCS substitution was
about $1 billion. They concluded that a deficiency payment program would be more
costly than an import quota. However, Marks (1993) stated that the study suffered since
price data on the HFCS were only available after 1975. He also questioned the composite
commodity assumption because glucose and dextrose are much less substitutable for
sugar than is HFCS.
Sudaryanto (1987) developed a trade model to analyze the effects of trade
liberalization by the US and EEC on international markets for sugar. In relation to the


8
studies of production and consumption of the commodity have been largely confined to
the domestic market. With the advancements in HFCS technology on both the supply and
demand sides and the reduction in trading barriers in the international arena, the prospects
of this commodity assuming a greater level of importance in trade have increased
considerably.
One aspect of the HFCS market is the dual nature of the product, whereby within
a given range of the sweetener demand curve it substitutes almost perfectly for sugar but
within another range it does not. Either of the sweeteners (HFCS or sugar) can be used in
soft drinks, but for some other uses that require a crystallized sweetener, HFCS is not a
viable substitute. This dual nature of the substitutability relationship has implications for
the extent to which the commodity will be traded internationally, the shape of the
sweetener demand curve and the econometric approach to be used to estimate HFCS
demand parameters.
Another related aspect is the issue of the extent to which the industry has reached
a level of maturity. Thomas (1985), pointed out that HFCS when considered by industrial
standard is regarded as a relatively new product. Among other things, he noted that this
implies that there will almost certainly be further technological improvements and that
the behavior of the industry can be expected to be different from that of a mature industry
such as sugar. If this is indeed the case, then there are implications with regard to the
econometric approach used to estimate the important economic parameters. A cursory
investigation into the history of the industry seems to support the point made by Thomas.
On both the supply and demand sides of the market advances in technology have resulted
in noticeable development of the product and product usage. For example, the price of


81
substitutes has market intelligence and knows the reaction function of his competitor but
the other is naive. Consequently, the producer with the market intelligence take
advantage of his competitor by supplying an amount at a price which will maximize his
profit at the expense of his competitor. Here, the HFCS producers could be regarded as
the sophisticated suppliers, while those supplying sugar are regarded as naive. Taking this
analogy a step further, as in the case of the Stackelberg model in which it is possible for
one competitor to force another out of the market by sufficiently differentiating the
product, in a similar manner such an opportunity exist for HFCS to force liquid sugar out
of the market.
More specifically, it is assumed that the profit maximizing objective fimction of
the HFCS suppliers can be represented as follows:
= (Ph-Ch)*Dh(.) (4.8)
where n represents the maximum industry profit; Ph is the price of HFCS; Ch is the cost
of producing HFCS and DH is the demand for HFCS as defined in equation 4.7.
Assuming other factors remain constant, the reaction fimction of the HFCS suppliers is
obtained by differentiating the profit function with respect to the price of HFCS, setting it
equal to zero and solving for the price of HFCS. The reaction function will therefore
reflect the best price to charge for HFCS given the price of sugar and the other exogenous
factors that affect profit of the HFCS industry.
Within the framework of the above conceptualization, it is assumed that: 1) the
demanders of HFCS are price takers; and 2) that the marginal cost of producing the
HFCS remains constant over the entire range of output. The first assumption seems
plausible since in comparison to the number of firms producing HFCS there are large


72
The US Sugar and HFCS Subsectors
Figure 4.1, presents a flowchart of the US sugar and HFCS sweeteners
subsectors. As stated earlier, HFCS is available mainly in a liquid form and is an almost
perfect substitute for liquid sugar, hence it is used in its entirety in the industrial sector.
Fig. 4.1. Flowchart of the US FIFCS and Sugar Subsectors
As evident from the Figure, useful byproducts are produced in the production of both
sweeteners. However, the revenue generated from the sales of the byproducts as a
proportion of the cost of the main input sugarcane, sugarbeet and com, respectivelyare
more substantial in the case of the production of HFCS.


CHAPTER 5
EMPIRICAL MODEL OF THE US DEMAND FOR HFCS
The preceding chapter outlines the conceptual model of the US sugar and HFCS
subsectors. This chapter focuses on the estimation of the derived demand for HFCS. The
first part deals with the specification of the model and gives a description of the variables
and data used. In the latter sections the results of the demand estimation and the model
implications are discussed.
Empirical Model
In keeping with earlier discussion of a derived demand function (Chapter 3) it is
postulated that the demand for HFCS is a function of the price of HFCS, the prices of
substitutes, and the prices of other inputs used in the production of the final HFCS-
containing goods. The total amounts of HFCS-42 and HFCS-55 utilized (millions of short
tons dry weight equivalent) were used to represent the quantity of HFCS demanded. This
amount reflects the total US domestic disappearancetotal supply less exports and stock
of HFCS. The HFCS-42 price (cents per pound dry weigh) was used as a proxy for the
price of HFCS. Using this series was necessary in light of the incompleteness in HFCS-
55 price series. This should not pose a serious problem, since both syrup prices tend to be
highly correlated.
The US wholesale beet sugar price series (cents per pound) was chosen to
represent the price of substitutes. Wholesale refined sugar prices rather than the
wholesale raw sugar prices were used, since the former is more readily comparable to the
103


CHAPTER 2
BACKGROUND
The purpose of this chapter is to provide background information on key players
and institutional arrangements involved with sugar and HFCS. The chapter contains a
brief description of the proposal to form a Free Trade Area of the Americas and
highlights some of the disparities among the potential members. This is followed by a
review of world and Western Hemispheric trends in the production and marketing of
sugar. Next is a discussion of the importance of the sugar industry to CARICOM. Key
marketing arrangements such as the EU/ACP Sugar Protocol and US Sugar Program are
examined. Implications of the GATT/WTO rules are also briefly examined as they relate
to these special marketing arrangements. The final section of the chapter focuses on the
HFCS and reviews world and Western Hemisphere production and marketing trends for
this commodity.
Free Trade Area of the Americas (TTAA)
A Free Trade Area of the Americas (FTAA) is one of several trade agreements
that the United States is encouraging and actively pursuing. The decision to establish an
FTAA and to conclude the negotiation for its implementation by the year 2005 was made
at the Summit of the Americas held in Miami, Florida in December 1994. Among other
things, the Plan of Action calls for the promotion of prosperity among the 34
participating countries through economic integration and free trade, the eradication of
poverty and discrimination, and the guaranteeing of sustainable development by the
14


76
given favorable price changes, and the technical characteristics of the individual
sweeteners.
Horizontally summing the two demand curves, CS and LS, in Figures 4.2a and
4.2b, gives us Figure 4.3, which represents the total market demand for refined sugar in
the US. Of significance in our model is the kinked nature of the resulting demand curve
ake. Point k on the curve, which represents the kink, can be interpreted as follows: At
prices above pK there is no demand for liquid sugari.e. it is completely substituted for
by HFCS. Below the kink however, there is both a demand for crystallized sugar (CS)
and liquid Sugar (LS), the latter in competition with HFCS.
Price
Fig. 4.3. Hypothetical US aggregate demand for sugar
In Figure 4.4, the US domestic sugar supply curve (Ss) is introduced and the
established US domestic support (loan) price, PT, set at approximately US 23 cents per
pound refined sugar. As drawn, the Figure indicates a situation in which there still


86
Price
Fig. 4.8. Hypothetical HFCS industry average costs, reflecting external economies
The Interrelationship between the US Sugar and HFCS Markets
As indicated earlier there is a unique relationship between the US sugar and
HFCS markets. Sugar can substitute in all cases for HFCS but HFCS can only replace
sugar in a selected number of usesalbeit in such cases it has proven to be quite an
effective substitute (Barros, 1992). The implication is that it is only one segment of the
demand curve for sugar that is affected by changes in the demand and supply conditions
within the HFCS subsector. As indicated earlier in relation to Figure 4.3, the kink in the
sugar demand curve is of importance since it indicates the price above which there would
be no demand for liquid sugar. Several factors affect the positioning of the kink. This
point is not stationary but shifts in relation to changes to these factors. The primary
factors determining the positioning of the kink are: 1) the availability of other substitutes;
2) the price of HFCS which itself is influenced by several factors including the price of


APPENDIX
STATE AND SMOOTEHED ESTIMATES 138
REFERENCES 140
BIOGRAPHICAL SKETCH 147
Vll


64
time invariant the Kalman filter will usually converge to a steady state, in that the
covariance matrix becomes time invariant and the estimates of the parameters are the
same as those obtained from fixed coefficient regression (Harvey, 1987).
Advantages of State-Space Approach and the Kalman Filter
Several authors (Harvey 1987,1997; Engle and Watson 1987; and Aoki 1990;
Hall et al, 1992) have alluded to the advantages of the state-space model over the
traditional parametrization of time series models such as ARMA, ARIMA, VARs and
VECMs. They noted that from a strictly theoretical point of view both types of models
are equivalent, because generic models in one representation can be transformed into the
generic ones in the other. However, when judged on other grounds such as numerical
stability, sensitivity with respect to small specification errors, statistical properties of
parameter estimators, or simply ease of dealing with nonstationary series, the state-space
models are superior. Moreover, they point out that in order to avail oneself of the
theoretical results and computational algorithms that have been developed over the years
in system literature, time series must be put in this newer state-space representation.
Hall et al (1992) in motivating the discussion for the use of the state-space
approach and the Kalman filter notes that although it is widely used in certain branches of
engineering and by applied statisticians, it is only just emerging as a possible useful tool
of the applied economists. He notes that the Kalman filter can be interpreted in terms of
agents forming expectations. He points out that while the paradigm for modeling
expectation is the rational expectation hypothesis (REH) where agents act as if they know
the true model of the economy up to a set of white noise errors, the consideration of the
information availability assumption has led critics to label it unrealistic. He notes that the


95
policy having no effect on the US demand for liquid sugar. Consumption of CS increased
marginally from X2 to X4. If instead the new price is below the kink in Ds2, then it could
imply a sharing of the market between HFCS and liquid sugar.
To complete the analysis we invoke the large country trade assumption and
construct the usual trade-diagram and rest of the world (ROW) exporters, Figures 4.10b
and 4.10c, respectively. Assuming that the increased demand for HFCS, resulting from
the opening of the industry to trade within the FTAA, works mainly through the demand
for sugar in the rest of the world (Figure 4.10c), this development can be modeled as a
combination of two effects. Firstly, the presence of a substitute for liquid sugar in the
market causes the demand for sugar in the ROW to become more elastic. This is modeled
as a slight rotation of the demand curve. Secondly, as the price of HFCS declines and
Price
Fig. 4.11. Hypothetical HFCS industry long-run demand and average cost
reflecting external economies non-US users make the switch, the demand for HFCS gains
at the expense of sugar. This results in a leftward shift in the rest of the world demand for
sugar, from RWDsi to RWDs2. In other words, the increased consumption of HFCS in the


I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
)avid A. Denslow
Distinguished Professor of Economics
This dissertation was submitted to the Graduate Faculty of the College of
Agriculture and Life Sciences and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctqrf>f Philosophy.
December, 1999
Den, College of Agriculture and Life
Sciences
Dean, Graduate School


27
(non-ACP) countries, presently granted Most Favored Nation (MFN)9 status, will be
enhanced thereby reducing the preference margin for exports from ACP. Secondly, as
support is reduced, particularly subsidization of exports to domestic producers in the EU,
the domestic prices paid to EU producers will begin to fall. This will in turn, trigger a
reduction in ACP prices since the Sugar Protocol guarantees that ACP quota holders will
be paid the same prices as EU producers.
Other factors which will or have begun to erode the value of the EU/ACP Sugar
Protocol include: 1) the pre-EU membership arrangements with the countries of Central
and Eastern Europe some of which are sugar producers; 2) a surge in EU centered
regional trade arrangements, in particular, preferential agreements in the pipeline with
some of the sugar producing countries in Latin America; 3) the fact that markets have
evolved significantly since the first Lome Convention in 1975, with the result that Europe
has become a leading sugar exporter; and 4) that cane sugar can be bought on the world
market at much lower prices than those guaranteed by the Protocol which now cost the
EU several hundred million Euros10 each year. A final issue that is unresolved is the
impact of the Euro, the EUs common currency, on the sugar prices that quota holders
receive.
The US Sugar Program and Policy
In spite of the less lucrative nature of the US market relative to the EU market, the
former nonetheless has provided CARICOM sugar producers with a much more viable
9 The Most Favored Nation (MFN) clause of a commercial treaty is a binding contract by the signatories to
confer upon each other all the most favored trade concessions that either may grant to any other nation
subsequent to the signing of the agreement.


116
since around 1996, there appears to have been a slow down and a leveling off in the
growth pattern of the demand for the product. This finding is important from two
perspectives. First, it corroborates the suggestion made by some industry experts that
within the framework of the current technology, the extent of the substitution between
HFCS and liquid sugar has been completed. Secondly, the finding has wider implications
with regard to the future of the HFCS industry and its expansion within a wider trading
environment occasioned by the proposed formation of the FTAA. It also support the
earlier conjecture with regard to possible rent seeking behavior on the part of the
suppliers within a much more liberalized trading environment. This latter issue is
discussed in greater detail in the next chapter.
Own-Price Elasticity (Bid
Figure 5.2 shows the time path adjustments for the own price elasticity of HFCS.
In particular, the evidence suggests that the elasticity of the demand has been becoming
less elastic at a decreasing rate. In other words, the demand for the product although still
elastic, has become relatively less elastic changing from -2.2 to -1.65. Again, this seems
plausible given the developments that have taken place on the supply side in
differentiating the product from liquid sugar and making it more user-friendly. Recalling
the sugar and HFCS competition within the framework of the Stackelberg model
(Chapter 4), the observed pattern would be consistent with the theory which suggests that
by differentiating a product to such an extent where it is perceived as completely different
from its competition, it is possible to drive the competitor out of the market. This has
been the case to some extent with the HFCS-liquid sugar competition. Judging from
statements made by some of the industry experts, HFCS is now considered to be a


70
for HFCS can be obtained. However, in light of the difficulties surrounding the
estimation of a reliable profit function, such as obtaining cost sensitive information and
quantities used, a more straight forward approach can be taken in which the derived
demand is estimated directly on the basis of the explanatory variables suggested by the
above theory. These explanatory variables include the price(s) of the final
commodity(ies) and the respective prices of the various inputs used in the production
process. This latter approach will be used in this current study.


100
owing to the presence of HFCS in the market, and the fact there is a strong possibility
that it can compete with sugar at world market prices, there is no guarantee that the
elimination of the US sugar program will necessarily result in an appreciable rise in the
world market price of sugar The exact outcome will be determined, among other things,
by the extent to which the importing countries open up their market and the degree of the
penetration achieved and sustained by HFCS. Additionally, it will depend on the long-run
ROW export supply price elasticity. Marks (1993) noted that in general the lower the
ratio of net exports to production, the higher should be the elasticity of export supply.
Price Price Price
Fig. 4.12a. US Sugar Market Fig. 4.12b. Trade Diagram Fig. 4.12c. ROW Sugar Market
Fig. 4.12. Trade implications under completely liberalized US sugar program


96
rest of the world will be at the expense of sugar. This leftward shift in the ROW demand
curve is explained by assuming that as the income of the members of the FTAA
improves, as a result of integration, there is for example a decrease in home baking and
an increase in eating outside of the home. Since less food is prepared at home, and since
it is more in the away-ffom-home food that there is a demand for HFCS, an indirect
substitution of HFCS for sugar occurs. Assuming iurther that the ROW sugar supply
remains the samegiven the limited incentive in the US market resulting from the change
in policythe net effect would be a rightward shift of the excess sugar supply function in
Figure 4.10b, from ES, to ES2, implying a fall in the world market price for sugar, from
Pwl to Pw2. The implication is that the partial liberalizing of the US sugar program within
the context of an FTAA, could in fact depress the world market price of sugar. This
conclusion differs from previous studies of the likely effects of a partial or full
liberalization of the US sugar policy which suggested that such action would increase the
world sugar price (Atlas, 1996; Marks, 1993; Schmitz and Christian, 1993).
Welfare implications
The welfare implications of the effect of a change in the US sugar policy within
scenario I, can be deduced from Figures 4.10 and 4.11. Notably, the producers of sugar in
the US would lose, while the producers of HFCS would gain substantially. The producers
of sugar in the US would lose as a result of lost revenue or income owing to eut back in
output and a reduction in the domestic price of sugar. The producers of HFCS however,
would gain from the lower cost of production and the increased market access within the
framework of the FTAA. Furthermore, assuming that such industries manage to maintain


80
the other inputs used in producing the final good, especially the price of sugar; 2) an
increase in the number of products using HFCS; 3) seasonality ; 4) changes in the price of
the final commodity; and 5) an improvement in technology which improves the
substitutability of HFCS for sugar. Consequently, the derived demand for HFCS can be
represented by the following general specification:
Demand for HFCS: Dh = /{Ph, Ps, Pf, T, Z) (4.7)
where:
Dh
=
demand for HFCS
Ph
=
price of HFCS
Ps
=
domestic price of sugar in the US
Pf
=
price index of sugar and HFCS containing products
T
=
technology (trend variable)
Z
a vector of other factors
With respect to the supply, it is important to note that the consensus of industry
experts is that the HFCS industry operates in an oligopolistic framework (Polopolus and
Alvarez, 1991, Thomas, 1985). A consequence of this is that even if the exact profit
function of such an industry were available it would not be possible to obtain the input
supply function of the industry by way of the duality theory, since the underlying
assumption of the theory is the existence of a perfectly competitive framework.
Moreover, in light of the prevailing circumstances under which the FfFCS is produced,
notably a kind of captive market resulting from the restrictive policies of the US sugar
program and the implicit floor price of sugar, it is argued that there is no supply function
for the industry but rather a reaction function. This reaction function can be viewed in the
contextalthough not identicalof the Stackelberg model in which one producer of the


104
price of HFCS on a sugar equivalency basis. While other com sweeteners such as
dextrose and glucose can be substituted for HFCS, in general owing to their much higher
price per unit of sweeteners there is limited substitution amongst these products. For
reasons discussed earlier, the non caloric (artificial) sweeteners were excluded from the
analysis. Consequently, only the price of sugar was included in the equation to capture
the effects of substitutes.
With respect to the final HFCS-containing products, only the price of soft drinks
was used. Although there are several such final products spanning several industries, the
models were limited to only the price of soft drinks, since this product utilizes the bulk of
the HFCS produced. In addition, sufficient information was not available to compute a
weighted index across the other products. Given the high degree of pair-wise correlation
among these indices, inclusion of more than one of these indices in the equation could
give rise to multicollinearity problems. While the soft drink index is not completely
representative of the range of final products, it was used as a proxy for those
unrepresented final products.
The final variable included in the derived demand equation is the price of flour.
Flour was used since a fair amount of HFCS is utilized in the baking industry. The
inclusion of this index represents a proxy for the other inputs used in the production of
final HFCS-sugar containing products. Finally, all prices in the equation are expressed in
real terms using the GDP deflator. The final model is expressed with a double log
specification representing the derived demand equation:
LQh, = Po + P.LRPh, + p2LRPSt + p3LRPDt + p4LRPFt + e, (5.1)


23
Table 2.5. Leading Hemispheric Sugar Exporters and Importers, 1997-98
Exporters
Importers
Country
Share (%)
of Hemispheric
Exports3
Exports as a %
of Domestic
Production
Country
Share (%)
of Hemispheric
Imports
Imports as a %
of Domestic
Consumption
Brazil
56.4
45.9
US
44.1
22.1
Guatemala
10.7
76.0
Canada
27.3
95.7
Mexico
10.0
23.3
Peru
5.8
28.2
Columbia
6.5
38.5
Venezuela
5.4
30.8
SACb
5.1
86.5
Chile
4.9
32.4
Dominican Rep.
2.3
51.2
Ecuador
2.9
32.5
El Salvador
1.6
45.4
Uruguay
2.1
86.4
a] Include re-exports
b] Total exports from six CARICOM countries
Source: International Sugar Organization
In terms of regional imports, the United States and Canada dominated the trade in
97/98. Together these two countries accounted for in excess of 70% of the region's sugar
imports. Peru and Venezuela were the other major importers, accounting for about
11.2% of the import market. The fourteen CARICOM countries import a total of about
0.1 million metric ton. This amount represents about 2.2% of the overall regional imports
and less than a third of the total CARICOM sugar requirements. The bulk of such imports
was from the world market and in some cases was done in order that the exporting
country would be able to satisfy its sugar export quota commitments.
Importance of the Sugar Industry to CARICOM Sugar Producers
Socioeconomic Importance
The sugar industry is of considerable importance to CARICOM countries. This is
particularly true for the six members that comprise SAC. Among other things, the
industry contributes substantially to valuable foreign exchange earned by these countries.
For example, in St. Kitts and Nevis the industry provides more than one-half of that


92
analysis which is predicated on the foregoing conceptual model of the US sweetener
market, considers two scenarios: 1) a partial liberalization; and 2) a complete
liberalization of the US sugar program.
Scenario I- Partial Liberalization of the US Sugar Program
Trade implications
In view of the sensitive nature of sugar and the still relatively strong sugar lobby
within the US, one possible scenario is a partial liberalization of the US, one possible
scenario sugar program. It is conceivable that within an FTAA negotiation framework the
US could offer initially, to reduce its current level of support to domestic sugar producers
with an agreement that other member countries of the FTAA open up their markets to
trade in HFCS For example, the US could propose lowering the domestic support price
to say 15 cents per pounda price most experts feel would be the true world free trade
price. The effect of this decision is reflected in Figure 4.10a in terms of our conceptual
model as a fall in the support price from Pti. Here, it is assumed that the initial situation
which existed in the US market was such that at the current support price there were no
demand for liquid sugar i.e. substitution had been completed. This is reflected by the fact
that the kink is placed below the support price Pti. With the lowering of the support price,
two cases are possible. The first case is where the new support price Pt2 is at or above the
kink in the US sugar demand curve (Dsi). For convenience it will be assumed initially
18 A similar agreement was negotiated within the framework of the NAFTA Agreement between the US
and Mexico whereby provisions were made for the phased opening up of the US sugar market to excess
supply from Mexico over a 15-year period, while agreeing that Mexicos import tariff on US exported


26
virtue of its non-reciprocal nature (Blanco, 1997). In particular, the Arrangement was
ruled to be discriminatory vis--vis other LDCs and thus not covered by the enabling
clause permitting GSP schemes. This GATT ruling led to the EU seeking a waiver,
which was subsequently granted by the WTO. However, the non-reciprocity permitted by
Table 2.7. CARICOM/EU Market Real Earnings and Quota Rent (1992=100)
Year
SAC
QUOTA
(MMT)
UK Price
(Pounds
/MT)
Real UK
Price raw
(USS/MT)
Real World
Price raw
(US$/MT)
Total Real
Earnings
(US $M)
Quota Rent
(US $M)
90-91
0.4281
346.10
628.35
204.27
269.00
181.55
91-92
0.4281
351.12
587.79
199.98
251.64
166.03
92-93
0.4281
403.00
629.91
215.66
269.67
177.35
93-94
0.4281
400.28
587.39
252.06
251.47
143.56
94-95
0.4281
420.56
608.52
273.07
260.51
143.61
95-96
0.4281
445.58
649.58
240.39
278.09
175.18
96-97
0.4281
414.44
577.45
227.90
247.21
149.64
97-98
0.4281
364.64
512.96
177.08
219.60
143.79
Source: SAC data
the GATT/WTO waiver is contrary to Lome principles and is only valid until the end of
the present Convention in the year 2000. What this means in practical terms for the ACP
countries and CARICOM in particular, is that the Sugar and Banana protocols will have
to stand alone as either viable WTO-compatible commercial arrangements between the
EU and the nations to which they apply, or they will disappear completely. The WTO
1999 negotiations will undoubtedly include strong calls for free trade in sugar and the
demise of the Protocol.
A second factor which stands to reduce the potency of the Sugar Protocol,
irrespective of the shape or form of any future EU-ACP Agreement, is the reductions in
domestic support measures by the EU as it seeks to implement its WTO liberalization
commitments. Firstly, as the EU implements its agreed tariff reductions, access by third


DYNAMICS OF THE US SUGAR AND HFCS MARKETS:
IMPLICATIONS FOR THE CARICOM SUGAR PRODUCERS WITHIN THE
FRAMEWORK OF THE PROPOSED FTAA
By
EDWARD ANTHONY EVANS
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1999


30
channels were cleared rather than forfeited to the Commodity Credit Corporation
(CCC). The MSP was set at a level of $0.22 per pound of raw sugar. However, under
the FAIR program the no net cost provision was eliminated and the MSP is no longer
set explicitly although it is expected that supply would be managed in such a way as
to ensure the market price would be above the loan rate.
3. Tariff-rate quota (TRQ)-the TRQ for sugar and sugar-containing products is not
technically a part of the domestic sugar support legislation (Buzzanell, 1998).
Nonetheless, it is one of the principal tools used to support the loan program and to
ensure that the domestic price does not come within the forfeiture range. The US
Trade Representative (USTR) allocates the TRQ for raw sugar to 40 countries. The
total quota varies from year to year in keeping with the objective to stabilize domestic
raw-sugar prices. Although the quota can vary from year to year, it cannot fall below
1.14 MMT, which is the commitment given by the United States under the Uruguay
Round of the GATT (USD A, Situation and Outlook Series, 1998). Of the 40 countries
participating in the program, 23 are from the Western Hemisphere and include all the
SAC sugar exporters. Country allocations are based on historic trade with the United
States. Haiti, and St. Kitts and Nevis were included in the allocations as minimum
quota-holding countries, and the allocation to Mexico was increased to fulfill
obligation pursuant to North American Free Trade Agreement (NAFTA). About 66
percent of the raw cane sugar TRQ are allocated to the hemispheric countries. The
Dominican Republic, Brazil, CARICOM, and Argentina account for the bulk of the
overall regional quota with TRQ shares of 17 percent, 14 percent, 5 percent, and 4.6
percent, respectively. All countries in the Hemisphere, with the exception of Brazil,


36
Table 2.10. Changes in US Per Capita Sweetener Consumption and Relative Market
Shares, (1975 1997) Pounds, Dry Weight Basis
Year
Refined
Sugar
Relative
Share (%)
HFCS
Relative
Share (%)
Total Corn
Sweet'rs
Relative
Share (%)
Honey & Edible
Syrups
Relative
Share (%)
Total Caloric
Sweet'rs
1975
89.2
76
5.0
4
27.5
23
1.4
1
118.1
1976
93.4
75
7.2
6
29.7
24
1.3
1
124.4
1977
94.2
74
9.5
7
31.2
25
1.4
1
126.8
1978
91.4
72
12.1
10
33.7
27
1.5
1
126.6
1979
89.3
70
14.9
12
36.4
29
1.4
1
127.1
1980
83.6
67
19.1
15
40.2
32
1.2
1
125.0
1981
79.4
63
23.2
19
44.5
36
1.2
1
125.1
1982
73.7
60
26.7
22
48.2
39
1.3
1
123.2
1983
71.1
57
30.7
25
52.2
42
1.3
1
124.6
1984
67.6
53
36.3
29
57.8
46
1.4
1
126.8
1985
63.2
49
44.6
35
63.9
50
1.5
1
128.6
1986
60.8
48
45.1
36
64.6
51
1.6
1
127.0
1987
63.1
48
47.1
36
66.8
51
1.7
1
131.6
1988
62.6
47
48.3
36
68.3
52
1.5
1
132.4
1989
62.8
47
47.5
36
68.0
51
1.6
1
132.4
1990
64.8
47
49.2
36
70.4
51
1.6
1
136.8
1991
64.4
47
50.0
36
72.0
52
1.6
1
138.0
1992
64.4
46
51.6
37
74.3
53
1.6
1
140.3
1993
64.6
45
54.4
38
77.8
54
1.6
1
144.0
1994
65.8
45
56.4
38
80.2
54
1.5
1
147.5
1995
66.2
44
58.4
39
82.6
55
1.5
1
150.3
1996
66.9
44
59.8
39
84.2
55
1.5
1
152.6
1997
67.1
43
61.4
40
86.3
56
1.5
1
154.9
Source:USDA, ERS, Commodity Economic Division, Sugar and Sweetener: Situation and Outlook Report,
various issues
Fig. 2.2. Changes in US per capita consumption of selected sweeteners, 1975-97


LIST OF REFERENCES
American Sugar Alliance. Initiation of Section 302 Investigation on Mexican Practices
Affecting HFCS, (Office of the United States Trade Representative) June 19,
1998.
Anderson, B and J Moore Optimal Filtering. Englewood Cliffs, NJ: Prentice Hall, 1979
Aoki, M. State Space Modeling of Time Series. 2nd edn. New York: Springer-Ver lag
1990.
Atlas, M. The Impact of Trade Liberalization by Major Importing and Exporting
Countries on World Sugar Trade. Ph.D. Dissertation, University of Colorado,
1996.
Balassa, B. The Theory of Economic Integration. London: George Allen & Unwin Ltd,
1969
Barros, A. R. Sugar Prices and HFCS Consumption in the United States. Journal of
Agricultural Economics, 43 (1992):64 -73.
Bernal, R. The Integration of Small Economies in the Free Trade Area of the Americas.
Paper prepared under the aegis of the Center for Strategic and International
Studies (CSIS), Washington: Volume IX Study 1, February 1998.
Bemirschka, M., W. Koo, and J. Lou. World Sugar Policy Simulation Model: Description
and Computer Program Documentation. Fargo department of Agricultural
Economics, Agricultural Experiment Station, North Dakota State University,
1996.
Blanco, M. ACP-EU Trade: Into the New Millennium. The Courier. 166(November-
December 1997): 64-67.
Borrell, B. and R. Duncan. A Survey of World Policies, in The Economics and Politics
of World Sugar Policies S. V. Marks and K. E. Maskus eds., p. 15-48 Ann
Arbor: The University of Michigan Press, 1993.
Bryan, A. Coping with the New Dynamics in The Caribbean: New Dynamics in Trade
and Political Economy, A. T. Bryan, ed., North-South Center, 1995.
Buzzanell, P. U.S. Sugar Trade Relations with Latin America. Paper presented at the
Third World Sugar and Sweeteners Conference, Sao Paulo, Brazil, October 1998.
140


69
7*'/v = -x](P,v,w,Z)<0 =>
derived demand for input one
Sk/p = v, w, z) > o =>
supply function
The main difference between the derived demand obtained by way of the profit function
and that obtained from the cost function is that in the former output is allowed to vary
whereas in the latter it is held constant.
Once the derived demand is obtained, the own price elasticity of demand for the
input as well as the cross price elasticity of demand for an input can be computed. In the
case of the former, the elasticity will always be negative once the input is not considered
to be inferior. This is because in production theory, unlike the consumption theory, the
substitution and output effects always work in unison. In the case of the cross price
elasticity the sign is ambiguous. The ambiguity is due largely to the fact that in this case
the substitution and the output effects may not always move in the same direction. The
magnitude of both the own-price and cross-price elasticities is less predictable and
depends on factors such as the elasticity of substitution, the importance of the input in
terms of its share of total cost of production of the final good, and the price elasticity of
demand for the good being produced. According to Nicholson (1995) in general, the
own- or cross-price elasticity of demand for any input will be greater (in absolute value):
1) the larger the elasticity of substitution of that input for other inputs; 2) the larger is the
share of total cost represented by expenditures on that input; and 3) the larger is the price
elasticity of demand for the good being produced.
Thus, it is possible that given the profit function of the industries utilizing HFCS
as an input, and the assumption that such industries are operating within a perfect
competitive framework with a profit maximization objective, the derived demand curve


52
spurious) and there is no loss of any valuable long-term information, which would result
if their differences were used instead.
Moreover, in such situations, it is possible to tie in the short-term behavior of a
variable obtained from differencing, with a long-term equilibrium relationship. This is
done with a special formulation which involves a mix of both level and difference terms.
Such models are referred to as error correction models (ECMs). The ECMs can therefore
be viewed as comprising the short-run transitory effects and the long-run relationship and
describes how the long-run solution is achieved by a combination of negative feedback
and error correction (Darnell, 1994). In this regard, it is seen as a way to model the short-
run dynamics without losing sight of the longer-term steady state as may be posited by
economic theory. In other word, the ECM specification provides a means by which the
short-run observed behavior of variables can be associated with their long-run
equilibrium growth paths. Kennedy(1998) notes that this is one of the reasons why
economists have shown such interest in the concept of cointegration, since it provides a
formal framework for testing for and estimating long-run (equilibrium) relationships
among economic variables. He suggests that if a co integration relationship can be found,
advantage should be taken of the ECM framework.
It should be pointed out however, that the use of a single representation, implicitly
assumes that all the explanatory variables are exogenous. If this is not the case, then it is
suggested in the literature that one needs to use a vector auto regression (VAR) approach.
Furthermore, as long as the equation contains more than two variables, there is the
possibility that there could exist more than one cointegrating relationship. In such cases,
the traditional estimation procedures become inappropriate and again there is need to


44
She conceded that conclusions in relation to the HFCS subsector could be a
reflection of rigid assumptions that were made in relation to the supply and cost of
production of HFCS. For example, an unlikely result is that HFCS does not effectively
compete with sugar and is replaced by sugar when free substitution is allowed. In further
analysis in the said study, however, she showed that the above result was highly sensitive
to the assumptions made with respect to the HFCS supply elasticities. She concluded that
further analysis is needed into the behavior of HFCS, but that, by and large, the signs of
the welfare change and the distributional implications would remain the same.
Marks (1993) constructed a partial-equilibrium model in which he assessed the
long-run static welfare effects of alternative US sugar policies. He evaluated two policies.
The first focused on estimating the overall economic impact of the US sugar program for
crop years (1984/85 to 1988/89)years in which the domestic-foreign price differential
varied widely. The analysis considered the long-run effect of the program on sugar prices
in the US and the rest of the world (ROW), the economic welfare of US and foreign
consumers and producers, and U. S. government revenues. His second policy analysis
examined the implication of a 10% real reduction in the US sugar loan rate, by which the
US provides support to its sugar producers. His approach to modeling the US sweetener
market differs from most of the previous studies in two ways. Firstly, he estimated the
total demand for refined sugar plus HFCS (US sweetener demand) as function of a
composite index of their prices. Secondly, the prices of refined sugar and HFCS in turn
were modeled as functions of the US raw sugar prices, which according to Marks is
targeted by the policy makers. In addition, he estimated equations for aggregate sugar
demand and supply for the rest of the world (ROW). From these equations he was able to


CHAPTER4
CONCEPTUAL MODEL OF US SUGAR AND HFCS MARKETS
As the process of development takes place, sweeteners in general are increasingly
being consumed as raw materials and inputs of industrial production and to a lesser extent
by households as essential foodstuffs15. Thus, in developed countries and regions such as
those of the EU, US, Canada and Japan, between 60%-70% of the sweeteners consumed
are in such goods as cakes, soft drinks and ice creams. However, in the developing
countries the situation is the reverse where only about 30%- 40% of the consumption of
sweeteners is in the form of sugar containing products (Atlas, 1996). This has obvious
implications for the patterns of demand, the direction of technical progress and the
underlying scientific investigation. The purpose of this chapter therefore is two fold. First
we construct a conceptual model of the US HFCS and sugar markets focusing on the
derived demand nature of the demand for sweeteners. While the model is essentially
static, some attempts are made to examine the dynamics of the relationship between
HFCS and sugar. Second, the conceptual model is used to assess some of the likely trade
and welfare implications resulting from changes to the US sugar program and an
expanded HFCS industry within the framework of the proposed FTAA. In carrying out
this assessment two scenarios are examined. The first focuses on the likely impact
resulting from a partial liberalizing of the US domestic sugar program, while the second
considers the more unlikely situation of the complete abandonment of the program.
15 Industrial (manufacturing ) uses include those for food (bakeries and cereals, confectioneries, soft drinks
and canned and frozen dairy products) and non-food items such as ethanol.
71


120
within such industries can be view more as a complement (joint inputs)21 rather than
substitute commodity. The declining degree of substitutability is perfectly consistent with
a product that eventually established its role as a reliable and safe alternative to liquid
sugar.
21 Recall that for some industrial uses HFCS is inappropriate and sugar must be used. It is in this broad
sense that the two products might be viewed as complements in the production of a representative industrial
product, given current prices and level of technology.


35
within the region are the US (91.5%), Canada (3.5%), Argentina (2.5%) and Mexico
(2.1%). Production in Mexico started in 1996 by US interests and already output level is
close to that of Argentina and Taiwan. At the moment, the bulk of the com used in the
production of this commodity in Mexico is imported from the US. The 1997 regional
consumption of HFCS was estimated at 8.32 MMT. Again the US was the main
consuming country accounting for 91.1% of total demand. This was followed by Mexico
(3.7%), Argentina (2.5%) and Canada (2.5%). Only a small quantity of HFCS is currently
consumed in CARICOM and none of the member states engage in the production of the
commodity.
The considerable success in HFCS as a substitute for sugar is best illustrated in
the case of the US. Table 2.10 shows the changes in the per capita consumption and the
relative market shares of caloric sweeteners consumed in the US over the period 1975 to
1997. And, Figure 2.2 shows a comparison of the trends in per capita consumption of
HFCS, total com sweeteners and sugar for the same time period. In particular, the Figure
reveals that between 1975 and 1997, the per capita consumption of sugar fell from 89.2
lbs. to 67.1 lbs. after reaching a low of 60.8 lbs. in 1986, while the per capita
consumption of HFCS increased from 5 to 61.4 lbs. over the same period. This meant that
whereas in 1975 sugar accounted for about 76 % of the caloric market shares and HFCS
only 4%, by 1997 the market share of sugar had fallen to 43% while in the case of HFCS
it had risen to 40%. The rapid and considerable increase in the production and
consumption of HFCS in the US and the concomitant displacement of a portion of the US
sugar demand, has come largely at the expense of the sugar quota holders and by
implication the US sugar refiners. This is easily seen when one considers the fall in US


11
5. Given the above, membership of CARICOM countries in the FTAA will in the
short to medium term have potentially negative economic impacts on the sugar
industry of the subregion.
Objectives
The study has two general objectives. Firstly, to qualitatively assess the potential
economic impact that likely changes to the US sugar program, and an expanded High
Fructose Com Syrup (HFCS) industry, might have on the CARICOM sugar industry
within the framework of the proposed Free Trade Area of the Americas (FTAA).
Secondly, to provide quantitative estimates for the HFCS demand configuration that can
be used in subsequent modeling exercises. Specific objectives are to:
1. Assess selected trends in the US sweeteners industry.
2. Develop a conceptual model of the US sugar and HFCS markets.
3. Qualitatively assess the trade and welfare implications of likely changes in the US
sugar program and an expanded HFCS industry on the sugar producing countries
of the FTAA, with specific reference to the CARICOM sugar producers.
4. Develop and estimate the derived demand for HFCS in the US sweeteners market.
5. Investigate the demand side dynamics within the HFCS industry, and
6. Identify critical policies regional planners might wish to take into consideration in
their ongoing integration negotiations, as they relate to the future of the
CARICOM sugar industry.


66
Relevance of the State-Space Model and Kalman Filter to the Current Study
The state-space framework was used in this study to estimate the derived demand
for HFCS. First, as pointed out elsewhere, the industry is relatively new which limits the
number of available observations (approximately 22). Also as discussed earlier, the VAR
based cointegration approach is biased with finite samples. In addition, one runs the risk
of encountering the vanishing degree of freedom problem referred to by Johnston and
Dinardo (1997). Second, there is strong evidence (Chapter 2 ) to suggest that the
parameters of the demand equation have been varying overtime For instance, the
growth rate of the demand for the product has varied considerably over the investigated
period. This suggests some kind of a stochastic trend rather than a deterministic trend.
Also, as pointed out elsewhere, the adoption of the product was gradual, implying that
parameters which do not vary over time, since the hyperparameters, which allow the parameter to vary, in
these cases would be set to zero, resulting in the estimation of fixed coefficients. These models in
forecasting put more weight on the most recent observations; the fasto- the level and slope change, the
more past observations are discounted.The state-space approach allows one to work in level data, rather
than differencing, even in cases where the data are nonstationary. Working in level form makes
interpretation of results much easier. He notes that traditional time series analysis, in situations where the
data is nonstationary, stresses the role of differences. Within the state-space framework it is not usually
necessary to difference in order to specify a suitable model. A similar view is expressed by Aoki (1990)
that while economic time series are usually nonstationary because circumstances facing optimizing
economic agents change with time and do not remain the same, the Kalman filter can deal more effectively
with nonstationary time series than can the traditional time series methods. Within the state-space
framework it is not necessary to test for unit root, as concerns over the degree of integration of a series are
not crucial. The state-space frame work is flexible enough that if the slope parameter is deterministic, the
hyperparameters which allow it to change over time will be estimated as zero or close to zero, hence little is
lost if indeed the series is not integrated The situations where VECMS can be usefully employed are quite
limited and one should have reservations about using them to provide a general vehicle for modeling
economic time series. These reservations stem from the fact that the VAR-based cointegration methods are
based on auto regressive models and like the unit root tests can have very poor statistical properties. In
addition, it is very difficult to fit autoregressive models to data with slowly changing seasonality or trend.
The strongest objection, centers not on the feet that when there are two or more cointegrating relationships
they can only be identified by drawing on economic knowledge, but rather that the VECM does not provide
a sensible vehicle for modeling the short-run...since it confounds long-run and short-run effects. As an
alternative to the VECMs Harvey proposes the use of multivariate structural time series model within the
state-space framework noting that such modeling can be formulated to incorporate long-run components of
cointegration directly by means of common trends. An added advantage of the state-space framework and
the Kalman filter is that the researcher need not understand the workings of the filter to apply the technique
and as such is free to concentrate on the selection of a suitable model and its interpretation.


16
Table 2.1. Ranking of Countries of the Western Hemisphere on Basis of Population,
Land Size and GDP
Rank
Population
1995
(MiniS"'
ns)
Land Size
(000 sq.
Km.)
GDP
1995
(US$ M
1990=100)
34
St. Kitts & Nevis*
0.04
Barbados
0.3
St. Kitts & Nevis
150
33
Antigua & Barbuda
0.06
Grenada
0.3
Dominica
178
32
Dominica
0.07
St. Kitts & Nevis
0.3
St. Vin. and Gren.
218
31
Grenada
0.09
Antigua & Barbuda
0.4
Grenada
229
30
St. Vin. and Gren.
0.11
St. Vin. and Gren.
0.4
Suriname
334
29
St. Lucia
0.14
St. Lucia
0.6
Antigua & Barbuda
366
28
Belize*
0.21
Dominica
0.7
Belize
491
27
Barbados*
0.26
Trinidad & Tobago
5.0
St. Lucia
509
26
Bahamas
0.27
Jamaica
11.0
Guyana
602
25
Suriname
0.42
Bahamas
14.0
Haiti
1,642
24
Guyana*
0.82
El Salvador
21.0
Barbados
1,717
23
Trinidad & Tobago*
1.26
Belize
23.0
Nicaragua
2,590
22
Jamaica*
2.53
Haiti
28.0
Bahamas
3,053
21
Panama*
2.63
Dom. Republic
49.0
Honduras
3,378
20
Uruguay*
3.19
Costa Rica
51.0
Jamaica
4,171
19
Costa Rica*
3.07
Panama
76.0
Trinidad & Tobago
5,707
18
Nicaragua*
4.54
Guatemala
109.0
Bolivia
6,496
17
Paraguay*
4.83
Honduras
112.0
Panama
6,570
16
El Salvador*
5.64
Nicaragua
130.0
El Salvador
6,674
15
Honduras*
5.95
Suriname
163.0
Costa Rica
7,027
14
Haiti*
7.18
Uruguay
177.0
Paraguay
7,177
13
Dom. Republic*
7.91
Guyana
215.0
Dom. Republic
7,341
12
Bolivia*
8.06
Ecuador
284.0
Guatemala
9,706
11
Guatemala*
10.62
Paraguay
407.0
Uruguay
11,431
10
Ecuador*
11.46
Chile
757.0
Ecuador
15,132
9
Chile
14.20
Venezuela
912.0
Peru
47,618
8
Venezuela
21.64
Bolivia
1,099.0
Chile
48,326
7
Peru*
23.53
Colombia
1,139.0
Colombia
56,379
6
Canada*t
29.61
Peru
1,285.0
Venezuela
64,980
5
Argentina*t
34.59
Mexico
1,958.0
Argentina
196,949
4
Colombia*
35.10
Argentina
2,767.0
Mexico
250,936
3
Mexico*t
94.78
Brazil
8,512.0
Brazil
432,433
2
Brazil*
159.22
United States
9,809.0
Canada
608,658
1
United States*f
263.06
Canada
9,976.0
United States
6,173,900
* Major sugar producing and exporting countries
t Major HFCS producers
Source: Adapted from Bernal (1998)
of these CARICOM countries are relatively small, open and undiversified economies
with over one-quarter of their total exports concentrated on one or two products. In the
case of countries such as St. Kitts and Nevis, Dominica and St. Vincent and the


Ill
(obtained from the GDP index). Information on the interest rates was obtained from the
Economic Report of the President (1997).
Estimation Results
The remainder of this chapter discusses the empirical results achieved using the
two-step estimation process, with equations 5.4 and equation 5.5. First, for simplicity
equation 5.4 was estimated as a double log using OLS. However, the diagnostic results
showed that it had a goodness of fit value, as reflected by the adjusted R2, of 0.63 with
two of explanatory variables being significant at the 5% level. Moreover, the Durbin-
Watson value was relatively low (1.09) indicating a possibility of the existence of serial
correlation among the error terms. The equation was re-estimated, assuming the existence
of first-order serial correlation, and the maximum likelihood technique was employed.
The new equation resulted in an adjusted R2 of 0.79 and the Durbin-Watson value
increasing to 1.72. The prices of com, sugar and the final good were all shown to have
significant impacts on the pricing of HFCS in a manner consistent with a priori
expectation. The results of both equations are shown in Table 5.1 as equations (5.4a) and
(5.4b), respectively.
With the information obtained from equation (5.4b), in particular the estimates of
price of HFCS(LRPhi*), the Kalman filter was used to estimate equation (5.5). The values
of the most recent set of estimates of the parameters together with their p-values are
presented in Table 5.220. This Table shows that the trend (pt), price of HFCS and the
20 Since the HFCS industry is one of imperfect competition the observation at any point in time t lies on the
demand curve corresponding to the price at which the MC = MR. The use of time series data to estimate the
derived demand curve implicitly assumes that all such points lie on a given demand curve. However, in


107
linear form or a suitable approximation of a linear form and then test such assumptions
using test procedures such as the Ramsey RESET or the Box-Cox. The latter approach
was used to test the linear specification versus the log-linear specification. The results of
the procedure showed that the likelihood function was maximized with the value of X =
.01. This indicated that the log-linear function was superior to the straight linear form and
provided the criterion for choosing this particular functional form to represent the derived
demand
A second issue relates to the exogenity of the independent variables. This issue
becomes important because it assist in determining the appropriate estimating technique
to be used and hence the validity of the results. For example, in applying OLS to estimate
a single equation, the assumption is made that all the independent variables are
exogenous to the system (Darnell, 1994). If such is not the case then it poses an
estimation problem since, in this case, the disturbance term in the equation would become
a determinant of both the price and quantity. In other words, a right-hand side variable is
correlated with the disturbance. In standard econometrics this can be handled in one of
several ways, including the use of instrumental variables, two stage least squares (2SLS)
or obtaining the reduced form of the price variable. Similarly to the case of the OLS
regression, one of the requirements for utilizing the state-space model with a single
equation is that the independent variables are exogenous to the system being estimated
(Harvey, 1987; Engle and Watson, 1987; Darnell, 1994).
Owing to concerns surrounding the price of HFCS, a test of exogeneity (Granger
causality test) was performed on the quantity of HFCS demanded and the price of HFCS.
The results confirm the suspicion that the HFCS price was not exogenous. The null


78
Demand for Sugar:
Ds = Dcs + Dls
(4.3)
Supply of Sugar:
Ss = Sc{)+Sb(.)
(4.4)
Import Demand:
Id = Ds Ss > 1.256tonnes
(4.5)
Equilibrium Condition: Ds = Ss + Id
where:
(4.6)
Ps
domestic price of sugar in the US
Pf
index of sugar and HFCS containing products
Ph
price of HFCS
Qh
quantity of HFCS supplied
Sc & Sb
supply of cane and beet sugar, respectively
Z
a vector of other factors
The US HFCS Subsector
Turning attention to the HFCS subsector, the demand curve for HFCS is
conceptualized, as shown in Figure 4.5. Such a demand configuration is based on the
following grounds: 1) the observation that although HFCS is almost a perfect substitute
for liquid sugar (LS) it has always been sold at a discount to LS on a sweetness
equivalency basis (Fig. 2.3); and 2) the dual nature for the commodity, as discussed in
chapter one, which suggests that HFCS can only be substituted for sugar within a certain
range of the sugar demand function. It should be pointed out that above the controlled
price for sugar, Pt, the particular shape assumed by the demand curve is very ambiguous.
This is due to the fact that the price of HFCS has never exceeded that of sugar or even
come close to it. If one is willing to assume the conclusion reached by Tanyeri-Abur


2
Caribbean and some of their Latin American neighbors vis--vis NAFTA-are likely to
face losses due to trade and investment diversions, higher non-tariff barriers (NTBs) and
possibly higher tariffs (Blanco, 1997).
As regionalism grows, many developing countries must choose among alternative
integration strategies. Accordingly, they are reassessing existing economic and trading
arrangements as well as considering the prospects of forming new relations and strategic
alliances so as to preserve, consolidate or establish stronger trading positions. The
Caribbean Community (CARICOM)2a group of 14 countries in the Western
Hemisphereis one such alliance of developing countries. With a population of
approximately 6.0 million and a total gross domestic product (GDP) at market prices of
only about US $14.0 billion, these countries represent a very small market by world
standards (Bryan, 1995). In addition to the Lome arrangement, these countries have
benefited a great deal from other marketing arrangements such as Caribbean and Canada
Agreement (CARIBCAN), Caribbean Basin Initiative (CBI), and GSP. Both the US and
the EU, however, have now separately proposed forming Free Trade Areas (FTAs) with
CARICOM with a tentative implementation date being the year 2005 or there about.
While details of such arrangements are still being refined, they will have far reaching
consequences for CARICOM's relationships with Europe, the other ACP countries, the
Americas and much of the developing world. Within CARICOM, these arrangements
2 Member countries of the CARICOM include: Antigua and Barbuda, The Bahamas, Barbados, Belize,
Dominica, Grenada, Guyana, Jamaica, Montserrat, St. Kitts/Nevis, St. Lucia, St. Vincent and the
Grenadines, Suriname, and Trinidad and Tobago. Haiti has formalized arrangements for full membership as
of August 1999 bringing the total membership to 15. However, in view of that countrys recent accession,
this study does not include it in the analysis.


89
substantially so that it falls well below the point of any kink in the demand curve. And, as
noted earlier, the location of the point of the kink is a function of the dynamics of the
rapid economic and technological changes occurring within the HFCS industry.
The third factor mentioned above which influences the positioning of the kink is
the supply behavior of HFCS. This factor is closely related to the previous two factors
discussed. Marks (1993) in alluding to the pattern of the substitution between HFCS and
sugar asserts that it is a function of both the demand for sweeteners and the supply
behavior of HFCS. Elaborating on the latter factor, he postulated that a reversal of the
time path of the real US sugar prices would hardly likely bring about the same
substitution pattern in reverse. His claim for the likely failure of the reverse pattern of
substitution to occur is based on the view that the HFCS industry is in place with
hundreds of millions of dollars of sunk costs to protect and the advantage of leaming-by-
doing and other scale economies (p. 80). The implication of his assertion is that due to
the advancement made in the technology of producing HFCS, the benefit of product
development and the need to protect investment, the shifts in the sugar demand could be
structural. Moreover, on the demand side there also exist switching costs associated with
changing from one form of sweetener to the other. For example, in switching from liquid
sugar to HFCS certain costs would have been incurred both in modifying recipes and in
putting in place certain physical infrastructure such as storage facilities. While these
factors do not guarantee complacency on the part of the HFCS supplier, they imply a
lowering of the kink and that the price of sugar would have to fall substantially to bring
about a reversal of the current trend of sweetener usage.


131
that is within the vicinity of the price of HFCS. The empirical findings therefore lend
supports to some of the earlier findings of the conceptual model, with regard to
uncertainty surrounding the possible outcome in the event that the US completely
abolishes its sugar program.
Policy Implications and Recommendations
In general, the findings of this study confirm inter alia, our hypotheses (See
Chapter 1). In particular, they supported the hypothesis that unqualified membership of
CARICOM countries in the proposed FTAA, could in the short to medium term have
potentially negative economic impacts on the subregion's sugar industry. Also, that an
expanded US HFCS industry could pose a threat to the subregional sugar industry. It
should be noted however, that the study did not specifically consider the competitiveness
of the CARICOM sugar industry vis-a-vis other sugar industries in the region. This was
not undertaken since it is generally accepted that despite the best efforts to modernize the
CARICOM sugar industry it can never be as competitive as segments of the US sugar
industry or those in countries such as Columbia, Brazil and Guatemala. As pointed out by
Kennedy et al (1998), the limitation of land and other resources are severe constraints in
CARICOM countries. A study conducted by LMC and reported by Haley (1998a)
suggests the field cost of production in the US is comparable to that of some of the
CARICOM countries. However, the US enjoys in some areas as much as a 40%
advantage over non-US countries factory costs. There are however, significant areas in
which the CARICOM sugar industry can improve its productivity and efficiency
considerably (Caribbean Update, 1999, James, {personal interview, 1999)).


88
liquid sugar, the HFCS substitute. It should also be noted that, other factors remaining
constant, the movement of the segment of the curve ke occurs along the dotted portion of
the demand curve akb as shown in Figure 4.9a for the case of a shift from ake to ake .
The nature of this movement has important trade implications. For one thing, it implies
that lowering the US support price for sugar (Pt) would induce only a slight increase in
the quantity of crystallized sugar demanded if the final support price is still above the
kink. This postulated movement is contrary to popular view of a substantial increase in
sugar demand associated with sugar price reduction in the US (Atlas, 1996; Schmitz and
Christian, 1993; Sudaryanto, 1987). This is unlikely, unless the price of sugar is lowered
Price Price
Fig. 4.9. Hypothetical supply and demand for sugar, and hypothetical demand for HFCS,
respectively


75
challenge by that industry to this segment of the market. The second demand curve
represents the demand for liquid sugar (LS). As argued earlier, it is liquid sugar for which
HFCS is almost a perfect substitute (Figure 4.2b). The presence of this substitute (HFCS)
causes the demand curve for LS to become relatively elastic (rotating to the right as
shown in Fig. 4.2b) reflecting the fact that manufacturers are much more apt than
households to switch to a lower cost substitute because of the need to remain
. 17
competitive .
This characterization of the demand for sugar although novel, is nonetheless
anchored by evidence from the literature. For instance, Sudaryanto (1987) in partitioning
the US sweeteners market into a direct and derived demand for sugar found that the
elasticity of the derived demand was far greater than that for direct consumption. His
explanation was that since the consumption of final uses (direct) had approached
saturation, the responsiveness to a change in the price of sugar would be minimal
(inelastic). On the other hand, the demand for sugar in the food and beverage industry
was more elastic since many more substitution possibilities were available. The
Sudaryanto explanation of the differences in sugar demand configuration is somewhat
similar to that advanced by Thomas (1985). Specifically, Thomas noted that the US
market favored the development of the HFCS industry, since there existed high price
elasticity of demand for sugar among industrial sweetener users. These users, he
suggests, are far more ready than households to substitute one sweetener for another,
17 Economic theory holds that the price elasticity of demand for any good depends on many economic,
social, and psychological factors. Included among the economic factors are the availability of close
substitutes and necessity of a good. In general, necessities tend to have inelastic demand whereas goods
with close substitutes tend to have a more elastic demand (Mankiw, 1997)


17
Grenadines and St. Lucia, one primary exported commodity, either raw sugar or bananas,
accounts for more than 50% of exports.
Review of World and Western Hemisphere Sugar Markets
Sugar4 is one of the few commodities that can be produced from crops grown in
both temperate and tropical climates. A consequence of this is that widespread
governmental interventions have long been a feature of the world sugar market, making it
one of the most volatile of all primary commodity markets (Borrell and Duncan, 1993).
To gain some independence from the volatile world market, producers in most countries
have lobbied their governments to operate intervention schemes aimed at controlling
domestic prices, supply, and demand. Numerous pricing tiers, and trade and stockholding
mechanisms (policy wedges) have been devised and have resulted in distorted
production, trade, consumption, and world price. The trends in real world and US market
prices of raw and refined sugars are shown in Figure 2.1 for the period 1985-98. Among
other things, Figure 2.1 reveals the relative volatility of prices in the world market
compared with those in the US market.
World Production and Market TrendsSugar
Global sugar output in 1998-99 is projected by the US Department of Agriculture
(USDA) to increase by about 1.14 million metric tons (MMT) (0.9 percent) to 126.5
MMT, whereas consumption is projected to increase by 0.4 percent to 127.5 MMT,
reflecting some degree of market tightening. The five leading world sugar producers and
4 In this study, the word sugar refers to sucrose obtained from sugarcane and sugar beets.


68
refers to the demand for a factor of production or product that is derived from the demand
for other goods. Neoclassical theory suggests that a profit maximizing firm will employ a
combination of inputs up to the point at which the additional cost of employing one more
unit of that input is equal to the additional revenue generated from the presence of that
that input. Given such a framework, a firms input demand functions (derived demands)
can be obtained in one of several ways. First, by solving the firm's profit maximization
objective function with respect to the given input. Alternatively, a system of derived
demands can be obtained by application of duality theory. This theory suggests that given
either the firms cost function or its profit function and assuming that such functions
satisfy a set of regularity conditions, the system of derived demands can be obtained by
applying Shephards lemma and Hotellings lemma to the given cost and profit function,
respectively (McFadden, 1978). For example, given a simple firm which produces one
output using three inputsthe third of which is fixed in the short runthe profit value
(optimized) function can be represented as follows:
n* = 7t* ( P, v, w, Z) = Max [Pq vxi WX2 : f(xi, X2, Z) > q ]
where n represents the maximum profit that the firm may obtain by allowing a subset of
inputs and output to vary while another subset of input is held fixed; P is the price of the
output; v and w are the prices of inputs xj and X2, respectively; xi and X2 are the levels of
the variable inputs used in the production process, while Z is the fixed input; q is the
quantity of output and f represents the production function. Since the profit function as
specify above is the result of a maximization process, then by applying Hotellings
lemma to the function the derived demands as well as the output supply can be obtained
as follows:


received from the Caribbean Agricultural Research and Development Institute (CARDI)
during my initial years. Special mention must be made of Mr. Hugh Saul whose faith in
me and timely intervention made it possible for me to receive the latter financial support
and take up the offer at the University of Florida.
Perhaps my greatest depth of gratitude however, is owed to my wife, Sabrina,
children, Latoya and Akeem, my mother and my godfather, Attli, for their encouragement
and for the sacrifice they have made in favor of this effort. To them this dissertation is
gratefully dedicated.
Finally, I wish to express my sincere gratitude to all those who stood in my comer
cheering my every move. Among them are, my dear friend Dr. Kenrick Jordan, my most
ardent supporter, Dr. Pauline Lawrence, Dr. Curtis McIntosh, Mr. Karl James, Greta and
Danny Roberts.
IV


109
where, LRPHt is as defined earlier; LRPct is the log of the real price of com; LRPei is the
log of the real price of energy; LRIt is the log of the real long term interest rates; vt is the
usual error term; and the remaining variables are as defined in the demand equation (5.1).
The real price of com (LRPct) was included since it constitutes the main input
used in the production of HFCS. The price series for the yellow dent com was chosen
since this represents the variety most commonly used in wet-milling process.
Consideration was given to using the net cost of com starch, (i.e., the price of the starch
after the returns from the major byproducts have been netted out) rather than the gross
price of the com. However, econometrics experiments with both variables showed that
the price of com was the better predictor of the price of HFCS. The cost of energy
(LRPEt) was included because of the highly capital intensive nature of the industry.
Energy cost is the second highest variable cost used in the production of the product.
Finally, long-term real interest rate (LRIt) was included to reflect the cost of borrowed
capital. This was computed as the difference between the long term nominal interest rate
and the inflation rate. Equation (5.4) was estimated by OLS and the predicted value of the
dependent variable LRP1It* was obtained and used as a proxy for the original HFCS price
variable. The respecified derived demand equation which utilized the Kalman filter is:
LQm ~ Pt + PitLRPnt + p2tLRPst + P3tLRPsDt + P4tLRPFt + et (5.5)
A third issue that arose had to do with supplying the values for the
hyperparameters and the initial values of the state vector within the framework of the
state-space representation (See Chapter 3, section on Kalman filter). In situations in
which the initial values of the state vector are not provided by the researcher, the first k
(the number of elements in the state vector) observations are regressed using OLS and the


60
m
y. =Mx +IXxt-¡+s
(3.7a)
i=0
M, = Mx-l+PxA + 7t
(3.7b)
A=A-,+£
(3.7c)
Sn =SA+vx
(3.7d)
where it is assumed that s, ~ NID(0,oE2), qt~ NID(0,Ot,2 ), £,~ NlD(0,c^2), and v,~
NID(0,ov2). The component pt is the trend, whereas p, is the slope of the trend and 6¡t are
the parameters of the explanatory variables, which are assumed to vary over time. The
introduction of the stochastic disturbance terms r|t and allows the level and slope of the
trend to change slowly over time. Likewise the disturbance term v,, allows 6it to be
generated by a random walk. More formally, the above state-space representation with
the simplifying assumption of no lags is:
Y, = (1 0 l)o, + e, (3.8)
where
1 1 0
Mx-i
7,
a, =
Px
=
0 1 0
PxA
+
kJ
0 0 1
k J
(3.9)
In terms of state-space description, such models usually consist of two parts: 1)
the measurement equation which describes how the data actually observed is generated
from the state variables (equation (3.7a) or equation (3.8)); and 2) the transition
(dynamic) equation which describes the evolution of a set of state variables (equations
(3.7b to 3.7d) or equation (3.9)). The vector at comprising the elements |it, pt, and 6, is
regarded as the state vector to be estimated and the measurements that are used to


77
remains a market demand for liquid sugar. Thus, Xi and X3 indicate the quantities of
sugar produced and consumed, respectively. As such X3 Xi represents the US tariff-rate
quota (TRQ), which cannot fall below 1.14 MMT. X2 represents the quantity of CS
demanded and X3- X2 is the quantity of LS demanded. It should be noted that the
quantity of LS demanded does not represent the total demand for liquid sweetener that
would include the amount being satisfied by HFCS. Finally, Pw gives an indication of the
Fig. 4.4. Hypothetical US supply and demand for sugar
world market price in relation to the support price. Thus, the US sugar market might be
represented by the following equations:
Demand for CS:
Dcs = f(Ps,PF,Z)
(4.1)
Demand for LS:
Dls = f(Ps,PH(.),Qn(.),PF,Z)
(4.2)


67
time was needed for firms to learn about the product, determine its compatibility with the
manufacturing process, to formulate new recipes as well as make adjustment to existing
ones and to assess new product attributes. Also, with regard to some of the other
parameters such as the own price and cross price elasticity, for similar reasons mentioned
above, there is a strong suspicion that these might have undergone changes over the time
period, in view of the considerable changes which have taken place in the development of
the product making it far more attractive to the end users and consumers.
Third, is the fact that the objective of the study is not so much concerned with
whether there exists a long term relationship among the chosen explanatory variables, as
it is to identify the dynamic path of adjustments which might have occurred in the
industry and to draw inferences. As a consequence, the issues of cointegration and unit
root, though important, are of lesser concern within the present context. As Engel and
Watson (1987) observed that in most of the studies where time-varying parameters have
been used the predominant consideration seems to be on the stability of the regression
equation, only in a few cases have the variations been interpreted as economically
important. The current study intends to add to those few studies.
Finally, the use of the state-space framework within this study is an attempt to
make use of some of the theoretical and practical advances that have been made in the
area of systems literature and to provide yet another case where the techniques have been
used in economics.
The Concept of a Derived Demand
The term derived demand is normally used to denote the demand for inputs that
are used to produce the final product. According to Shim and Siegel (1995) the term


125
similar to the situation in the Stackelberg model in which one player was sophisticated
(HFCS producers) and the other was naive. The users of HFCS were assumed to be price
takers. Suppliers established the best-response price, with consideration to the situation in
the sugar market and their objectives of maximizing profit and/or maintaining market
shares. Industry experts believe the long-term prospects of the HFCS industry appear
more favorable than its sugar substitute. This belief played a central role in modeling the
long-run situation in the HFCS industry. This view stemmed from the fact that the
product is the outcome of biotechnology with prospects of further significant
improvements in technologies, compared with sugar that is essentially a chemical
product. A likely implication is that the cost of producing HFCS could decrease
considerably as external economies are realized, providing real justification on the part of
the com refiners to actively seek expansion of the market for HFCS products.
Theoretical Findings
The impact of a partial liberalization of the US sugar program and a complete
liberalization of the program within the framework of the FTAA were also considered in
Chapter 4. The conceptual trade model utilized the typical large country trade
assumptions and assumed that the effect of an expanded HFCS industry could be
captured through its impact on the excess sugar supply and demand curves. The results of
the first investigation revealed the importance of the kink in the sugar demand
configuration. If in a situation of partial liberalization of the US sugar program, the US
domestic support price were to be reduced to a level at which it was still above the kink
in the sugar demand then, there would be an increase in the US demand for sugar and
most likely an increase in the levels of imports. However, the increase in the demand for


79
(1990) that if the price of HFCS should ever exceed that of sugar the industries would
revert to using sugar completely, then this would suggest that the demand curve kinks at
or just above the controlled price and become highly or even infinitely elastic. However,
if one were to accept the arguments put forward by Marks (1993, p. 80 ) that "a reversal
of the time path of the real US sugar prices would hardly likely bring about the same
substitution pattern in reverse" then one might conceive of the portion of the demand
curve above the controlled sugar price, tapers off as shown in Figure 4.5.
Price
Several factors could account for a shift in demand for HFCS. As discussed in Chapter 3
in the section on derived demand, the demand for an input depends considerably on the
demand for the final good, hence any change in the demand for the latter will cause the
derived demand for the input to also shift. Thus, a shift in the demand for HFCS could
come about owing to any or a combination of the following: 1) changes in the prices of


101
Using his estimate of 13.12 as a rough guide, the implication is that the opening up of the
US market should not seriously affect the world market price since a 100 % increase in
the quantity imported by the US would only result in an approximately 7.6 % increase in
the world market price, ceteris paribus.
Figure 4.12a also indicates a final outcome in which, it is possible that both liquid
sugar and HFCS are consumed with sugar regaining some of the market share that was
lost to HFCS. One can envisage a situation similar to that which exists in Canada, where
HFCS is still consumed in an environment where there is no special protection given to
HFCS and sugar prices reflect those of world market prices. Thomas (1985) points out
that in spite of such a situation, HFCS consumption captured approximately one-third of
the sweeteners market in Canada.
Welfare implications
The welfare effects of scenario II are similar to those that occurred in the partial
liberalization case under scenario I, although the magnitude of the benefits and losses will
differ considerably. Noticeably, the sugar producers in the US are expected to experience
considerable loss, while the gains to the HFCS producers would be less than in the
previous case. This latter situation will result since HFCS can be expected to lose some of
the market share as it competes with liquid sugar. The situation with regards to producer
in the rest of the world, including the FTAA is not clear, but the indications are that they
would again suffer losses or at best be made no worse off than previously. There would
however, be a noticeable redistribution of the benefits and losses among the producers,
with the relatively high cost CARICOM sugar producers being unfavorably disposed.


124
a justification given for use of the methodology in this study. This methodology: 1)
allows for modeling of time-varying parameters, if the situation so warrants and provides
estimates at any point within the sample permitting one to estimate the path parameters
have taken over time; 2) deals fairly easily with nonstationary time series ; and 3) allows
modeling to be done in level versus differenced form which facilitates interpretation. The
flexibility of the state-space model makes it appropriate for estimation in situations, such
as in the case of derived demand for HFCS, where there are indications that the industry
is still evolving. The chapter was concluded with a brief theoretical discussion of the
concept of a derived demand.
A major part of the dissertation was concerned with constructing the conceptual
model of the HFCS and sugar markets used in assessing the implications. This was done
in Chapter 4. For convenience, the US demand for sugar was separated into the demand
for crystallized sugar (CS) and liquid sugar (LS). The CS represents the demand for
tabletop sugar plus sugar used in the industrial sector in processes in which it was not
technically feasible to use HFCS. The LS represents the demand for sugar for which
HFCS provides an almost perfect substitute. This kind of classification results in a kinked
demand for sugar. At prices above the kink there is no effective demand for liquid sugar.
The model is essentially static and therefore only provides a first approximation of the
likely occurrence. Dynamics were inserted through a comparative static framework with
regard to the interrelationship between the sugar and HFCS price. In modeling the HFCS
market, the imperfectly competitive framework within which the industry operates was
recognized. Consequently, it was argued that a supply function for HFCS did not exist
and that the supply could be better explained in terms of a reaction function, somewhat


48
Serial Correlation
Serial Correlation (autocorrelation) is the extent to which a time series variable
(usually the error term of the time series model), lagged one or more time periods, is
correlated with itself (Shim and Siegel, 1995). When such a situation exist it has
implication for the classical linear regression (OLS) model. The OLS estimator of the
parameters remains unbiased but no longer efficient and in fact is asymptotically
inefficient. Moreover, the standard formula used by the OLS to compute the estimate of
the variance of the errors is no longer unbiased and could either over- or underestimate
the true variance. The implication is that all tests of reliability, using the t- and F-
distribution, which rely on the estimated variance-covariance matrix of the estimated
parameters, become invalid.
Several tests such as the Durbin Watson (DW) statistic are available to detect the
presence of serial correlation and depending on the causes, various remedial actions can
be taken. In pure times series analysis, taking account of the pattern of serial correlation
usually allow for better forecast of future observations. Darnell (1994) notes that
detection of a pattern of first-order positive autocorrelation by means of the residuals can
result from several factors other than autocorrelated true errors (unobserved). These
factors include mispecification of functional and structural breaks in the relationship
(time-varying parameters). Consequently, he suggests that a respecification of the model
might be more appropriate and only when the evidence strongly suggest autocorrelated
true errors should the model be re-estimated by feasible generalized least squares or
maximum likelihood.


87
com, the prices of the by-products, the price of domestic sugar; and 3) the supply
behavior of HFCS. These three factors are discussed below in highlighting some of the
dynamics of the sweetener market.
First, assuming other factors remain constant, it can be expected that an increase
in the number of available substitutes and the closer the substitutes on a per unit
sweetness equivalency basis the lower will be the positioning of the kink. This follows on
the assumption that there does not exist any switching cost for the manufacturers (user of
sweeteners) in utilizing one input over another. This latter assumption however is not
entirely accurate since such costs do exist in one form or the other. However, the general
principle applies in cases where such costs are negligible compared with the anticipated
increase in profits resulting from the change.
Second, and of greater importance is the relative price movement. For instance a
decrease in the price of HFCS is expected to cause the kink in the sugar demand curve to
shift downward, from ketoke , implying a lower level trigger price that sugar would
have to be sold in order for it to become an effective HFCS substitute (Figures 4.9a and
4.9b). The converse situation would also hold, that is, an increase in the price of HFCS
relative to the price of sugar shifts the kink, ke, upwards, implying an increase in the
trigger price at which sugar becomes an effective substitute for HFCS. These movements
postulated are consistent with economic theory which holds that a reduction in the price
of one commodity that is a close substitute for another commodity will cause a downward
shift in the demand for the substitute commodity. Thus, the change in the relative price of
HFCS only caused that segment of the sugar demand, (below the point of the kink in
Figure 4.9a, which is ke) to shift, since it is this segment which represents the demand for


141
Caribbean Update. Kal Wagenheim: Maplewood. Vol 14 #11 Dec. 1998.
Caribbean Update. Kal Wagenheim: Maplewood. Vol 14 #11 July 1999.
Carman, F. "A Trend Projection of High Fructose Com Syrup Substitution for Sugar."
American Journal of Agricultural Economics, 64(1982):625-633.
Cheung, Y. and K. Lai. Finite-Sample Sizes of Johansens Likelihood Ratio Tests for
Cointegration. Oxford Bulletin of Econometrics and Statistics, 55(1993):313-28.
Congress of US, The 1997 Economic Report of the President: Economic and Budget
Outlook: Hearing Before the Joint Economic Committee, Congress of the US,
One Hundred Fifth Congress, First Session. Washington DC: US GPO, 1997
Contrearas, A. Agricultural Issues Relating to the 1999 World Trade Organization
Multilateral Trade Negotiation. House Agriculture Committee Field Hearing,
January 22, 1998.
Cooley, T. and E. Prescott "An Adaptive Regression Model," International Economic
Review, 14(1973):364-71.
Darnell, A. A Dictionary of Econometrics. Aldershot: Edward Elgar, 1994
Engle, R. F. and M. Watson A Time Domain Approach to Dynamic Factor Analysis
and Mimic Models, Les Cahiers du Seminaire dEconometric, 22,(1980): 109-22.
Engle, R. F. and M. Watson The Kalman Filter: Applications to Forecasting and Rational
Expectations Model In T.F. Bewley (ed,)Advances in Econometrics: Fifth
World Congress. Cambridge: Cambridge University Press, 1(1987): 245-84.
Flemming, J. Inflation, Oxford: Oxford University Press, 1976
Friedman, M. Optimal Expectations and the Extreme Information Assumptions
of Rational Expectations Macro Models, Journal of Monetary Economics, 5
(1979): 23-41.
Gemmill, G. The World Sugar Economy : An Econometric Analysis of Production and
Policies. Michigan State University. Agricultural Economics Report No. 313,
1976.
Gonzales, A. Europe and the Caribbean: Toward a Post-Lome Strategy," in The
Caribbean: New Dynamics in Trade and Political Economy, A. T. Bryan, ed.,
North-South Center, 1995.
Group of Experts, Overcoming Obstacles and Maximizing Opportunities: Smaller
Economies and Western Hemispheric Integration, http://ncfap.org/ftaa-
wtol.htm. 1998.


Evaluation and Critique 46
Time Series Analysis 47
Serial Correlation 48
Stationary 49
Cointegration and Error Correction Models 51
Time-Varying Parameters 55
The State-Space Framework 57
The State-Space Model 57
Kalman Filter 61
Advantages of State-Space Approach and the Kalman Filter 64
Relevance of the State-Space Model and Kalman Filter to the Current Study 66
The Concept of a Derived Demand 67
4 CONCEPTUAL MODEL OF US SUGAR AND HFCS MARKETS 71
The US Sugar and HFCS Subsectors 72
The US Sugar Subsector 73
The US HFCS Subsector 78
The Interrelationship between the US Sugar and HFCS Markets 86
Trade Implications of Changes in the US Sugar Program within an FTAA 90
Free Trade Area 90
Scenario I- Partial Liberalization of the US Sugar Program 92
Trade implications 92
Welfare implications 96
Scenario II- Complete Liberalization of the US Sugar Program 98
Trade implications 98
Welfare implications 101
5 EMPIRICAL MODEL OF THE US DEMAND FOR HFCS 103
Empirical Model 103
Econometric and Statistical Issues 106
Description of Data 110
Estimation Results 111
Model Implications 113
Intercept Adjustment Over Time (gt) 114
Own-Price Elasticity (Pit) 116
Cross-Price Elasticity (P2O 118
6 SUMMARY, POLICY RECOMMENDATIONS AND SUGGESTIONS FOR
FURTHER RESEARCH 121
Summary and Conclusions 121
Theoretical Findings 125
Empirical Findings 128
Policy Implications and Recommendations 131
Limitations of the Study and Suggestions for Further Research 135
vi


6
economic arrangement with any developed countries which would result in such
countries receiving terms and conditions more favorable than those given to the EU
(Group of Experts, 1998). As indicated earlier, more than three-quarters of the
subregion's exports of sugar goes to the EU market where producers receive a high
enough price to cover the relatively high cost of production. Becoming part of the FTAA
could therefore mean that CARICOM might either have to sacrifice its preferential
market in the EU or at best offer the EU similar terms and conditions as those given to
the US and Canada, the developed members of the proposed FTAA. Resolving this issue
within the framework of an FTAA might not be as simple as it seems in view of the fact
that the US took a similar stance, within the framework of its CBI initiative, to that taken
by the EU in their Lome Agreement.
Second, an FTAA, which includes the subregion, would place the CARICOM
sugar industry into direct competition with the larger and more efficient producers in the
hemisphere. The study conducted by LMC, referred to above, notes that three of the
world most efficient sugar producers, Brazil, Columbia and Guatemala are in the
hemisphere. These low-cost producers could easily supply the sugar requirements of the
other members of the FTAA.
Third, is the fact that the US while being the largest sugar importing country
within the hemisphere is also a major producer of all the principal sweetenerscane
sugar, beet sugar, com sweeteners and several non-caloric sweeteners. Its sugar and
sweetener industry is reported to be the largest in the world, with annual consumption of
caloric sweeteners approaching 20 million short tons (Lord, 1995). Moreover, with the
technological advances in the com sweetener industry, the United States has become the



PAGE 1

DYNAMICS OF THE US SUGAR AND HFCS MARKETS: IMPLICATIONS FOR THE CARICOM SUGAR PRODUCERS WITHIN THE FRAMEWORK OF THE PROPOSED FTAA By EDWARD ANTHONY EVANS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1999

PAGE 2

To Sabrina, Latoya, Akeem, Attli and ma dodo

PAGE 3

ACKNOWLEDGMENTS I would like to express my profound gratitude to Dr. Carlton Davis, chairman of my supervisory committee, for all his help and encouragement throughout the course of this study. Dr. Davis' patience, understanding, guidance, assistance and constructive criticisms given to me during the preparation of this dissertation are greatly appreciated. His desire for excellence, and steadfast refusal to lower the bar kept my shoulders to the wheel and has contributed immensely to my education and self-development. Drs. Max Langham and Ron Ward, members of my committee, deserve special mention. I have benefited a great deal from Dr. Langham's criticisms and vast experience, and even though we did not always agree, I am often reminded that the beauty of a choir comes from the fact that not everyone sings the same melody. Dr. Ward has provided useful suggestions and has welcomed me into his home on several occasions during the night. To the other members of my committee, Drs. R. Emerson, A. Schmitz and D. Denslow I am eternally grateful. Without their criticisms and useful suggestions this effort would not have been possible. Special thanks is also due to Drs C. Moss and R. Kilmer who made worthwhile contributions in the initial stages of the dissertation. I gratefully acknowledge the financial support I received from the Department of Food and Resource Economics. In this regard I must again say thanks to Dr. Davis and will always remember the words he spoke which jolted me into finally accepting his offer~"there are more dogs than bones." I would also like to acknowledge the stipend I iii

PAGE 4

received from the Caribbean Agricultural Research and Development Institute (CARDI) during my initial years. Special mention must be made of Mr. Hugh Saul whose faith in me and timely intervention made it possible for me to receive the latter financial support and take up the offer at the University of Florida. Perhaps my greatest depth of gratitude however, is owed to my wife, Sabrina, children, Latoya and Akeem, my mother and my godfather, Attli, for their encouragement and for the sacrifice they have made in favor of this effort. To them this dissertation is gratefully dedicated. Finally, I wish to express my sincere gratitude to all those who stood in my corner cheering my every move. Among them are, my dear friend Dr. Kenrick Jordan, my most ardent supporter, Dr. Pauline Lawrence, Dr. Curtis Mcintosh, Mr. Karl James, Greta and Danny Roberts. iv

PAGE 5

TABLE OF CONTENTS page ACKNOWLEDGMENTS iii LIST OF TABLES viii LIST OF FIGURES ix ABSTRACT x CHAPTERS 1 INTRODUCTION 1 Statement of Problem 3 Problematic Situation 3 Specific Problem 7 Hypotheses 10 Objectives 1 1 Methodology 12 Organization of the Study 13 2 BACKGROUND 14 Free Trade Area of the Americas (FTAA) 14 Review of World and Western Hemisphere Sugar Markets 17 World Production and Market Trends— Sugar 17 Western Hemisphere Production and Market Trends—Sugar 21 Importance of the Sugar Industry to CARICOM Sugar Producers 23 Socioeconomic Importance 23 The EU/Lome ACP Agreement 25 The US Sugar Program and Policy 27 Review of World and Western Hemisphere HFCS Markets 31 World Production and Market Trends-HFCS 33 Western Hemisphere Production and Market Trends— HFCS 34 3 LITERATURE REVIEW AND THEORETICAL BACKGROUND 39 Previous Empirical Studies 39 Sugar and HFCS Related Studies 39 v

PAGE 6

Evaluation and Critique 46 Time Series Analysis 47 Serial Correlation 48 Stationary 49 Cointegration and Error Correction Models 5 1 TimeVarying Parameters 55 The StateSpace Framework 57 The StateSpace Model 57 Kalman Filter 61 Advantages of State-Space Approach and the Kalman Filter 64 Relevance of the State-Space Model and Kalman Filter to the Current Study 66 The Concept of a Derived Demand 67 4 CONCEPTUAL MODEL OF US SUGAR AND HFCS MARKETS 71 The US Sugar and HFCS Subsectors 72 The US Sugar Subsector 73 The US HFCS Subsector 78 The Interrelationship between the US Sugar and HFCS Markets 86 Trade Implications of Changes in the US Sugar Program within an FTAA 90 Free Trade Area 90 Scenario IPartial Liberalization of the US Sugar Program 92 Trade implications 92 Welfare implications 96 Scenario IIComplete Liberalization of the US Sugar Program 98 Trade implications 98 Welfare implications 101 5 EMPIRICAL MODEL OF THE US DEMAND FOR HFCS 103 Empirical Model 103 Econometric and Statistical Issues 106 Description of Data 110 Estimation Results Ill Model Implications 113 Intercept Adjustment Over Time (m) 1 14 Own-Price Elasticity (p ]t ) 116 Cross-Price Elasticity (P2O 118 6 SUMMARY, POLICY RECOMMENDATIONS AND SUGGESTIONS FOR FURTHER RESEARCH 121 Summary and Conclusions 121 Theoretical Findings 125 Empirical Findings 128 Policy Implications and Recommendations 131 Limitations of the Study and Suggestions for Further Research 135 vi

PAGE 7

APPENDIX STATE AND SMOOTEHED ESTIMATES 138 REFERENCES 140 BIOGRAPHICAL SKETCH 147 vii

PAGE 8

LIST OF TABLES Table page Table 2. 1 . Ranking of Countries of the Western Hemisphere on Basis of Population, Land Size and GDP 16 Table 2.2. Leading World Sugar Producers and Consumers, 1997-98 19 Table 2.3. Leading World Sugar Exporters and Importers, 1997-98 20 Table 2.4. Leading Hemispheric Sugar Producers and Consumers, 1997-98 22 Table 2.5. Leading Hemispheric Sugar Exporters and Importers, 1997-98 23 Table 2.6. Socioeconomic Importance of CARICOM Sugar Industries 24 Table 2.7. CARICOM/EU Market Real Earnings and Quota Rent (1992=100) 26 Table 2.8. CARICOM/US Market Real Earnings and Quota Rent (1992=100) 28 Table 2.9. World HFCS Production, 1989-97 ('000 metric tons dry basis) 34 Table 2.10. Changes in US Per Capita Sweetener Consumption and Relative Market Shares, (1975 1997) Pounds, Dry Weight Basis 36 Table 5. 1 . Estimation of Price Equation 1 12 Table 5.2 Final Estimates Using Kalman Filter Over the Period 1977-98 1 12 viii

PAGE 9

LIST OF FIGURES Figure page Fig. 2.1. World and US refined and raw real sugar prices, 1985-98 (1992=100) 18 Fig. 2.2. Changes in US per capita consumption of selected sweeteners, 1975-97 36 Fig. 2.3. Real HFCS and refined sugar prices, 1982-98 (1992=100) 37 Fig. 4.1. Flowchart of the US HFCS and Sugar Subsectors 72 Fig. 4.2. Hypothetical US demand for crystallized sugar (CS) and liquid sugar (LS), respectively 74 Fig. 4.3. Hypothetical US aggregate demand for sugar 76 Fig. 4.4. Hypothetical US supply and demand for sugar 77 Fig. 4.5. Hypothetical US demand for HFCS 79 Fig. 4.6. Shift in hypothetical HFCS short-run marginal cost curve 82 Fig. 4.7. Hypothetical US HFCS short-run marginal cost and demand configuration 83 Fig. 4.8. Hypothetical HFCS industry average costs, reflecting external economies 86 Fig. 4.9. Hypothetical supply and demand for sugar, and hypothetical demand for HFCS, respectively 88 Fig. 4.10. Trade implications under partial liberalization of US sugar program 93 Fig. 4.1 1. Hypothetical HFCS industry long-run demand and average cost 95 Fig. 4.12. Trade implications under completely liberalized US sugar program 100 Fig. 5.1. Dynamic path of HFCS intercept over time, 1982-98 115 Fig. 5.2. Dynamic path of adjustment in HFCS own-price elasticity, 1982-98 117 Fig. 5.3. Dynamic path of adjustment of cross-price elasticity of HFCS with respect to sugar, 1982-98 119 ix

PAGE 10

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy DYNAMICS OF THE US SUGAR AND HFCS MARKETS: IMPLICATIONS FOR THE CARICOM SUGAR PRODUCERS WITHIN THE FRAMEWORK OF THE PROPOSED FTAA By Edward Anthony Evans December 1999 Chairman: Carlton G. Davis Major Department: Food and Resource Economics To cope with growing regionalism, many developing countries must choose between alternative and competing integration strategies. The Caribbean Community (CARICOM) is faced with such a dilemma. The United States and the European Union have proposed the formation of separate free trade areas (FT As) with the subregion with a tentative implementation date of 2005. The consensus emerging in CARICOM favors participation in the proposed Free Trade Area of the Americas (FTAA)--a potential grouping of 34 disparate countries in the Western Hemisphere. While the anticipated benefits of such a group are unquestionable, there will be losses at the sector and industry levels. The CARICOM sugar industry is of considerable importance to the subregion but could be adversely affected by the FTAA. First, it is possible that CARICOM could violate the Lome Agreement through FTAA membership, denying continued access to the lucrative EU sugar market. Second, CARICOM sugar producers could be forced to x

PAGE 11

compete with larger and more efficient sugar-producing FTAA member countries—such as Brazil, Colombia and Guatemala—for a share of the regional sugar market. Third, and the focus of this dissertation, a substantial portion of the regional sugar market might be eroded as a result of the successful penetration of high fructose corn syrup (HFCS) produced in the United States. A qualitative assessment of the potential economic impact of changes to the US sugar program and an expanded HFCS industry on the CARICOM sugar industry was conducted. First, a conceptual model of the US sugar and HFCS markets was developed, and potential trade and economic impacts were discussed. Second, an empirical model of the demand for HFCS in the United States was constructed. To capture the dynamics within the industry a state-space representation utilizing the Kalman filter was employed. The maturity of the HFCS industry was assessed, and implications for the various sugar concerns were drawn. The findings substantiated the view that CARICOM membership in an FTAA could have potential negative economic impact on the CARICOM sugar industry and that an FTAA could further depress, rather than increase, the world market sugar price. Finally, it was concluded that the US HFCS industry is entering a mature phase, forcing the need for corn refiners to actively seek external markets for their products. xi

PAGE 12

CHAPTER 1 INTRODUCTION The global economic environment is in a state of rapid transition. The new dispensation in both the developed and developing countries is the opening up of national and regional economies to market forces. In the developed countries, the principal forces leading to trade liberalization stemmed from: 1) the growing budgetary pressure of having to maintain protective regimes; 2) growing regionalism; and 3) multilateral negotiations such as the General Agreement on Tariff and Trade (G ATT/WTO). Developing countries, on the other hand, have tended to liberalize their trading regimes either unilaterally or as a result of following the dictates of multilateral financial and developmental institutions (Wint, 1997). Whatever the reason, the result has been a dramatic reduction in trade barriers. For developing countries, the wide spread process of trade liberalization and globalization is only one element of the new global dispensation to which they must respond. An important related element is the erosion of non-reciprocal preferential treatment under the Generalized System of Preferences (GSP) or other preferential regimes such as Lome 1 , which in many cases has been a mainstay of several of the economies. At the same time, LDCs excluded from regional blocs~as in the case for the 1 Lome Convention is a contractual non-reciprocal trading arrangement between African, Caribbean and Pacific (ACP) countries and The European Union (EU). The first Lome Convention (Lome 1) was established in 1 975 and the Convention is now in its fourth round (Lome 4). It expires in the year 2000 and is currently being re-negotiated albeit, under new philosophical circumstances. 1

PAGE 13

2 Caribbean and some of their Latin American neighbors vis-a-vis NAFTA-are likely to face losses due to trade and investment diversions, higher non-tariff barriers (NTBs) and possibly higher tariffs (Blanco, 1997). As regionalism grows, many developing countries must choose among alternative integration strategies. Accordingly, they are reassessing existing economic and trading arrangements as well as considering the prospects of forming new relations and strategic alliances so as to preserve, consolidate or establish stronger trading positions. The Caribbean Community (CARICOM) 2 ~a group of 14 countries in the Western Hemisphere— is one such alliance of developing countries. With a population of approximately 6.0 million and a total gross domestic product (GDP) at market prices of only about US $14.0 billion, these countries represent a very small market by world standards (Bryan, 1995). In addition to the Lome arrangement, these countries have benefited a great deal from other marketing arrangements such as Caribbean and Canada Agreement (CARIBCAN), Caribbean Basin Initiative (CBI), and GSP. Both the US and the EU, however, have now separately proposed forming Free Trade Areas (FT As) with CARICOM with a tentative implementation date being the year 2005 or there about. While details of such arrangements are still being refined, they will have far reaching consequences for CARICOM's relationships with Europe, the other ACP countries, the Americas and much of the developing world. Within CARICOM, these arrangements 2 Member countries of the CARICOM include: Antigua and Barbuda, The Bahamas, Barbados, Belize, Dominica, Grenada, Guyana, Jamaica, Montserrat, St. Kitts/Nevis, St. Lucia, St. Vincent and the Grenadines, Surinam e, and Trinidad and Tobago. Haiti has formalized arrangements for full membership as of August 1999 bringing the total membership to 15. However, in view of that country's recent accession, this study does not include it in the analysis.

PAGE 14

3 will change how the subregion conducts its business with the rest of the world and how it earns its foreign exchange. Faced with a set of perplexing choices, the CARICOM countries have begun to systematically weigh the benefits and costs of embarking on one of the following three strategies: 1) becoming a part of a Free Trade Area of the Americas (FTAA) with the US as the central figure; 2) seeking to form a Free Trade Area with the EU as part of the new Lome arrangement; and 3) remaining outside of the formation of any free trade area with the US or the EU. The emerging consensus favors (1). Irrespective of which strategy is chosen, there will be serious internal economic conflicts because of gains and losses at the sectoral and industrial levels. Moreover, the decisions made could have irreversible economic consequences. The CARICOM sugar industry, a major contributor to the economic and social stability of the subregion, is undoubtedly one of the more important industries that will be affected as the new economic arrangements emerge. Within the framework of the proposed FTAA, for example, the industry is likely to be faced with the additional economic challenges associated with significant structural changes that are occurring within the US--one of its major sugar markets. In particular, this challenge could come as a result of the rapid development of the US High Fructose Corn Syrup (HFCS) Industry and its expansion in a much freer trading environment. Statement of Problem Problematic Situation The CARICOM countries are characterized as having certain features which, when taken in combination, makes them extremely vulnerable. These features include

PAGE 15

4 having: 1) small economies with a relatively high degree of openness as reflected in trade-to-output ratios; 2) undiversified sectoral and general economic structures; 3) export concentrations in one or two products; 4) small firms with limited opportunity to take advantage of economies of scale; 5) a high degree of dependency on non-reciprocal trading arrangements for historical reasons; and 6) a limited-resource based sugar industry with technologies heavily oriented toward labor intensity and high cost productions structures (West Indian Commission, 1993). As indicated earlier, these countries have benefited appreciably from the preferences they enjoyed in various marketing and trading arrangements such as the EU/ACP Sugar Program and the US Sugar Program. A consequence has been a degree of complacency and technical obsolescence on the part of regional producers that have led in part to an appreciable reduction in the levels of production and marketing efficiencies 3 . A recent study conducted by LMC International and reported by the Economic Research Service (ERS) of the United States Department of Agriculture (USD A) indicated that none of the CARICOM countries can be regarded as low cost sugar producers. Barbados and Trinidad and Tobago were among the highest cost producers in the world. The remaining four CARICOM producers (Belize, Guyana, Jamaica and St. Kitts/Nevis) were classified as medium cost producers (Haley, 1998a). Production cost in Jamaica is reported to be in the vicinity of 33 cents per pound, compared to the US and Brazil prices of 22 and 14 cents, respectively (Caribbean Update, 1998; USDA, 1998). To complicate 3 This point has been conceded by representatives of the Sugar Association of the Caribbean (SAC>the official producer association. They pointed out however, that while current initiatives are being taken to modernize the industry and improve both field and factory levels efficiencies, efforts to attract much needed capital investments are being hampered by the uncertainty which surrounds the future of the subregion's industry. They have also indicated that levels of efficiencies that can be obtained will not match some of those reported in parts of the US due to their economies of scale (James, 1999 Personal Interview).

PAGE 16

5 matters further, a considerable amount of the land used in the subregion to produce sugar cane is marginal with respect to sugar cane production but has few viable alternatives. Also, from a wider socioeconomic point of view, labor and capital could become socially disruptive in these small undiversified economies if other alternatives were pursued too quickly (Haraksingh, 1997). Notwithstanding these drawbacks, the industry remains of considerable import to the subregion, provides much of the valuable foreign exchange earned, and is a major contributor to the employment of labor. For example, the industry provides over one half of the foreign exchange earnings in St. Kitts and Nevis and employs about 97 thousand persons or about 5 % of the active working population within CARJCOM (James, 1998). Moreover, the socioeconomic importance of sugar production to these countries goes much further than the above figures might suggest as many of these industries do play a vital role in the maintenance and upgrading of rural infrastructure, health, recreational and educational facilities. With the general consensus that CARICOM should seek to become part of the proposed FTAA, there has been a genuine concern with regards to the future of its sugar industry (Group of Experts, 1998). Such a move toward integrating the economies brings with it the challenge of competing in a liberalized trading environment without the benefit of preferences to which the subregion has grown accustomed. The following three interrelated issues with respect to the subregional sugar industry are of importance. First, there is the possibility that should the subregion pursue such a path it could be in violation of the terms and conditions of its Lome Agreement. Specifically, article 174 section 2a of the Agreement forbid recipients to enter into any

PAGE 17

economic arrangement with any developed countries which would result in such countries receiving terms and conditions more favorable than those given to the EU (Group of Experts, 1998). As indicated earlier, more than three-quarters of the subregion's exports of sugar goes to the EU market where producers receive a high enough price to cover the relatively high cost of production. Becoming part of the FTAA could therefore mean that CARICOM might either have to sacrifice its preferential market in the EU or at best offer the EU similar terms and conditions as those given to the US and Canada, the developed members of the proposed FTAA. Resolving this issue within the framework of an FTAA might not be as simple as it seems in view of the fact that the US took a similar stance, within the framework of its CBI initiative, to that taken by the EU in their Lome Agreement. Second, an FTAA, which includes the subregion, would place the CARICOM sugar industry into direct competition with the larger and more efficient producers in the hemisphere. The study conducted by LMC, referred to above, notes that three of the world most efficient sugar producers, Brazil, Columbia and Guatemala are in the hemisphere. These low-cost producers could easily supply the sugar requirements of the other members of the FTAA. Third, is the fact that the US while being the largest sugar importing country within the hemisphere is also a major producer of all the principal sweeteners—cane sugar, beet sugar, corn sweeteners and several non-caloric sweeteners. Its sugar and sweetener industry is reported to be the largest in the world, with annual consumption of caloric sweeteners approaching 20 million short tons (Lord, 1995). Moreover, with the technological advances in the corn sweetener industry, the United States has become the

PAGE 18

world's lowest-cost producer of sweeteners. The world average cost of production is 20.91 cents per pound. It is 13.73 cents per pound in the most efficient sugar producing countries. In comparison the US produces HFCS in 1994/95 at an average cost of only 10.60 cents per pound dry weight equivalence (Haley, 1998a). HFCS competes directly with sugar in most industrial uses. In a situation of no foreign tariffs, duties, taxes, quotas, and other border barriers against corn sweetener imports, the US could export its technology and/or corn sweetener product to some of the higher-cost markets within the FTAA. The US would also benefit as the world's largest and cheapest producer of corn-the main input used in the production of HFCS. The potential of HFCS to displace sugar has already been seen in the US and to a lesser degree in Canada. And, though it has not yet adversely affected the other sugar producing and consuming countries, those in the planned FTAA are likely to feel its impact. Specific Problem This study is concerned with the threat that an expanded HFCS industry could pose to the CARICOM sugar industry within the framework of a liberalized trading environment. The US sugar industry and its sugar program have been adequately dealt with in the literature. The same is not true in the case of the HFCS industry. Not much has been written on the likely impact that an expanded HFCS industry could have on world and regional prices, production, consumption and trade patterns of sugar. Also, the dynamics of the US sugar-HFCS interrelationships and the implications of such relationships for trade in an extended market have not been fully explored. Thirdly, as noted by Tanyeri-Abur (1990) there appears to be an absence of studies to form a basis for studying the trade implications of an expanded HFCS industry. As a consequence,

PAGE 19

8 studies of production and consumption of the commodity have been largely confined to the domestic market. With the advancements in HFCS technology on both the supply and demand sides and the reduction in trading barriers in the international arena, the prospects of this commodity assuming a greater level of importance in trade have increased considerably. One aspect of the HFCS market is the dual nature of the product, whereby within a given range of the sweetener demand curve it substitutes almost perfectly for sugar but within another range it does not. Either of the sweeteners (HFCS or sugar) can be used in soft drinks, but for some other uses that require a crystallized sweetener, HFCS is not a viable substitute. This dual nature of the substitutability relationship has implications for the extent to which the commodity will be traded internationally, the shape of the sweetener demand curve and the econometric approach to be used to estimate HFCS demand parameters. Another related aspect is the issue of the extent to which the industry has reached a level of maturity. Thomas (1985), pointed out that HFCS when considered by industrial standard is regarded as a relatively new product. Among other things, he noted that this implies that there will almost certainly be further technological improvements and that the behavior of the industry can be expected to be different from that of a mature industry such as sugar. If this is indeed the case, then there are implications with regard to the econometric approach used to estimate the important economic parameters. A cursory investigation into the history of the industry seems to support the point made by Thomas. On both the supply and demand sides of the market advances in technology have resulted in noticeable development of the product and product usage. For example, the price of

PAGE 20

HFCS since its inception has been lower than that of sugar on a sweetness equivalency basis. Nevertheless, the adoption of this product had been cautious. It took a decade for the major soft drink industries to convert from liquid sugar to HFCS. On the demand side the adoption of the product has been aided by the producers of HFCS working in concert with the users, providing technical assistance and adapting the product to the special requirements of the various clients. It is in this context that HFCS-55 was developed as a more effective substitute for liquid sugar than HFCS-42 in the soft drink industry. Likewise on the supply side there has been a constant effort aimed at improving the technology involved in the production of the commodity, with a view to producing an effective substitute sweetener at the lowest cost. The implication is that the industry is evolving and as such the parameters of the industry might be changing. This timevarying issue must be recognized in characterizing the industry in any analytical model. Ward and Tilley (1980) pointed out that in markets, such as the one we are concerned with where significant adjustments are occurring, the parameters are likely to be timevarying. They noted that technological changes result in structural changes in economic phenomenon, giving rise to varying parameters. They cautioned that a failure to take into consideration such variations in the estimated parameters could cause serious errors in projections and policies drawn from them. This issue of the degree to which the HFCS industry in the US can be considered mature is therefore important from an estimation point of view. It is also important in the context of this study since the dynamic paths of adjustments of various economic parameters have such implications for the CARICOM sugar industry such as: 1) the extent of saturation of HFCS in the US market and by extension the size of the US

PAGE 21

10 domestic sugar requirement; and 2) likely rent-seeking behavior on the part of the corn refiners. From the above it is clear that any assessment of the likely impact of an FTAA on the CARICOM sugar industry must include a thorough understanding of developments occurring within the US sweeteners industry. To focus only on sugar per se would be to overlook an important component of the dynamics occurring within the region's sweeteners market. Understanding the dynamics between HFCS and sugar and generating reliable parameters for the HFCS industry, are required for a meaningful assessment of the manner in which the HFCS industry might respond in a freer trading environment. Hypotheses 1 . The parameters of the supply and demand configuration of the HFCS industry have been varying over time because of the relative "newness" of the HFCS industry and the influence of both scale and technological factors. 2. The market for HFCS is approaching an asymptotic level, giving rise to the need to seek external markets for the product, and a basis for a new round of rent seeking by corn refiners. 3. If the FTAA becomes a reality, the US could become a major exporter of HFCS in the near future with a sizable proportion of the product going to the countries in the regional trading blocs. 4. If HFCS remains competitive and can be substituted for sucrose causing a structural shift in the demand and supply of sugar, there will be an excess supply of sugar in the FTAA resulting in depressed regional and world prices.

PAGE 22

11 5. Given the above, membership of CARICOM countries in the FTAA will in the short to medium term have potentially negative economic impacts on the sugar industry of the subregion. Objectives The study has two general objectives. Firstly, to qualitatively assess the potential economic impact that likely changes to the US sugar program, and an expanded High Fructose Corn Syrup (HFCS) industry, might have on the CARICOM sugar industry within the framework of the proposed Free Trade Area of the Americas (FTAA). Secondly, to provide quantitative estimates for the HFCS demand configuration that can be used in subsequent modeling exercises. Specific objectives are to: 1 . Assess selected trends in the US sweeteners industry. 2. Develop a conceptual model of the US sugar and HFCS markets. 3. Qualitatively assess the trade and welfare implications of likely changes in the US sugar program and an expanded HFCS industry on the sugar producing countries of the FTAA, with specific reference to the CARICOM sugar producers. 4. Develop and estimate the derived demand for HFCS in the US sweeteners market. 5. Investigate the demand side dynamics within the HFCS industry, and 6. Identify critical policies regional planners might wish to take into consideration in their ongoing integration negotiations, as they relate to the future of the CARICOM sugar industry.

PAGE 23

12 Methodology The purpose of this study is to assess the potential economic impact that likely changes to the US sugar program, and an expanded High Fructose Cora Syrup (HFCS) industry, might have on the CARICOM sugar industry within the framework of the proposed Free Trade Area of the Americas (FTAA). This will be accomplished in two stages. First, a conceptual model of the US sugar and HFCS markets will be constructed. Although constructed within a static framework, the model will attempt to highlight some of the dynamics between sugar and HFCS. Based on the model, the implications of possible changes in the US sugar program, and the impact of an expanded HFCS industry on non-US sugar producers will be discussed. The second stage of the exercise will be to empirically model the demand for HFCS in the US with a view of assessing the maturity of the industry, and drawing implications for the various sugar concerns. The focus will be on the dynamic paths of adjustments of various parameters of the estimated demand function. To facilitate such an investigation a state-space modeling approach and the Kalman filter will be employed. Among the advantages of utilizing the state-space form and the Kalman filter are that: 1) it allows for modeling of timevarying parameters, if the situation so warrants, and can provide an estimate at any point within the sample and hence a valuable way of examining how a trend might have evolved; 2) it can deal fairly easily with nonstationary time series which traditional econometric methods have difficulty in handling; and 3) it allows modeling to be done in level form (rather than differenced) in situation where the data is nonstationary thus facilitating interpretation of results. The flexibility of the statespace model thus makes it an appropriate tool for estimation in situations, such as the

PAGE 24

13 demand for HFCS, in which the phenomenon being considered might be nonstationary and still evolving. Organization of the Study This dissertation consists of six chapters. Following this introduction, chapter 2 provides background information on the trends and developments and the institutional settings of the sugar and HFCS markets. It therefore places the study within the wider context, highlighting the complex nature of the issue on hand and the varied interests. Chapter 3 begins with a review of the literature, focusing on studies that have attempted to include the HFCS industry in their modeling exercises. This is followed by a review of the theoretical background to some of the issues, which are considered and used in the subsequent chapters. In particular, a detailed overview of the state-space model and Kalman filter is provided. In Chapter 4 a conceptual model is constructed of the US sugar and HFCS subsector. The model is then used to examine a few scenarios within the framework of the proposed FTAA. Implications of the results are drawn for the non-US sugar producers, in particular the CARICOM producers. Chapter 5 takes the analysis a step further and presents the two-stage estimation of the derived demand for HFCS in the US. The first section of the chapter focuses on the specification of the model while the latter section presents and discusses the implications of the results. The final chapter summarizes the conclusions from the research and makes suggestions with regard to some of the critical policies that political and regional planners might wish to take into consideration in their ongoing integration negotiations as they relate to the future of the CARICOM sugar industry.

PAGE 25

CHAPTER 2 BACKGROUND The purpose of this chapter is to provide background information on key players and institutional arrangements involved with sugar and HFCS. The chapter contains a brief description of the proposal to form a Free Trade Area of the Americas and highlights some of the disparities among the potential members. This is followed by a review of world and Western Hemispheric trends in the production and marketing of sugar. Next is a discussion of the importance of the sugar industry to CARICOM. Key marketing arrangements such as the EU/ACP Sugar Protocol and US Sugar Program are examined. Implications of the GATT/WTO rules are also briefly examined as they relate to these special marketing arrangements. The final section of the chapter focuses on the HFCS and reviews world and Western Hemisphere production and marketing trends for this commodity. Free Trade Area of the Americas (FTAA) A Free Trade Area of the Americas (FTAA) is one of several trade agreements that the United States is encouraging and actively pursuing. The decision to establish an FTAA and to conclude the negotiation for its implementation by the year 2005 was made at the Summit of the Americas held in Miami, Florida in December 1994. Among other things, the Plan of Action calls for the promotion of prosperity among the 34 participating countries through economic integration and free trade, the eradication of poverty and discrimination, and the guaranteeing of sustainable development by the 14

PAGE 26

15 conservation of natural environment for future generations. The process has five elements: 1) liberalization of barriers to trade in goods and services; 2) elimination of restraints on investments; 3) provisions of free labor movement for specialized workers; 4) harmonization of tax and monetary policies; and 5) establishment of supra regional institutions to administer the arrangements and to engage in dispute resolution. Bryan (1995) notes that the new integration process goes beyond the traditional concept of countries simply extending reciprocal preferences to their trading partners. He cautions, however, that while absolute reciprocity may be the ultimate goal of the hemispheric trade liberalization, many of the smaller Caribbean economies simply cannot compete with countries having more productive infrastructures and technologies. Stated differently, it might not be possible for them to offer absolute reciprocity to industrial countries at least in the short run, and there will be the need for some compromise on the part of the more developed trade partners (Bryan, 1995). To better appreciate some of the disparities among the countries, Table 2. 1 shows a relative ranking of the 34 countries that will comprise the FTAA with respect to the selected indicators of population, land area and GDP. In general, the CARICOM countries (underlined in Table) occupy the lower levels in all three cases. The subregion accounts for only 0.67% of the Hemisphere's land area, 0.82% of its population and approximately 0.2% of total GDP. The Table also reveals that the most populous and highest income country is the United States and the least populous and lowest income country is St. Kitts and Nevis. Barbados has the smallest land area while Canada has the largest. What is not obvious from the Table however, is the fact that many

PAGE 27

16 Table 2.1. Ranking of Countries of the Western Hemisphere on Basis of Population, Land Size and GDP Rank Population 1995 (Millio ns) Land Size ('000 sa Km.) GDP 1995 CUSS M \ V-/ V k^J 1TI 1990=100^ 1 A St. Kitts & Nevis* 0.04 Barbados A "> 0.3 St. Kitts & Nevis 150 33 Antigua & Barbuda 0.06 Grenada 0.3 Dominica 178 32 Dominica 0.07 St. Kitts & Nevis 0.3 St. Vin. and Gren. 218 j l Grenada c\ no. Antigua & Barbuda A A 0.4 Urenada 229 30 St. Vin. and Gren. All 0.1 1 St. Vin. and Gren. 0.4 Suriname 334 bt. Lucia A 1 A 0.14 St. Lucia 0.6 Antigua & Barbuda 366 Belize* u.zi Dominica A T 0. / Belize 491 97 z / Dal UaUUi>*' ft 9A u.zo i rinicidu oc i ouaso «. fl oi. Lucia CAA 509 9f> Rq n QtYl Q c oallalllao ft 97 u.z / JdlilalCd 1 1 fl 1 1 .U ouvana £AO 602 ZJ ^ i inn otYi £i ounnanic U.'IZ Bahamas i A fl 14. U Haiti 1,642 AH 1 w 1 11 14 t t"l -if VJUyaila* fl 87 U.oZ El Salvador Z 1 XI Barbados 1,717 Zj Trinidad & Tobago* 1.Z0 Belize T2 A Nicaragua 2,590 99 ZZ JalllaHa* 7 SI Z. jj Haiti nam 7fi ft Zo.U Bahamas O ACO 3,053 91 z l Mom om a A r allalila* 7 AT Z.OJ uom. Kepuuiic 40 fl 4y.u Honduras 3,378 zu Uruguay* j. iy Costa Rica CIA 51.0 Jamaica 4,171 1 Q Costa Rica* "5 AT J.U/ Panama 76.0 1 rinidad & Tobago 5,707 1 o 18 Nicaragua* 4.54 Guatemala 109.0 Bolivia 6,496 1 / Paraguay* 4.83 Honduras 1 12.0 Panama 6,570 1 a 1 0 fc,l oalvaaor* 5.04 Nicaragua 130.0 El Salvador 6,674 1 J Honduras* C AC Suriname 163.0 Costa Rica 7,027 1 A Maiti* /.Is Uruguay 1 11 A 177.0 Paraguay 7,177 1 1 uom. Kepuuiic* /.y 1 Guyana T 1 C A Zlj.O Dom. Republic 7,341 1 Z Bolivia* o.uo Ecuador lOi A 284.0 Guatemala 9,706 i i 1 1 Guatemala* 1 A /CO Paraguay A f\1 A 407.0 Uruguay 11,431 1 fl Ecuador* 1 1 At* 1 1 .40 Chile 1C1 A 757.0 Ecuador 15,132 9 Chile 14.20 Venezuela 912.0 Peru 47 <\ \ a H /,0 15 8 Venezuela 21.64 Bolivia 1,099.0 Chile 48,326 7 Peru* 23.53 Colombia 1,139.0 Colombia 56,379 6 Canada*t 29.61 Peru 1,285.0 Venezuela 64,980 5 Argentina*t 34.59 Mexico 1,958.0 Argentina 196,949 4 Colombia* 35.10 Argentina 2,767.0 Mexico 250,936 3 Mexico*t 94.78 Brazil 8,512.0 Brazil 432,433 2 Brazil* 159.22 United States 9,809.0 Canada 608,658 1 United States*t 263.06 Canada 9,976.0 United States 6,173,900 * Major sugar producing and exporting countries t Major HFCS producers Source: Adapted from Bernal (1998) of these CARICOM countries are relatively small, open and undiversified economies with over one-quarter of their total exports concentrated on one or two products. In the case of countries such as St. Kitts and Nevis, Dominica and St. Vincent and the

PAGE 28

17 Grenadines and St. Lucia, one primary exported commodity, either raw sugar or bananas, accounts for more than 50% of exports. Review of World and Western Hemisphere Sugar Markets Sugar 4 is one of the few commodities that can be produced from crops grown in both temperate and tropical climates. A consequence of this is that widespread governmental interventions have long been a feature of the world sugar market, making it one of the most volatile of all primary commodity markets (Borrell and Duncan, 1993). To gain some independence from the volatile world market, producers in most countries have lobbied their governments to operate intervention schemes aimed at controlling domestic prices, supply, and demand. Numerous pricing tiers, and trade and stockholding mechanisms (policy wedges) have been devised and have resulted in distorted production, trade, consumption, and world price. The trends in real world and US market prices of raw and refined sugars are shown in Figure 2.1 for the period 1985-98. Among other things, Figure 2.1 reveals the relative volatility of prices in the world market compared with those in the US market. World Production and Market TrendsSugar Global sugar output in 1998-99 is projected by the US Department of Agriculture (USDA) to increase by about 1.14 million metric tons (MMT) (0.9 percent) to 126.5 MMT, whereas consumption is projected to increase by 0.4 percent to 127.5 MMT, reflecting some degree of market tightening. The five leading world sugar producers and 4 In this study, the word sugar refers to sucrose obtained from sugarcane and sugar beets.

PAGE 29

18 consumers in 1997-98 are shown in Table 2.2. The European Union, followed by Brazil, continues to dominate production. The United States is listed as the fifth largest sugar producer, reflecting growth in domestic sugar beet production. In terms of consumption, the leading country is India, followed by the European Union and China. The United States occupies the fourth position in spite of its massive consumption of HFCS. As seen from the Table, both production and consumption are relatively concentrated, with 4 I I I I I I I I I I I I I I + £ £ N # f £ £ £ N # £ £ N # £ £ Years Fig. 2.1. World and US refined and raw real sugar prices, 1985-98 (1992=100) Source: Compiled with USDA sugar data and GDP deflator the top five producing countries (out of 120 sugar-producing countries) and top five consuming countries (out of all countries) accounting for 53 percent and 45 percent of global production and consumption, respectively. Finally, it should be observed that developing countries occupy three of the top five positions in both cases, reinforcing the importance of this commodity to these economies.

PAGE 30

19 Table 2.2. Leading World Sugar Producers and Consumers, 1997-98 Producers Consumers Conntrv Ouantitv % of World Country Quantity % of World (MMT) Production (MMT) Consumption European Union 19.3 15.4 India 16.7 13.1 Brazil 16.6 13.2 European Union 14.4 11.3 India 14.5 11.6 China 9.0 7.0 China 8.6 6.8 United States 8.9 7.0 United States 7.3 5.8 Brazil 8.8 6.9 World 125.4 100.0 World 126.9 100.0 Source: USDA, Sugar and Sweeteners Situation and Outlook Yearbook. Only a relatively small proportion of the large world production of sugar is traded each year. Schmitz (1995, p 54) notes that it is common to divide the market for sugar into three distinct markets: 1) the domestic market within the sugar-producing countries. This market, he contends, is the largest and accounts for about 75% of all world sugar production; 2) the market for the various international agreements between certain importers and certain exporters. These agreements include the import quotas under the US program, the bilateral agreements between Cuba and the former Soviet Republics and the agreements of various countries and groups of countries such as the ACP with the EU. This market makes up about 10% of world production; and 3) the market for the residual "free market" in world sugar, which accounts for the remaining 15% of world production. For the fiscal year 1997-98, approximately 35.6 MMT (28.3 percent of world production) of sugar was traded globally (Table 2.3). The five most important sugar exporters-Brazil, European Union, Australia, Thailand, and Cuba-accounted for 65.1 percent of global exports. Imports, on the other hand, have been less concentrated, with the share of the top five sugarimporting countries and regions in 1 997-98-Russia, United States, European Union, Japan, and Korea— amounting to only 30.1 percent.

PAGE 31

20 Table 2.3. Leading World Sugar Exporters and Importers, 1997-98 Exporters Importers Country Quantity % of World Country Quantity % of World (MMT) Exports (MMT) Imports Brazil 7.2 20.2 Russia 3.8 10.7 European Union 6.2 17.4 United States 2.0 5.6 Australia 4.6 12.9 European Union 1.8 5.1 Thailand 2.9 8.1 Japan 1.6 4.5 Cuba 2.3 6.5 Korea 1.5 4.2 World 35.6 100.0 World 35.6 100.0 Source: USDA, Sugar and Sweeteners Situation and Outlook Yearbook. Owing to domestic government policies, significant changes have occurred in the status of many of the major players in the market. For example, the European Union has changed its status from being a net sugar importer through 1976 to becoming the world's largest exporter. In the United States, sugar imports have declined considerably from a share of 20 percent of world imports in the 1970s to the current 5.6 percent. As noted by Hannah and Spence (1997), the United Stateswhich until 1975 had been the largest sugar importer by a wide margin-had slipped to second position during the 1975-80 period and to fifth position by 1994. Japan-once the world's second largest importer of sugar-has reduced its imports considerably, owing mainly to domestic support policy. Table 2.3 however, does not reflect the significant change that has occurred in the structure of the global sugar import market. The market was previously dominated by developed countries but is now dominated by developing countries. Indeed, Hannah and Spence (1997) reported that in 1994 the developing countries accounted for approximately 68 percent of net imports, compared with only 25% in 1975. These authors noted that this change in the structure of global import demand has implications for the extent of the price volatility since, unlike the developed countries, these countries

PAGE 32

21 have relatively low income and high price elasticities. Consequently, the change has resulted in greater stability of world market prices. Western Hemisphere Production and Market Trends— Sugar Of the thirty-four countries comprising the FTAA, twenty-five are major sugar producers/exporters. Thus in the 97/98 production year the Hemisphere produced about 39.61 MMT or about 31.6% of world production, estimated at about 125.38 MMT. Consumption of sugar within the Hemisphere was estimated at 31.38 MMT, which represented 24.7% of world consumption. This implies that the region 5 is a net surplus producer of sugar of approximately 8.23 MMT. If production of sugar in Cuba 6 were to be added, output would increase by another 3.0 MMT bringing the regional surplus to 10.54 MMT. Table 2.4 shows the top six sugar producing countries plus CARICOM and the top seven consuming countries in the region together with their respective market shares. On the production side the largest producer is Brazil with a share of 39.6%, followed by the US and Mexico with shares of 1 8.4% and 13.9%, respectively. Total CARICOM production, comprising output from Barbados, Belize, Guyana, Jamaica, St., Kitts and Nevis, and Trinidad and Tobago, amounted to less than one million metric ton in the 97/98 fiscal year 7 . This amount represents about 1.9% of regional output and less than 1% of world output. 5 In the document the word "region" will be used to refer to the Western Hemisphere, and the "subregion" will be used to refer to the CARICOM countries, unless otherwise define. 6 Cuba is excluded from the prospective countries slated to form the FTAA. 7 These countries are referred to as the CARICOM sugar quota holders and comprise the Sugar Association of the Caribbean (SAC).

PAGE 33

22 On the consumption side the US and Brazil were the main consumers each accounting for about 28% of regional consumption. Next in line were Mexico, Argentina, Columbia and Canada, respectively. The fourteen CARICOM countries consumed 0.32 MMT or about 1% of regional consumption. Table 2.4. Leading Hemispheric Sugar Producers and Consumers, 1997-98 Producers Consumers Country Quantity %of Country Quantity %of (MMT) Hemispheric (MMT) Hemispheric Production Consumption Brazil 15.70 39.6 US 8.91 28.4 US 7.28 18.4 Brazil 8.80 28.0 Mexico 5.49 13.9 Mexico 4.24 13.5 Columbia 2.15 5.4 Argentina 1.45 4.6 Guatemala 1.79 4.5 Colombia 1.37 4.4 Argentina 1.75 4.4 Canada 1.27 4.0 CARICOM 8 0.75 1.9 Peru 0.91 2.9 a] Represents the total output of six CARICOM countries Source: International Sugar Organization With respect to the hemispheric trade, in the 97/98 period total exports amounted to 12.77 MMT representing about 35.9% of world exports, estimated at 35.59 MMT. In contrast, total imports amounted to just 4.46 MMT or about 12.5% of world imports. Consequently, the region is a net exporter of 8.3 MMT (excluding net exports from Cuba of 2.3 MMT). Table 2.5 shows the leading sugar exporters and importers in the region. The Table also shows exports as a percentage of domestic production and imports as a percentage of consumption. The leading sugar exporters within the region are Brazil, Guatemala, Mexico and Columbia, respectively. Of interest however, is the fact that the CARICOM exports, while accounting for only 5. 1% of total regional exports, represent more than 85% of their domestic production~an indication of the importance of this commodity in trade for these countries.

PAGE 34

23 Table 2.5. Leading Hemispheric Sugar Exporters and Importers, 1997-98 Exporters Importers Country Share (%) Exports as a % Country Share (%) Imports as a % of Hemispheric of Domestic of Hemispheric of Domestic Exports 3 Production Imports Consumption Brazil 56.4 45.9 US 44.1 22.1 Guatemala 10.7 76.0 Canada 27.3 95.7 Mexico 10.0 23.3 Peru 5.8 28.2 Columbia 6.5 38.5 Venezuela 5.4 30.8 SAC b 5.1 86.5 Chile 4.9 32.4 Dominican Rep. 2.3 51.2 Ecuador 2.9 32.5 El Salvador 1.6 45.4 Uruguay 2.1 86.4 a] Include re-exports b] Total exports from six CARICOM countries Source: International Sugar Organization In terms of regional imports, the United States and Canada dominated the trade in 97/98. Together these two countries accounted for in excess of 70% of the region's sugar imports. Peru and Venezuela were the other major importers, accounting for about 1 1 .2% of the import market. The fourteen CARICOM countries import a total of about 0. 1 million metric ton. This amount represents about 2.2% of the overall regional imports and less than a third of the total CARICOM sugar requirements. The bulk of such imports was from the world market and in some cases was done in order that the exporting country would be able to satisfy its sugar export quota commitments. Importance of the Sugar Industry to CARICOM Sugar Producers Socioeconomic Importance The sugar industry is of considerable importance to CARICOM countries. This is particularly true for the six members that comprise SAC. Among other things, the industry contributes substantially to valuable foreign exchange earned by these countries. For example, in St. Kitts and Nevis the industry provides more than one-half of that

PAGE 35

24 country's foreign exchange earnings. In Jamaica, sales from exports of sugar were in excess of US $100 million in 1996 (Caribbean Update, 1998). In the subregion, the industry directly employs about 97,000 persons, or about 5 percent of the active working population, and contributes appreciably to the Gross Domestic Product (GDP) of the respective countries (Table 2.6). Moreover, the socioeconomic importance of sugar production to these countries is much more than these figures might suggest since these industries play a vital role in the maintenance and upgrading of rural infrastructure, health, recreational, and educational facilities (James, 1998). Table 2.6. Socioeconomic Importance of CARICOM Sugar Industries Country Revenue Revenue Foreign Earnings % of Active Hectares as % of GDP As % of Agric. as % of Total Work Force Cultivated Production Barbados 1.7 33.2 17.0 3.6 11,700 Belize 10.5 61.1 32.8 45.3 24,000 Guyana 26.0 41.0 27.7 9.6 42,000 Jamaica 2.6 29.7 8.0 4.1 45,000 St. Kitts/Nevis 49.0 85.0 64.1 12.5 4,000 Trinidad & 1.4 67.0 1.6 3.8 19,000 Tobago Source: James 1998 Historically, sugar in the Caribbean has been marketed under preferential trading arrangements. Currently, these arrangements are the Lome Agreement (via the Sugar Protocol) and the US Preferential Quota (via the Sugar Import Tariff Quota). In 1997-98, of total SAC countries' sugar output of 0.75 MMT (raw sugar) produced, 0.66 MMT (86.5 percent) were exported. Of this amount, 79.9 percent went to the European Union, and 1 1.3 percent went to the United States. The remaining 8.8 percent of sugar exports from CARICOM were sold on the subregional and world markets, with each accounting for an equal share. Prices obtained in both the EU and US markets were significantly higher than the world market price (James, 1 998).

PAGE 36

25 The EU/Lome ACP Agreement The EU provides special support to ACP countries, which includes CARICOM, through a series of Protocols annexed to the Lome Convention. These Protocols give free access to EU for fixed quantities of bananas, sugar, beef, veal and rum. Producers usually receive EU prices, which, by virtue of the Union's Common Agricultural Policy (CAP), are usually two to three times the world market price. Of all the Protocols, the Sugar Protocol is considered unique in that it is characterized by its "indefinite period," and its claimed "autonomy." Signatory countries benefit from a formal contractual commitment by the EU to buy specific quantities of cane sugar at guaranteed prices. The quota is fixed at about 1.3 MMT (of which 0.43 MMT is allocated to CARICOM) and ACP countries are permitted to import sugar for domestic purposes if their surplus in a year is insufficient to fill the quota. Additional preferential access was granted to ACP sugar producers on July 1995, for a six-year period, under the Special Preferences Sugar (SPS) quota bringing the total allocation to the CARICOM subregion to 0.53 MMT (James, 1998). Table 2.7 gives an indication of the real earnings and the quota rent 8 from sale of sugar to the EU market. The values are calculated on the basis of only the main quota (does not include shipment under the SPS) and assumes SAC countries fulfill their quotas. Trade liberalization puts at risk the benefits of the Protocol to the signatory parties and by implication its very existence. One problem stems from the fact that the Lome Convention was ruled in 1994 by a GATT panel to be inconsistent with GATT rules by 8 Quota rent is used here loosely to refer to the extra revenue (surplus) earned by the exporting country-in excess of what the country would have earned selling at the world market price-from selling in a particular market where the price is higher than world market price because of import restrictions.

PAGE 37

26 virtue of its non-reciprocal nature (Blanco, 1997). In particular, the Arrangement was ruled to be discriminatory vis-a-vis other LDCs and thus not covered by the enabling clause permitting GSP schemes. This GATT ruling led to the EU seeking a waiver, which was subsequently granted by the WTO. However, the non-reciprocity permitted by Table 2.7. CARICOM/EU Market Real Earnings and Quota Rent (1992=100) Year SAC UK Price Real UK Real World Total Real Quota Rent QUOTA (Pounds Price raw Price raw Earnings (US $M) (MMT) /MT) (US$/MT) (US$/MT) (US $M) 90-91 0.4281 346.10 628.35 204.27 269.00 181.55 91-92 0.4281 351.12 587.79 199.98 251.64 166.03 92-93 0.4281 403.00 629.91 215.66 269.67 177.35 93-94 0.4281 400.28 587.39 252.06 251.47 143.56 94-95 0.4281 420.56 608.52 273.07 260.51 143.61 95-96 0.4281 445.58 649.58 240.39 278.09 175.18 96-97 0.4281 414.44 577.45 227.90 247.21 149.64 97-98 0.4281 364.64 512.96 177.08 219.60 143.79 Source: SAC data the GATT/WTO waiver is contrary to Lome principles and is only valid until the end of the present Convention in the year 2000. What this means in practical terms for the ACP countries and CARICOM in particular, is that the Sugar and Banana protocols will have to stand alone as either viable WTO-compatible commercial arrangements between the EU and the nations to which they apply, or they will disappear completely. The WTO 1999 negotiations will undoubtedly include strong calls for free trade in sugar and the demise of the Protocol. A second factor which stands to reduce the potency of the Sugar Protocol, irrespective of the shape or form of any future EU-ACP Agreement, is the reductions in domestic support measures by the EU as it seeks to implement its WTO liberalization commitments. Firstly, as the EU implements its agreed tariff reductions, access by third

PAGE 38

27 (non-ACP) countries, presently granted Most Favored Nation (MFN) 9 status, will be enhanced thereby reducing the preference margin for exports from ACP. Secondly, as support is reduced, particularly subsidization of exports to domestic producers in the EU, the domestic prices paid to EU producers will begin to fall. This will in turn, trigger a reduction in ACP prices since the Sugar Protocol guarantees that ACP quota holders will be paid the same prices as EU producers. Other factors which will or have begun to erode the value of the EU/ACP Sugar Protocol include: 1) the pre-EU membership arrangements with the countries of Central and Eastern Europe some of which are sugar producers; 2) a surge in EU centered regional trade arrangements, in particular, preferential agreements in the pipeline with some of the sugar producing countries in Latin America; 3) the fact that markets have evolved significantly since the first Lome Convention in 1975, with the result that Europe has become a leading sugar exporter; and 4) that cane sugar can be bought on the world market at much lower prices than those guaranteed by the Protocol which now cost the EU several hundred million Euros 10 each year. A final issue that is unresolved is the impact of the Euro, the EUs common currency, on the sugar prices that quota holders receive. The US Sugar Program and Policy In spite of the less lucrative nature of the US market relative to the EU market, the former nonetheless has provided CARICOM sugar producers with a much more viable 9 The Most Favored Nation (MFN) clause of a commercial treaty is a binding contract by the signatories to confer upon each other all the most favored trade concessions that either may grant to any other nation subsequent to the signing of the agreement.

PAGE 39

28 alternative than selling sugar on the world market. It has also permitted the subregional sugar industry to survive and, through inter-industry economic linkages, has fostered growth in services and other sectors. Table 2.8 gives a rough indication of the real quota rent earned by CARICOM as a result of the US program over the period 1990-98. The data are based on the assumption that CARICOM fulfilled its quota commitment to the US, which is approximately 4.8 percent of the total tariff rate quota (TRQ)". Also, no Table 2.8. CARICOM/US Market Real Earnings and Quota Rent (1992=100) Year TOTAL SAC Real US Real World Total Real Quota Rent TRQ TRQ Price raw Price raw Earnings (US $M) (MMT) (MMT) (US$/MT) (US$/MT) (US $M) 90-91 2.1027 0.1051 487.41 204.27 51.24 29.77 91-92 1.3863 0.0693 468.82 199.98 32.50 18.63 92-93 1.1347 0.0567 464.86 215.66 26.37 14.14 93-94 1.1347 0.0567 458.00 252.06 25.98 11.68 94-95 1.3241 0.0662 466.49 273.07 30.88 12.81 95-96 1.9702 0.0985 439.92 240.39 43.34 19.66 96-97 1.9065 0.0953 414.98 227.90 39.56 17.83 97-98 1.4545 0.0727 403.56 177.08 29.35 16.47 Source: Compiled from USDA data. adjustments have been made to accommodate shipping expenses. As can be seen, the amounts of quota rents in real terms are substantial and their loss would have serious negative social repercussions. However, as indicated in Fig. 2.1, prices have trended upward in the world market and downward in the US market during the period. An obvious implication of this observation is that the real premium enjoyed by quota holders and US producers has been declining. As will be seen later, this factor — coupled with a The Euro is currently valued at approximately 1 Euro = US $1.06. 11 The Tariff Rate Quota (TRQ) system is a key component of the current US sugar program. This component of the program is elaborated on below.

PAGE 40

29 reduction in the quota levels—implies a considerable loss of real earnings for the quota holders. A stated objective of the US Sugar Program is to ensure a reliable supply of sugar to American consumers at competitive prices, while providing some stability for American sugar interests (growers and processors). Key components of the program include: 1) loan rates; (2) an implicit market stabilization price (MSP); and 3) Tariff-Rate Quota (TRQ). These components include: 1 . Loan rates~a loan rate of US $0. 1 8 per pound of raw sugar is made available to processors. The loan rate of 22.9 cents per pound for refined beet sugar was historically set in relation to raw sugar with a prescribed formula. Loans were previously non-recourse in that, at any time, the processors could refuse to repay the loan and the government would accept the sugar in lieu of cash. However, with the implementation of the Federal Agricultural Improvement and Reform Act (FAIR) of 1996 current loan rate programs are to operate on a recourse basis under certain conditions. Whenever the tariff-rate quota (TRQ, discussed below) is 1 .36 MMT or less, the USDA can demand repayment of the loan with interest at maturity, regardless of the price of sugar. The loans are considered to be non-recourse only when the TRQ is above the stipulated 1 .36 MMT. In addition, loan rates were fixed at the 1995 levels of 18.0 and 22.9 cents per pound for raw and refined sugar, respectively. 2. Market stabilization price (MSP)--a major stipulation of the program in the past was that it should run at no cost to the government. To ensure that such provision was met, an MSP was explicitly established at a level to ensure that commercial market

PAGE 41

30 channels were cleared rather than forfeited to the Commodity Credit Corporation (CCC). The MSP was set at a level of $0.22 per pound of raw sugar. However, under the FAIR program the no net cost provision was eliminated and the MSP is no longer set explicitly although it is expected that supply would be managed in such a way as to ensure the market price would be above the loan rate. 3. Tariff-rate quota (TRQ)--the TRQ for sugar and sugar-containing products is not technically a part of the domestic sugar support legislation (Buzzanell, 1998). Nonetheless, it is one of the principal tools used to support the loan program and to ensure that the domestic price does not come within the forfeiture range. The US Trade Representative (USTR) allocates the TRQ for raw sugar to 40 countries. The total quota varies from year to year in keeping with the objective to stabilize domestic raw-sugar prices. Although the quota can vary from year to year, it cannot fall below 1.14 MMT, which is the commitment given by the United States under the Uruguay Round of the GATT (USDA Situation and Outlook Series, 1998). Of the 40 countries participating in the program, 23 are from the Western Hemisphere and include all the SAC sugar exporters. Country allocations are based on historic trade with the United States. Haiti, and St. Kitts and Nevis were included in the allocations as minimum quota-holding countries, and the allocation to Mexico was increased to fulfill obligation pursuant to North American Free Trade Agreement (NAFTA). About 66 percent of the raw cane sugar TRQ are allocated to the hemispheric countries. The Dominican Republic, Brazil, CARICOM, and Argentina account for the bulk of the overall regional quota with TRQ shares of 17 percent, 14 percent, 5 percent, and 4.6 percent, respectively. All countries in the Hemisphere, with the exception of Brazil,

PAGE 42

31 receive preferential treatment (zero-tariff) in that they are able to sell their in-quota sugar to the United States at the established US domestic price and retain the export quota rent. In the case of Brazil, a small tariff of 0.625 cent/pound is charged on the in-quota exports. Imports outside of the quota have been subject to a high tariff of 17.62 cents per pound since January 1, 1995. In accordance with GATT/WTO commitments, this tariff is expected to decline to 1 5.36 cents per pound by the year 2000. Since the quota is reserved for selected countries and thus can be considered discriminatory towards the LDC countries, there is the view that the TRQ will be seriously questioned in the 1999 WTO review round (Contrearas, 1998). In addition, there is increasing pressure within the US to eliminate the current sugar program in keeping with the Federal Agricultural Improvement and Reform (FAIR) Act of 1996, also known as the "Freedom to Farm Act." FAIR seeks to eliminate or greatly reduce government intervention into the agriculture sector as a means of making the sector more competitive. While the issue of the elimination of the US sugar program is somewhat debatable, there is the unconfirmed view that the once strong lobby, comprising the growers of cane and beet sugar and the manufacturers of caloric and non-caloric sweeteners, is losing its political clout, in part because manufacturers of HCFS no longer need the protection accorded in order to make their operations viable. Review of World and Western Hemisphere HFCS Markets High Fructose Corn Syrup is a liquid caloric sweetener made from ordinary cornstarch. It can be substituted for sugar (sucrose) in most liquid uses. As a consequence of its relative cheapness in comparison to other forms of caloric sweeteners, it has been

PAGE 43

32 used in a wide range of processed food products such as beverages, baked goods, dairy products and jams and jellies. Although converting cornstarch into sweet substances was discovered as early as 181 1, it was not until the late nineteen-century that sweeteners such as glucose corn syrup and dextrose were produced in corn wet milling. However, these products were only about 70% as sweet as sugar and hence were not competitive with sugar on a sweetness equivalency basis. The search to find a sweetener comparable to sugar did continue and although promising results were obtained in the early 1960s when scientists discovered an effective method of obtaining fructose from glucose ~a substance which is 1 10 to 170 percent sweeter than sucrose --the break through did not occur until 1967 and was based on research carried out in Japan and the US. However, commercialization of the product did not occur until 1972 12 . HFCS comes in two strengths, HFCS-42 and HFCS-55. The numbers indicate the strength of the fructose percentage. Thus, in the case of HFCS-42, the product contains 42% fructose, and about 50% and 8% of dextrose and other saccharides, respectively. This product is approximately 90% as sweet as sugar. HFCS-55 contains 55% of fructose and about 40% and 5% of dextrose and other saccharides, respectively. This product has sweetness of about 110%. Commercial production of HFCS-42 began in 1972 while that of HFCS-55 began in 1977. In 1985, through further processing of HFCS-55 a crystalline form of the product was prepared for commercial use. However, certain technical and economical blotches still hamper its manufacture and limit its widespread use as a direct substitute for crystallized (tabletop) sugar. The main problems have to do with its still relatively high cost of production compared with sugar and the fact that its sweetness 12 For details on the invention process of the product and facts about its diffusion see Zitt (1998).

PAGE 44

33 appears to vary depending upon the particular use (Polopolus and Alvarez, 1991 ; Thomas, 1985). Although both syrups share certain favorable features such as stability, high osmotic pressure or crystallization control, each offer special qualities to food manufactures and consumers. Thus HFCS-42 is popular in canned fruits, condiments and other processed foods which need mild sweetness that won't mask natural flavors. The sweeter HFCS-55 is used mainly in the soft drinks industry, in ice cream and frozen desserts. The main drawbacks to HFCS are that it is available mainly in liquid form, and therefore restricted to only certain industrial uses. In addition, due to its high water content it costs more to transport than an equivalent quantity of sugar, and it is difficult to handle, since it must be maintained at 80° -100° F temperature when stored or transported (American Sugar Alliance, 1998). World Production and Market Trends— HFCS Table 2.9 shows world HFCS production for the period 1989-1997. It shows that in 1997 world output of HFCS amounted to 10.41 MMT which is about 8.5% of the combined world output of sugar and HFCS. The US dominates production with output of 7.71 MMT or 74% of total output in 1997. Japan's production of 0.8 MMT accounted for another 7.6% of total output. Thus, together these two countries accounted for almost 82% of world output. It should be pointed out however, that while production of HFCS in the US is unrestricted, production levels in both Japan and the EU (third largest producer) are restricted.

PAGE 45

34 Owing to the problems alluded to earlier with regards to storing the commodity, very little is held in stock. Of particular interest is the fact that, between 1990 and 1997, world consumption of HFCS increased at an annual rate of 4.0% while consumption of sugar grew at a lackluster rate of 1 .5% per annum. With the demand growth for sugar failing to keep pace, HFCS has captured an increasing share of the combined sugar market for these sweeteners, increasing from 7.2% in 1990 to an estimated 8.5% in 1997 (LMC International, 1998). Ta ble 2.9. World HFCS Production, 1989-97 ('000 metric tons dry basis) Country 1989 1990 1991 1992 1993 1994 1995 1996 1997 US 5,345 5,677 5,852 6,038 6,459 6,813 7,143 7,408 7,711 Japan 744 783 794 747 727 806 789 792 800 EU 276 280 284 286 288 290 303 303 303 Korea 244 270 276 263 246 260 277 279 280 Canada 239 245 252 250 255 255 265 260 295 Argentina 146 161 184 184 195 207 190 198 210 Taiwan 51 67 110 125 150 171 182 195 210 Mexico 0 0 0 0 0 0 0 110 175 Others 156 210 197 229 258 308 354 388 423 World 7,201 7,693 7,950 8,122 8,579 9,109 9,493 9,933 10,407 Adapted from LMC International, (1998) Currently, only a small amount of HFCS output enters into trade. As indicated earlier, this is due to the technical difficulties and high costs associated with transporting the liquid sweeteners. Consequently, trade is confined mainly to cross-border transactions. In this regard, the only region where considerable quantities are traded internationally is in North America. Western Hemisphere Production and Market Trends-HFCS With respect to the Western Hemisphere, total output of HFCS in 1997 amounted to 8.43 MMT, representing 81.0% of world production. The main producing countries

PAGE 46

35 within the region are the US (91.5%), Canada (3.5%), Argentina (2.5%) and Mexico (2.1%). Production in Mexico started in 1996 by US interests and already output level is close to that of Argentina and Taiwan. At the moment, the bulk of the corn used in the production of this commodity in Mexico is imported from the US. The 1997 regional consumption of HFCS was estimated at 8.32 MMT. Again the US was the main consuming country accounting for 91.1% of total demand. This was followed by Mexico (3.7%), Argentina (2.5%) and Canada (2.5%). Only a small quantity of HFCS is currently consumed in CARICOM and none of the member states engage in the production of the commodity. The considerable success in HFCS as a substitute for sugar is best illustrated in the case of the US. Table 2.10 shows the changes in the per capita consumption and the relative market shares of caloric sweeteners consumed in the US over the period 1975 to 1997. And, Figure 2.2 shows a comparison of the trends in per capita consumption of HFCS, total corn sweeteners and sugar for the same time period. In particular, the Figure reveals that between 1975 and 1997, the per capita consumption of sugar fell from 89.2 lbs. to 67.1 lbs. after reaching a low of 60.8 lbs. in 1986, while the per capita consumption of HFCS increased from 5 to 61.4 lbs. over the same period. This meant that whereas in 1975 sugar accounted for about 76 % of the caloric market shares and HFCS only 4%, by 1997 the market share of sugar had fallen to 43% while in the case of HFCS it had risen to 40%. The rapid and considerable increase in the production and consumption of HFCS in the US and the concomitant displacement of a portion of the US sugar demand, has come largely at the expense of the sugar quota holders and by implication the US sugar refiners. This is easily seen when one considers the fall in US

PAGE 47

36 Table 2.10. Changes in US Per Capita Sweetener Consumption and Relative Market Shares, (1975 1997) Pounds, Dry Weight Basis Year Refined Relative HFCS Relative Total Corn Relative Honey & Edible Relative Total Caloric Sugar Share (%) Share (%) Sweet'rs Share (%) Syrups Share (%) Sweet'rs 1975 89.2 76 5.0 4 27.5 23 1.4 1 118.1 1976 93.4 75 7.2 6 29.7 24 1.3 1 124.4 1977 94.2 74 9.5 7 31.2 25 1.4 1 126.8 1978 91.4 72 12.1 10 33.7 27 1.5 1 126.6 1979 89.3 70 14.9 12 36.4 29 1.4 1 127.1 1980 83.6 67 19.1 15 40.2 32 1.2 1 125.0 1981 79.4 63 23.2 19 44.5 36 1.2 1 125.1 1982 73.7 60 26.7 22 48.2 39 1.3 1 123.2 1983 71.1 57 30.7 25 52.2 42 1.3 1 124.6 1984 67.6 53 36.3 29 57.8 46 1.4 1 126.8 1985 63.2 49 44.6 35 63.9 50 1.5 1 128.6 1986 60.8 48 45.1 36 64.6 51 1.6 1 127.0 1987 63.1 48 47.1 36 66.8 51 1.7 1 131.6 1988 62.6 47 48.3 36 68.3 52 1.5 132.4 1989 62.8 47 47.5 36 68.0 51 1.6 132.4 1990 64.8 47 49.2 36 70.4 51 1.6 136.8 1991 64.4 47 50.0 36 72.0 52 1.6 138.0 1992 64.4 46 51.6 37 74.3 53 1.6 140.3 1993 64.6 45 54.4 38 77.8 54 1.6 144.0 1994 65.8 45 56.4 38 80.2 54 1.5 147.5 1995 66.2 44 58.4 39 82.6 55 1.5 150.3 1996 66.9 44 59.8 39 84.2 55 1.5 152.6 1997 67.1 43 61.4 40 86.3 56 1.5 154.9 Source:USDA, ERS, Commodity Economic Division, Sugar and Sweetener: Situation and Outlook Report, various issues 100 n Years Fig. 2.2. Changes in US per capita consumption of selected sweeteners, 1975-97

PAGE 48

37 sugar imports from a level of 5.0 MMT in the seventies to current level of about 1.6 MMT (Hannah & Spence, 1997). The growth of the industry in the US is attributed to several factors including: 1) a marketing environment with no restriction on supply; 2) advancements in technology; and 3) the relative cheapness of the product in relation to sugar on a sweetness equivalency basis. Since its inception in the US the product has sold at a discount to sugar. Figure 2.3 shows a comparison of the real price of HFCS-42 and the real world and US prices of refined sugar. The evidence indicates that the pricing of HFCS is now much closer to that of the world market price of refined sugar. Fig. 2.3. Real HFCS and refined sugar prices, 1982-98 (1992=100) Source: USDA, ERS, Commodity Economic Division and Dept. of Commerce, Bureau of Economic Analysis

PAGE 49

38 As noted before, most of the international trade in HFCS occurs within North America. Although the Canadian and the US-HFCS industries are difficult to separate, from a custom point of view, Canada is a net exporter to US, which in turn is a net exporter to Mexico. The growth in this trade is reportedly due to increases in the Mexican demand (USD A, Situation and Outlook Report, 1998). However, owing to a dispute concerning the legitimacy of the US-Mexican Side Agreement on trade in sugar under the North American Free Trade Agreement (NAFTA), the Mexicans have recently (January 1998) placed a 100% tariff on imports of HFCS from the US. The issue is now being addressed by both the WTO and the NAFTA dispute settlement bodies. The tariff has retarded the growth in trade but the US has maintained current levels of exports to Mexico. Total exports of HFCS from the US increased by 37 % in 1998 and now account for as much as 6.2 % of domestic production compared with 3.7% and 4.7 % in 1996 and 1997, respectively. Finally, even with the shipping problems of the liquid product, an increasing amounts of HFCS are currently being shipped to markets such a Japan and Trinidad and Tobago (USDA Situation and Outlook Report, 1997). Industry experts believe that shipping problems are surmountable. Within CARICOM, Trinidad and Tobago is the only country, which currently imports HFCS for use in its soft drink industry, however there is the possibility that other countries might soon follow.

PAGE 50

CHAPTER 3 LITERATURE REVIEW AND THEORETICAL BACKGROUND Several studies have been done on modeling the US Sugar/Sweetener subsector and drawing economic/welfare implications for various groups including domestic sugar producers, processors, sweeteners' users, consumers and trading partners. Most of these studies have either not taken into consideration the impact of HFCS in their analysis or have only done so in a cursory manner. The first section of this chapter therefore undertakes a critical review of those studies in which the HFCS industry was taken into consideration. The review provides a basis for the conceptual model of the US sugar and HFCS subsector developed in chapter 4. Because of the problem of time-varying parameters in the HFCS industry and the need to provide more reliable parameters for the HFCS industry, the second section of the chapter focuses on some critical time series issues such as serial correlation, unit root and cointegration. The methodology of statespace model and the Kalman filter, allows for estimation of time-varying parameters and is employed later in estimating the derived demand for HFCS. Consequently, this methodology is discussed in detail. The chapter ends with a discussion on the nature of a derived demand and the general approaches used to estimates such functions. Previous Empirical Studies Sugar and HFCS Related Studies One of the earliest studies to include HFCS in the analysis of the US sweetener's market is a study conducted by Carman (1982). This study had as its objective the 39

PAGE 51

40 projection of the rate of adoption of HFCS in the US and an assessment of its economic implications for major sweetener industry participants. The study used a logistical trend model with estimates of a ceiling market share of HFCS and total demand for caloric sweeteners to carry out the projections. Among other things, the study concludes that the total quantity of sugar demanded in the United States would decrease for several years as HFCS demand increases. Also, that the impact of the HFCS on domestic sugar producers, under the restrictive sugar policy, would be minimal since the bulk of the costs would be borne by those countries exporting sugar to the US. Carman assumed that the saturation of the HFCS sweetener market would occur between 25 -30% of the total caloric sweetener demand. This assumption has proven to be too conservative since HFCS currently accounts for as much as 40% of the caloric sweeteners' market (Table 2.10). Lopez and Sepulveda (1985) developed and estimated a demand model for sugar and on the basis of the model's results derived implications for sugar import policies. In that model they addressed what they considered to be some shortcomings of earlier studies. They: 1) estimated separate demand functions for industrial and nonindustrial sectors; 2) compared the periods before and after the introduction of high fructose corn syrup in industrial uses; and 3) utilized a partial adjustment framework in an attempt to measure the speed of demand adjustment. Their empirical results indicated that: 1) changes in consumer preferences and the availability of the cheaper HFCS in the food processing segment were exerting downward pressure on sugar demand; and 2) the demand for sugar had become less responsive to sugar price change after the introduction of HFCS. They estimated that the short-run own price elasticity went from -0.15 to 0.04 and the long-run (full adjustment) own price elasticity went for -0.3 1 to -0.06 after

PAGE 52

41 the introduction of HFCS. They attributed the decreased in own price elasticity of sugar to the substitution taking place in the more price elastic segment of the sweetener market such as the soft drinks and canning industries. They concluded that as the US demand for sugar decreases and the food industry adjusts faster to sweetener choices, the US government would have to impose more restrictive import barriers to maintain prices to domestic sugar and HFCS producers. Leu, Schmitz, and Knutson (1987) used a general equilibrium model to analyze the policy options for the US Sugar program and to empirically estimate the substitutability between sugar and HFCS by including the price of HFCS as a determinant of sugar demand. They used a composite price of corn sweeteners glucose, dextrose and HFCS-in order to generate a series dating back to 1955. Demand elasticities were estimated for sugar and HFCS while supply elasticities were obtained from previous studies. Their results indicated that the net social cost of the quota program with an elasticity of excess supply of 2.37 was $203 million when substitution with HFCS was allowed and $253 million with no HFCS substitution. Under the assumption of a perfectly elastic excess supply curve the net social cost with HFCS substitution was about $1 billion. They concluded that a deficiency payment program would be more costly than an import quota. However, Marks (1993) stated that the study suffered since price data on the HFCS were only available after 1975. He also questioned the composite commodity assumption because glucose and dextrose are much less substitutable for sugar than is HFCS. Sudaryanto (1987) developed a trade model to analyze the effects of trade liberalization by the US and EEC on international markets for sugar. In relation to the

PAGE 53

42 US, he pointed out that analysts in previous sugar studies ignored the derived nature of the demand for sugar by treating all sugars as if they were consumed directly. Utilizing Fourier's flexible functional forms he separately estimated, a derived demand and direct demand for sugar. He then computed an aggregate elasticity of demand for sugar using the weighted average of elasticities obtained from the two models. While he did take into consideration the impact of the possible substitution between HFCS and sugar in estimating the various demand functions, he followed previous studies and utilized a composite price index of HFCS, glucose and dextrose. The results of his analysis indicated that in the United States, sugar for final consumption is far less elastic than sugar used as intermediate input. Sudaryanto's results suggest that consumption for final uses had reached the near saturation point and a change in sugar price had little effect on consumption. In contrast, the demand for sugar in the food processing industry was more elastic because of available substitutes. The result of his trade model showed that unilateral trade liberalization in the US market would force an increase in US imports by 114% and would be accompanied by a decrease in both production and supply price by 8.3% and 1 1.2%, respectively. Complete removal of protection in the EEC was estimated to cause a reduction in supply price and production in that region of 13.2% and 5.6%, respectively. Rendleman and Hertel (1989) examined the impact of policy changes on sugar and HFCS within a general equilibrium framework. The results of their analysis showed that any substantial increase in the production of corn sweeteners would cause an increase in the production of byproducts, which would have the effect of lowering the byproducts prices and thus raising the cost of production of corn sweeteners. They

PAGE 54

43 estimated that free trade would result in the US prices for sugarcane, sugar beet and corn sweeteners declining by 34%, 31% and 10%, respectively. Corn sweetener byproducts would decline by 2% but prices would increase by 12 %. In terms of the changes to output, they estimated that US sugarcane production would decline by 77%, sugar beet production by 47%, and corn sweetener production by 2%. This study was an improvement over many of the previous studies in that an attempt was made to seriously analyze the interrelationship among the various segments of the sweetener market. Tanyeri-Abur (1990) analyzed the implications of current and potential changes in the US sugar policy with a price endogenous agricultural sector model. The sugar and HFCS markets and the industries, which used these two sweeteners, were modeled. Several policy runs were made under varying assumptions pertaining to the level of imports, target prices and substitution relationship between sugar and HFCS. The results were compared with the based year (1996) in which imports of raw sugar was restricted to 1.6 million short tons. Her work indicated that: 1) the removal of US sugar quotas would benefit US consumers by $657 million and increase net foreign surplus by $350 million; and 2) that US producers' surplus would decline by $288 million leaving a net welfare gain of about $718 million. However, when the agricultural sector with all commodities was considered, there was a reduction in net social welfare for the sector of $1.2 billion dollars despite an increase in total consumer surplus of $1.4 billion. She explained that this result was due mainly to the increase in government payments that would result as the price of cotton and rice fall as land was shifted from sugarcane to these crops.

PAGE 55

44 She conceded that conclusions in relation to the HFCS subsector could be a reflection of rigid assumptions that were made in relation to the supply and cost of production of HFCS. For example, an unlikely result is that HFCS does not effectively compete with sugar and is replaced by sugar when free substitution is allowed. In further analysis in the said study, however, she showed that the above result was highly sensitive to the assumptions made with respect to the HFCS supply elasticities. She concluded that further analysis is needed into the behavior of HFCS, but that, by and large, the signs of the welfare change and the distributional implications would remain the same. Marks (1993) constructed a partial-equilibrium model in which he assessed the long-run static welfare effects of alternative US sugar policies. He evaluated two policies. The first focused on estimating the overall economic impact of the US sugar program for crop years (1984/85 to 1988/89)~years in which the domestic-foreign price differential varied widely. The analysis considered the long-run effect of the program on sugar prices in the US and the rest of the world (ROW), the economic welfare of US and foreign consumers and producers, and U. S. government revenues. His second policy analysis examined the implication of a 10% real reduction in the US sugar loan rate, by which the US provides support to its sugar producers. His approach to modeling the US sweetener market differs from most of the previous studies in two ways. Firstly, he estimated the total demand for refined sugar plus HFCS (US sweetener demand) as function of a composite index of their prices. Secondly, the prices of refined sugar and HFCS in turn were modeled as functions of the US raw sugar prices, which according to Marks is targeted by the policy makers. In addition, he estimated equations for aggregate sugar demand and supply for the rest of the world (ROW). From these equations he was able to

PAGE 56

45 infer a price elasticity of ROW export supply. The price elasticity is important because it gives an indication of the extent to which changes in the US imports would affect the world market price. Marks (1993) investigation reveals that in all years considered, the unilateral liberalization of the US program would result in the world market price increasing, by an amount which depends on the price differential between the US and world market price. For example, the US program's depressive effect on the world market price in the year 1984/85, when the price differential was high, was approximately 16% whereas in the year 1988/89, when world market price was relatively high, it was estimated at less than 9%. Among his other conclusions was the suggestion that the US sugar program benefited the rest of the world in the aggregate over the period by an average of $0.33 1 billion. The program depressed the world raw sugar price, and cost foreign sugar producers an annual average of $1,867 billion. However, these costs were outweighed by $2,109 billion in gains to foreign consumers. The paradox he explained, was due to the effects of the rent to foreign quota holders who profit from the price differential between the US and the rest of the world. In his analysis of a marginal reform to the US sugar program the result indicated that a 10% reduction in the real US loan rate would lead to higher welfare in the aggregate in both the United States and the rest of the world. Marks (1993) study although novel in some respects, suffered from two weaknesses. First, the approach in estimating the US sweetener demand used the historical annual market shares of refined sugar and HFCS as weights in computing the composite sweetener price index. This method assumed a kind of fixed proportion good. Second, in estimating the effects of future policy changes the erroneous assumption was

PAGE 57

46 made that the technical substitution of HFCS for liquid sugar was complete and that the existing market share would continue into the future. Evaluation and Critique A common feature of the studies reviewed here is that all utilized time series data in their analyses to estimate parameters in supply and demand equations. With the exception of the study by Lopez and Sepulvelda (1987), none considered the issue of time-varying parameters. Secondly, although some attempt was made to take into consideration the substitution effect between HFCS and sugar in the US sweeteners market, none of the studies directly included HFCS in their trade analysis. Thirdly, all studies inferred that if the US were to liberalize its sugar program unilaterally, it would have a positive (upward) impact on the existing world market price. Fourthly, all the trade-related studies were either spatial or nonspatial equilibrium models. According to Thompson (1981) although spatial equilibrium models represent one of the most popular approaches to agricultural trade modeling, especially for comparative statics analysis of the effect of a change in policy, their adequacy for this purpose is questionable. He noted that such models were not consistent with how many international agricultural markets normally operate and hence do not do well in accounting for actual trade flows. Chief among the reasons given for the real world inconsistency of the models was their faulty assumption of a homogeneous traded product, ignoring the fact that many importers differentiate among exporters on historic or political grounds. Consequently, changing from one source to another is usually much more difficult than assumed by such models. The second class of models used was the nonspatial equilibrium types in which simulation rather than optimization procedures are used to give the net trade position of

PAGE 58

47 each country. Such models are usually less efficient as a means of examining the effects of changes in policy variables, such as tariffs and transport costs. The main advantage of these models is that they are usually much easier to solve (Thompson, 1981). Both spatial and non-spatial models are based on a competitive framework. However, as shown by authors such as Krugman (1979), models with imperfect competition features might be more realistic when modeling international trade. Although it is recognized by this author that of necessity any model is an abstraction of the real world, an attempt will be made to address a few of the shortcomings alluded to. Time Series Analysis Aoki (1990) states that the main aims of time series analysis are one or a combination of the following: 1) to describe the data behavior succinctly; 2) to explain the behavior of the time series in terms of exogenous variables; and 3) to forecast and in some cases to control. Harvey (1987) makes a distinction between a pure time series model and a structural time series model. The distinguishing feature of the two is that no attempt is made in the former to formulate behavioral relationships among variables, in contrast to the latter. Rather, the movements in the variable are 'explained' solely in terms of its own past, or by its position in relation to time. In utilizing time series data to carry out analysis several issues must be confronted including a determination of whether the data are: 1) serially correlated; 2) stationary; 3) possess a deterrninistic or stochastic trend; and 4) the stability of the parameters (timevarying parameters). In cases where two or more time series are to be regressed on each other there is the added concern of cointegration and guarding against spurious regression. The following discussion provides brief insight into each of these issues.

PAGE 59

48 Serial Correlation Serial Correlation (autocorrelation) is the extent to which a time series variable (usually the error term of the time series model), lagged one or more time periods, is correlated with itself (Shim and SiegeL, 1995). When such a situation exist it has implication for the classical linear regression (OLS) model. The OLS estimator of the parameters remains unbiased but no longer efficient and in fact is asymptotically inefficient. Moreover, the standard formula used by the OLS to compute the estimate of the variance of the errors is no longer unbiased and could either overor underestimate the true variance. The implication is that all tests of reliability, using the tand Fdistribution, which rely on the estimated variance-covariance matrix of the estimated parameters, become invalid. Several tests such as the Durbin Watson (DW) statistic are available to detect the presence of serial correlation and depending on the causes, various remedial actions can be taken. In pure times series analysis, taking account of the pattern of serial correlation usually allow for better forecast of future observations. Darnell (1994) notes that detection of a pattern of first-order positive autocorrelation by means of the residuals can result from several factors other than autocorrelated true errors (unobserved). These factors include mispecification of functional and structural breaks in the relationship (time-varying parameters). Consequently, he suggests that a respecification of the model might be more appropriate and only when the evidence strongly suggest autocorrelated true errors should the model be re-estimated by feasible generalized least squares or maximum likelihood.

PAGE 60

49 Stationarity Stationarity is an important aspect of a time series. Formally, an infinite sequence of random variables is said to be strictly stationary if the joint probability function of any finite subset is identical to any other subset (Darnell, 1994). Less formal, when a series is generated by stationary process it tends to fluctuate around a constant level and there is no tendency for its spread to increase or decrease over time. In other words, the mean of say Y t , its variance, and its co variance with other Y values, say Y,. k , are invariant to time and the covariance between say Y, and Y t+S is a function of s only, for all t. Hence an immediate implication of a stationary time series is that it has a time independent mean and is homoscedastic. Kennedy (1998) points out that although many scientific time series data are stationery, most economic time series data are nonstationary since they tend to exhibit some kind of trend over time (i.e., the mean changes over time). One reason for this (Aoki, 1990) is that the circumstances facing optimizing economic agents change with time. Another is that in several cases the dynamic structures generating the data are mostly nonlinear. Nonstationarity has implications for the standard regression model in econometrics since the model usually makes assumptions regarding the stationarity of the error term as well as that of the variables in the regression. It has been shown that running regression on nonstationary data can give rise to misleading (or spurious) values of R 2 , Durbin Watson, and t-statistics, which may lead to an erroneous understanding of the relationships which exist among the regression variables. For instance, it has long been recognized (Yule, 1926) that regressing one nonstationary (trended) variable on another could result in spurious regression.

PAGE 61

Nonstationary time series have been subjected to one of two detrending procedures in order to make them stationary before subsequent analysis. The two procedures are: 1) regressing the time series as a simple linear (or higher order) function of time and then using the residuals as the detrended series; and 2) an approach which involves using first (or higher) order differencing of the time series. The first method works only if the series say Y t has a deterministic trend and is generated according to the following equation: where e t is white noise. If however, the time series is generated by say a random walk or random walk with a drift as for example: where v t is white noise, the first approach will not work. More formally, equations (3.1) and (3.2) are referred to as trend-stationary process (TSP) and difference stationary process (DSP) with a trend, respectively. The second approach has had some measured success even though as Kennedy (1998) points out some analysts have adopted it without any justification based on economic theory. Within this approach, a variable is said to be integrated of order d, written as 1(d) if it has to be differenced d times to be made stationary. A stationary variable is integrated of order zero. Economic variables are seldom integrated of order greater than two and if non-stationary they are usually 1(1) (Kennedy, 1998). Equation (3.2) above represents a series, which is integrated of order one. Taking Y M from both sides of the equation gives equation (3.3): Y, = a +pt + e, (3.1) Y, = Y,., + P + v t (3.2) AY, = p +v, (3.3)

PAGE 62

51 which is a stationary process. If P equals zero then the series is said to be integrated (random walk) without a drift and if P 0 then it is said to be integrated of order one with a drift. Because of the serious econometric implications of modeling variables which are nonstationary, as pointed out earlier, tests for stationarity and the degree of integration have all but become routine to econometric analysis when working with time series data. The most popular tests continue to be the unit root tests which are designed to test the order of integration of the variable. Kennedy(1998) points out that although a wide variety of unit root tests are available, none of them is very powerful. The most popular unit root tests are the Dickey Fuller or the Phillips-Perron in which the null hypothesis is that the variable is non-stationary. But as Kwaitowski et al (1992) have shown conflicting results could come about from the use of these tests and the one they proposed with the null hypothesis being that the variable is stationary. Cointegration and Error Correction Models A major problem stemming from the use of variables made stationary by differencing is that valuable information pertaining to the long-run equilibrium properties of the data would be lost. The solution to this problem appears to be in a concept known as cointegration. Two times series Y t and X t are said to be cointegrated if both series are individually integrated of the same order say 1(1) and there exists a linear combination of the two variables which is not integrated. In such circumstances it will be possible to carry out a regression on levels of the variables and the results are meaningful (not

PAGE 63

52 spurious) and there is no loss of any valuable long-term information, which would result if their differences were used instead. Moreover, in such situations, it is possible to tie in the short-term behavior of a variable obtained from differencing, with a long-term equilibrium relationship. This is done with a special formulation which involves a mix of both level and difference terms. Such models are referred to as error correction models (ECMs). The ECMs can therefore be viewed as comprising the short-run transitory effects and the long-run relationship and describes how the long-run solution is achieved by a combination of negative feedback and error correction (Darnell, 1994). In this regard, it is seen as a way to model the shortrun dynamics without losing sight of the longer-term steady state as may be posited by economic theory. In other word, the ECM specification provides a means by which the short-run observed behavior of variables can be associated with their long-run equilibrium growth paths. Kennedy(1998) notes that this is one of the reasons why economists have shown such interest in the concept of cointegration, since it provides a formal framework for testing for and estimating long-run (equilibrium) relationships among economic variables. He suggests that if a cointegration relationship can be found, advantage should be taken of the ECM framework. It should be pointed out however, that the use of a single representation, implicitly assumes that all the explanatory variables are exogenous. If this is not the case, then it is suggested in the literature that one needs to use a vector auto regression (VAR) approach. Furthermore, as long as the equation contains more than two variables, there is the possibility that there could exist more than one cointegrating relationship. In such cases, the traditional estimation procedures become inappropriate and again there is need to

PAGE 64

53 switch to the VAR approach in which each variable is modeled in terms of lagged values of all the other variables (Kennedy ,1998). Within such a general framework, testing can be done to determine the number of cointegration relationships and the exogeneity of the variables. The most common method is that of Johansen (1988). Manipulation of the single equation VAR can produce a vector error correction model (VECM) in which the vector terms are expressed as lagged differenced vector terms plus a lagged level term which represents the error correction mechanism. As in the case of the ECM discussed earlier, the intent is to exploit the steady state conditions resulting from cointegration relationship and the short-term effects from the differenced variable in a single framework. A survey of the literature reveals, however, that there are unresolved issues with the concept of cointegration and the use of ECMs and VECMs. For one thing, although ECMs and VECMs are used widely in the literature, several analysts (Harvey, 1997; Hall et aL 1992; Aoki, 1990) have expressed serious reservations about the mixing of level and difference variables in a given equation. Secondly, Monte Carlo studies have shown that the estimates of the cointegration regression have considerable small sample bias in spite of excellent large sample properties ( "superconsistency"). In this connection, Cheung and Lai (1993) point to several finite-sample shortcomings in the Johansen method, and have attempted to address some of these shortcomings by adjusting upwards the critical values of the Johansen test. However, an implication of this is that the likelihood of finding cointegration with finite sample correction is much more difficult. Thirdly, in situations in which there are more than two variables in the regression equation there is the likelihood that there could be more than one cointegration

PAGE 65

54 relationship. In such situations, the standard OLS procedure does not produce estimates that are consistent (Kennedy, 1998). Also, in these circumstances when there are more than one cointegrating relationship there is usually a certain degree of ambiguity surrounding the interpretation of the estimated cointegrating vectors (Johnston and Dinardo, 1997; Kennedy, 1998). Kennedy (1998) further points out that the interpretation of multiple cointegrating vectors can be frustrating. He refers to a practice among some researchers, in dealing with this issue, of ignoring those cointegrating vectors that seems not to make good economic sense as tantamount to imposing slightly false restrictions to improve mean square error. Fourthly, according to Johnston and Dinardo (1997) VARs have serious limitations as a tool for the analysis of economic systems and are susceptible to the problem of vanishing degrees of freedom, since the number of unknown coefficients can rapidly approach the available sample size. An implication of this is that researchers wishing to use this technique, but having limited data (observations), are forced to be quite parsimonious in their model specification. However, this could be disastrous since the failure to include an important variable could result in absence of any detection of cointegrating relationship among the variables (Kennedy, 1998). On the other hand, as pointed out by Johnston and Dinardo (1997), as more variables are added to the VARs, problems arise in testing the number of cointegrating relationships. This comes about because the test statistics are nonstandard distribution and require simulation. At present the available tables can only account for eleven variables.

PAGE 66

55 TimeVarying Parameters Another potential problem with time series regression models is that the estimated parameters may change. A review of the literature indicates that econometricians have been aware of the problems of structural change since the late sixties as evidence by the work on random coefficients (Hildreth and Houck, 1968; Langham and Mara, 1973). However, it was not until the work of Rosenberg (1973) and Coo ley and Prescott (1973) that much more attention was focused on this particular phenomenon. Cooley and Prescott (p. 463, 1973), in making a case for such analysis wrote: "In recent years economic theory has increasingly abandoned the relative security of static equilibrium and perfect certainty... [as] it has become increasingly clear that to assume behavioral and technological relationships are stable over time is in many cases not only heroic, but completely untenable on the basis of economic theory." Since then there has been a proliferation of articles in the literature dealing with these issues (see for example survey by Nichols and Pagan, 1983). Ward and Myers (1979) for example, used a distributed lagged advertising model with coefficients that had random and systematic adjustments to demonstrate the dynamic effects of advertising on consumer demand. Ward and Tilley (1980), using timevarying parameters with random component, illustrated quite clearly the problems that can occur when parameters that have changed are, in fact, ignored in the specification of the econometric model. Two other interesting examples include the work of Engle and Watson (1987) in forecasting electricity sales with timevarying parameters, and the work of Harvey et al (1986) on stochastic trends in dynamic regression models, focusing on the employment—output equation. In all such models one or more of the parameters is allowed to evolve over time, and estimation techniques (ranging from simple models of varying parameters to

PAGE 67

56 more complex ones involving state-space framework and Kalman filter (discussed below)) are used for both the fixed operators and the varying parameters. Timevarying parameters is regarded as a problem because standard regression analysis of economic phenomenon assumes that the relationship being studied is stable within the sample of estimation. For example, in the simple regression model of Y, = Xp + u t it is presumed that the P vector, which represents the effect of changes in the explanatory variables upon the dependent variable, is constant. However, there are good reasons why this might not be so and in fact should be allowed to vary. Ward and Myers (1979) grouped the possible sources of parameter variation under three broad categories: 1) structural changes in the economic phenomenon being studied; 2)model mispecification; and 3) aggregation. They noted that technology and institutional changes always cause structural changes in economic phenomena. The Lucas (1976) critique of economic analysis provides a good example. Lucas' critique is to the effect that if the policy regime is changed and economic agents takes account of the general policy environment into their decision-making, then agents will adjust their behavior and the coefficients that held in the previous regime will change to new values. Sources of misspecification are varied but include omission of independent variables, wrong functional forms and the use of proxy variables. For example, if the true relation is: Y t = ai + a 2 X, + a 3 X 2 , + u t (3.4) and the analyst considers the linear relationship give by Y, = Pi + p 2 X, + v, (3.5)

PAGE 68

57 then it is clear that the effect of B 2 which is equal to a 2 + 2a 3 Xt is not a constant. In the case of aggregation as source of parameter variation, Ward and Myers (1979) cites as an example the fact that over time, the relative importance of microeconomic agencies will change and that such changes might not get reflected in the aggregate weight. In addition, they noted that aggregated(macro) variables are discrete (Le., indexed as discrete time points) whereas the underlying microvariables may be continuous, providing another reason why parameter variation are likely to occur in macromodels. In the literature, parameter variations are classified under the two broad headings of stochastic and nonstochastic parameter variation. Each of these classifications is further broken down into sub-categories. Hence, nonstochastic parameter variations, which are caused by structural change in the economic phenomenon being studied, can be divided into: 1) discrete variation or switching regression; and 2) systematic variation. Likewise, stochastic models are divided into: 1) those which are stationary and are referred to as random coefficient models; and 2) those which are nonstationary and are termed sequential (Markovian) models 13 . The State-Space Framework The State-Space Model Kennedy (1998) states that the state-space model can be regarded as a generalization of the linear regression model and provides a unifying framework for all dynamic linear models used in econometrics. A similar view is expressed by Harvey 13 For additional information on the various aspects of these types of models see Ward and Myers, 1979 and Nichols and Pagan, 1983.

PAGE 69

58 (1993, 1997) that the state-space model is fundamental to dynamic modeling, and that since most times series models can be put in this framework they provide a more robust alternative to the modeling of time series analysis. The state-space framework is a concept borrowed from engineering and is based on the original work of Kalman (1960) and Kalman and Bucy (1961). The concept centers around the possibility of tracking the economy in a similar manner as engineers would track the "state" of a system, such as the location of a satellite or a tanker, using noisy measurements. The Kalman filter (discussed below ) was used to derive optimal estimate of the state, given knowledge of certain parameters. Engle and Watson (1987) note however, that there is a slight difference between the approach used by engineers and that used by economists. In the main, engineers usually have qualitative theories that describe the equation of motion of physical systems and were primarily interested in the "state" of the system obtain from noisy measurements. Economists on the other hand, were less fortunate in not having such laws of motion of the economy at their disposal and consequently were much more interested in discovering them from noisy data, rather than in merely estimating the state of the economy. Consequently, economists had to overcome the initial problem of not having the parameters of process in estimating the state. The focus for economists therefore shifted slightly to finding a suitable methodology by which such parameters could be estimated. The solution to the problem came in the mid-1970s when economists, with the use of the Kalman filter were able to evaluate the likelihood function in complex cases. Engle and Watson (1987) note that this development allowed the parameters to be estimated using maximum likelihood methods as in standard econometrics. It therefore opened up the way for the widespread use of the technique in

PAGE 70

59 applied economics and the model became a natural generalization of latent variable models to a full dynamic framework A brief taxonomy of empirical applications of such modeling in economics can be found in Engel and Watson (1980). To illustrate the state-space model, first consider the following standard single equation regression: m y, = a + /ft + + £ t (3.6) i=0 where y t is a scalar observable data series, Xt, is the single observable explanatory variable with the structure of the lagged coefficients being 6o...6 m , t is the trend variable with level and slope coefficients a and P, respectively, and e t is the stochastic disturbance term distributed such that e, ~ NID(0,o e 2 ). Harvey (1987) asserts that while equation (3.6) is typical of many regression equations and that while many economic time series show a trend, it is equally apparent that unless the time period is fairly short the trends cannot be adequately captured by a deterministic (straight line) representation. He observed that a great deal of applied economic work fall prey to the above error, since they start off by detrending the data by regressing on time, thus making all that follows invalid. He further points out that there is no reason in principle, why the parameters of the explanatory variables of time series should not also be allowed to vary over time. He suggests that a better approach would be to formulate the model in a state-space framework. Such a representation could account for any unobservable components as well as timevarying parameters. Following along the lines suggested by Harvey (1987), reparametizing equation (3.6) in terms of a simplified version of the state-space model, appropriate for timevarying parameters yields the following:

PAGE 71

60 (3.7a) 1 = 0 M, =A-i+A-i +T Jt ( 3 7b > where it is assumed that e, ~ NID(0,o e 2 ), n,~ NID^o,, 2 ), NID(0,o 5 2 ), and v,~ NID(0,o v 2 ). The component ^ t is the trend, whereas p t is the slope of the trend and 8 it are the parameters of the explanatory variables, which are assumed to vary over time. The introduction of the stochastic disturbance terms r| t and £ t allows the level and slope of the trend to change slowly over time. Likewise the disturbance term v, , allows 6 it to be generated by a random walk. More formally, the above state-space representation with the simplifying assumption of no lags is: Y, = (1 0 l)a t + s, (3.8) where Mt 1 1 0 fix a, = Px 0 1 0 + A. 0 0 1 A. . (3.9) In terms of state-space description, such models usually consist of two parts: 1) the measurement equation which describes how the data actually observed is generated from the state variables (equation (3.7a) or equation (3.8)); and 2) the transition (dynamic) equation which describes the evolution of a set of state variables (equations (3.7b to 3.7d) or equation (3.9)). The vector ot t comprising the elements |i t , Pt, and 8 t is regarded as the state vector to be estimated and the measurements that are used to

PAGE 72

61 estimate this state are the y t . . The statistical algorithm used to derive these state parameters is the Kalman filter. Y t is the measurement equation. The parameters on the right hand side of the transition equations (3.7a to 3.7c) or more precisely the system matrix in equation 3.9, are described as the hyperparameters. They are regarded as being fixed and known although they may change over the period in question. Engle and Watson (1987) points out that on the basis of empirical experience and theoretical arguments, in most cases these transition equations should to have unit root. Concerning equations 3.7b to 3.7c, it should be noted that in the special case where o,, 2 , Og 2 , and o v 2 equal zero, the hyperparameters remain constant and equation (3.7) collapses to the standard regression equation (3.6) with the deterministic trend and fixed regression parameters. Harvey (1997) also showed that if o e 2 = o^ 2 = o v 2 = 0 and if Orf > 0, then the first difference formulation is obtained as: Ay, = P + ZS.Axt.i + r,, (3.10) Kalman Filter The Kalman filter is simply a statistical algorithm, which allows certain computation to be carried out for a model cast in the state-space form. Following the approach of Harvey (1987) the typical model for which the Kalman filter is applied takes the form of: Y, = xt6\ + 8, (3.11) 5, = Gt5,.i + ©t (3.11a) where as previously noted, equations (3.1 1) and (3.1 la) comprise the state-space system in which the former is defined as the measurement equation and the latter as the dynamic

PAGE 73

62 equation. Also, it is assumed that e t ~ NID (0, o € 2 ), and ©, ~ NID (0, aj ). Although jq , G t , ,o e 2 and a J may ultimately depend on a set of unknown parameters , they are for the purpose of the Kalman filter regarded as being fixed and known. G t , o € 2 and aj are defined as the hyperparameters and are normally supplied by the researcher. In cases where such information is not provided each defaults to the identity matrix. The parameter of interest in the model is the state variable 8 t , which evolves over time. The Kalman filter uses a recursive estimation procedure to update the state as new observations become available. In order to begin the procedure one must specify values for ,o e 2 , Oo, 2 , x, , G t and the starting value for 5o. However, since such values are not usually available the usual procedure is to construct starting values from the observations themselves. Thus, in the case of a classical linear regression model with k explanatory variables the starting values can be obtained by applying ordinary least squares (OLS) to the first k observations with G = I and a J = 0. The operation of the Kalman filter is easier understood when interpreted from a Bayesian point of view. In this regard the goal of estimating St can be though of as being carried out in two stages (Shen et al, 1999). In the first stage the 6\ is estimated prior to observing y t through equation 3.11a. This can be represented as: *,=G t 4 (3.11b) where, S t is regarded as the best guess of the true value based on all information available up to time t-1. In the second stage when the actual value of y t becomes available, S x is updated as follows: S t = G t <£,_, + A t e, (3.11c)

PAGE 74

63 where, A t e t is considered to be the correction term. This term comprise two components. The first component, the prediction error et , is obtained as the difference between the actual y t and the estimated or forecast value , y t , which is based on the prior (unupdated) estimate of 8 t i.e. d = y t y, and y, = x t S t + i , . The second term, A, , can be considered as the regression coefficient of the regression of 6t on et (Shen et al, 1 999). The entire process can be viewed as one in which the Kalman filter takes the starting estimate of 6\ and corrects it as new information becomes available, through a one step at a time process, so that the estimator at any time t is a better predictor of the actual observation. As mentioned earlier, from a practical standpoint, the starting values are usually obtained from the computation of an OLS regression. Consequently, if the equation contains k variables then k observations will be required to construct the starting values. Thus, the first k estimates are usually disregarded. Having obtained these starting values the procedure continues in a recursive manner until the final set of information at time T becomes available, producing Tk one-step-ahead prediction errors. At the stage when all information has been accounted for, a better estimator can be obtained. The techniques for computing such an estimator are known as smoothing. There are three basic smoothing algorithms; fixed point , fixed lag, and fixed interval (Anderson and Moore, 1979). The one most commonly used is the fixed interval smoother, which starts with the final Kalman filter and works backwards. It thus ensures that an optimal estimator is available at all point in time based on all information up to and including the final observation, Yt. This information as pointed out by Harvey (1987) can be quite valuable for examining the way in which a component such as trend or elasticity evolved in the past. Finally, it should be noted that when the model is linear and

PAGE 75

64 time invariant the Kalman filter will usually converge to a steady state, in that the covariance matrix becomes time invariant and the estimates of the parameters are the same as those obtained from fixed coefficient regression (Harvey, 1987). Advantages of State-Space Approach and the Kalman Filter Several authors (Harvey 1987,1997; Engle and Watson 1987; and Aoki 1990; Hall et al, 1992) have alluded to the advantages of the state-space model over the traditional parametrization of time series models such as ARMA, ARIMA, VARs and VECMs. They noted that from a strictly theoretical point of view both types of models are equivalent, because generic models in one representation can be transformed into the generic ones in the other. However, when judged on other grounds such as numerical stability, sensitivity with respect to small specification errors, statistical properties of parameter estimators, or simply ease of dealing with nonstationary series, the state-space models are superior. Moreover, they point out that in order to avail oneself of the theoretical results and computational algorithms that have been developed over the years in system literature, time series must be put in this newer state-space representation. Hall et al (1992) in motivating the discussion for the use of the state-space approach and the Kalman filter notes that although it is widely used in certain branches of engineering and by applied statisticians , it is only just emerging as a possible useful tool of the applied economists. He notes that the Kalman filter can be interpreted in terms of agents forming expectations. He points out that while the paradigm for modeling expectation is the rational expectation hypothesis (REH) where agents act as if they know the true model of the economy up to a set of white noise errors, the consideration of the information availability assumption has led critics to label it unrealistic. He notes that the

PAGE 76

65 Kalman filter can be used to address the concerns of critics of the REH, such as Friedman (1979), with regard to the information-available assumption. This he states can be done by extending Friedman's framework in which he advocates that given the true model y t = x,p + Ut (ut is white noise) agents may sequentially update their estimate of the fixed true parameter vector p as more information on (y t , Xt) becomes available (e.g. time-varying parameter models). Hall et al (1992) suggested that this could be done by: 1) assuming that agents have some prior information about (3 (at time t = 0); and 2) by allowing p to vary stochastically. Thus, agents are not assumed to know instantaneously the true model but they do use the information optimally (efficiently), hence making the information available assumption of the REH much more plausible. They further point out that in certain models, the Kalman filter can be viewed as mimicking a learning process by agents as in the case of adaptive expectation where the adjustment parameter is updated each period, based on new information. This he notes formalizes Flernming's (1976) idea of a 'change in gear' when forming expectations. He makes the point that while the adaptive expectation has been shown to be optimal (in the sense of producing unbiased forecasts) only when the data generation process is of the form of an integrated moving average, IMA (1,1), or ARIMA (1,1,1), the Kalman filter is optimal under more general conditions, and in fact produces minimum mean square estimators (MMSE) under the normality assumption. Harvey (1987,1997) has pointed out several other advantages of the state-space framework and the Kalman filter. These are summarized below 14 . 14 The approach allows unobserved components to be incorporated in a model and the Kalman filter provides the means of estimating them by way of a likelihood function. Estimates of the unobserved components can then be obtained by smoothing. It therefore allows the entire model to be set up in terms of components, which have direct interpretation.lt provides a framework in which time-varying parameters can be estimated. For example, a stochastic trend can be estimated in which the level and slope are allowed to evolve over time. Moreover, the framework is flexible enough to facilitate estimation of those

PAGE 77

66 Relevance of the State-Space Model and Kalman Filter to the Current Study The state-space framework was used in this study to estimate the derived demand for HFCS. First, as pointed out elsewhere, the industry is relatively new which limits the number of available observations (approximately 22). Also as discussed earlier, the VAR based cointegration approach is biased with finite samples. In addition, one runs the risk of encountering the vanishing degree of freedom problem referred to by Johnston and Dinardo (1997). Second, there is strong evidence (Chapter 2 ) to suggest that the parameters of the demand equation have been varying overtime . For instance, the growth rate of the demand for the product has varied considerably over the investigated period. This suggests some kind of a stochastic trend rather than a deterministic trend. Also, as pointed out elsewhere, the adoption of the product was gradual implying that parameters which do not vary over time, since the hyperparameters, which allow the parameter to vary, in these cases would be set to zero, resulting in the estimation of fixed coefficients. These models in forecasting put more weight on the most recent observations; the faster the level and slope change, the more past observations are discounted.The state-space approach allows one to work in level data, rather than differencing, even in cases where the data are n on stationary. Working in level form makes interpretation of results much easier. He notes that traditional time series analysis, in situations where the data is nonstationary, stresses the role of differences. Within the state-space framework it is not usually necessary to difference in order to specify a suitable model. A similar view is expressed by Aoki (1990) that while economic time series are usually nonstationary because circumstances facing optimizing economic agents change with time and do not remain the same, the Kalman filter can deal more effectively with nonstationary time series than can the traditional time series methods. Within the state-space framework it is not necessary to test for unit root, as concerns over the degree of integration of a series are not crucial. The state-space frame work is flexible enough that if the slope parameter is deterministic, the hyperparameters which allow it to change over time will be estimated as zero or close to zero, hence little is lost if indeed the series is not integrated The situations where VECMS can be usefully employed are quite limited and one should have reservations about using them to provide a general vehicle for modeling economic time series. These reservations stem from the fact that the VAR-based cointegration methods are based on auto regressive models and like the unit root tests can have very poor statistical properties. In addition, it is very difficult to fit autoregressive models to data with slowly changing seasonality or trend. The strongest objection, centers not on the fact that when there are two or more cointegrating relationships they can only be identified by drawing on economic knowledge, but rather that the VECM does not provide "a sensible vehicle for modeling the short-run... since it confounds long-run and short-run effects." As an alternative to the VECMs Harvey proposes the use of multivariate structural time series model within the state-space framework noting that such modeling can be formulated to incorporate long-run components of cointegration directly by means of common trends.An added advantage of the state-space framework and the Kalman filter is that the researcher need not understand the workings of the filter to apply the technique and as such is free to concentrate on the selection of a suitable model and its interpretation.

PAGE 78

67 time was needed for firms to learn about the product, determine its compatibility with the manufacturing process, to formulate new recipes as well as make adjustment to existing ones and to assess new product attributes. Also, with regard to some of the other parameters such as the own price and cross price elasticity, for similar reasons mentioned above, there is a strong suspicion that these might have undergone changes over the time period, in view of the considerable changes which have taken place in the development of the product making it far more attractive to the end users and consumers. Third, is the fact that the objective of the study is not so much concerned with whether there exists a long term relationship among the chosen explanatory variables, as it is to identify the dynamic path of adjustments which might have occurred in the industry and to draw inferences. As a consequence, the issues of cointegration and unit root, though important, are of lesser concern within the present context. As Engel and Watson (1987) observed that in most of the studies where timevarying parameters have been used the predominant consideration seems to be on the stability of the regression equation, only in a few cases have the variations been interpreted as economically important. The current study intends to add to those few studies. Finally, the use of the state-space framework within this study is an attempt to make use of some of the theoretical and practical advances that have been made in the area of systems literature and to provide yet another case where the techniques have been used in economics. The Concept of a Derived Demand The term derived demand is normally used to denote the demand for inputs that are used to produce the final product. According to Shim and Siegel (1995) the term

PAGE 79

68 refers to the demand for a factor of production or product that is derived from the demand for other goods. Neoclassical theory suggests that a profit maximizing firm will employ a combination of inputs up to the point at which the additional cost of employing one more unit of that input is equal to the additional revenue generated from the presence of that that input. Given such a framework, a firm's input demand functions (derived demands) can be obtained in one of several ways. First, by solving the firm's profit maximization objective function with respect to the given input. Alternatively, a system of derived demands can be obtained by application of duality theory. This theory suggests that given either the firm's cost function or its profit function and assuming that such functions satisfy a set of regularity conditions, the system of derived demands can be obtained by applying Shephard's lemma and Hotelling's lemma to the given cost and profit function, respectively (McFadden, 1978). For example, given a simple firm which produces one output using three inputs—the third of which is fixed in the short run—the profit value (optimized) function can be represented as follows: 7i* = 7i* ( P, v, w, Z) = Max [Pq vxi WX2 : f(xi, X2 , Z) > q ] where n* represents the maximum profit that the firm may obtain by allowing a subset of inputs and output to vary while another subset of input is held fixed; P is the price of the output; v and w are the prices of inputs xi and X2 , respectively; xi and x 2 are the levels of the variable inputs used in the production process, while Z is the fixed input; q is the quantity of output and f represents the production function. Since the profit function as specify above is the result of a maximization process, then by applying Hotelling's lemma to the function the derived demands as well as the output supply can be obtained as follows:

PAGE 80

69 /fa ~ x \ (P* v, w, Z) < 0 => derived demand for input one /gp 0 => supply function The main difference between the derived demand obtained by way of the profit function and that obtained from the cost function is that in the former output is allowed to vary whereas in the latter it is held constant. Once the derived demand is obtained, the own price elasticity of demand for the input as well as the cross price elasticity of demand for an input can be computed. In the case of the former, the elasticity will always be negative once the input is not considered to be inferior. This is because in production theory, unlike the consumption theory, the substitution and output effects always work in unison. In the case of the cross price elasticity the sign is ambiguous. The ambiguity is due largely to the fact that in this case the substitution and the output effects may not always move in the same direction. The magnitude of both the own-price and cross-price elasticities is less predictable and depends on factors such as the elasticity of substitution, the importance of the input in terms of its share of total cost of production of the final good, and the price elasticity of demand for the good being produced. According to Nicholson (1995) in general the ownor cross-price elasticity of demand for any input will be greater (in absolute value): 1) the larger the elasticity of substitution of that input for other inputs; 2) the larger is the share of total cost represented by expenditures on that input; and 3) the larger is the price elasticity of demand for the good being produced. Thus, it is possible that given the profit function of the industries utilizing HFCS as an input, and the assumption that such industries are operating within a perfect competitive framework with a profit maximization objective, the derived demand curve

PAGE 81

70 for HFCS can be obtained. However, in light of the difficulties surrounding the estimation of a reliable profit function, such as obtaining cost sensitive information and quantities used, a more straight forward approach can be taken in which the derived demand is estimated directly on the basis of the explanatory variables suggested by the above theory. These explanatory variables include the price(s) of the final commodity(ies) and the respective prices of the various inputs used in the production process. This latter approach will be used in this current study.

PAGE 82

CHAPTER 4 CONCEPTUAL MODEL OF US SUGAR AND HFCS MARKETS As the process of development takes place, sweeteners in general are increasingly being consumed as raw materials and inputs of industrial production and to a lesser extent by households as essential foodstuffs 15 . Thus, in developed countries and regions such as those of the EU, US, Canada and Japan, between 60%-70% of the sweeteners consumed are in such goods as cakes, soft drinks and ice creams. However, in the developing countries the situation is the reverse where only about 30%40% of the consumption of sweeteners is in the form of sugar containing products (Atlas, 1996). This has obvious implications for the patterns of demand, the direction of technical progress and the underlying scientific investigation. The purpose of this chapter therefore is two fold. First we construct a conceptual model of the US HFCS and sugar markets focusing on the derived demand nature of the demand for sweeteners. While the model is essentially static, some attempts are made to examine the dynamics of the relationship between HFCS and sugar. Second, the conceptual model is used to assess some of the likely trade and welfare implications resulting from changes to the US sugar program and an expanded HFCS industry within the framework of the proposed FTAA. In carrying out this assessment two scenarios are examined. The first focuses on the likely impact resulting from a partial liberalizing of the US domestic sugar program, while the second considers the more unlikely situation of the complete abandonment of the program. 15 Industrial (manufacturing ) uses include those for food (bakeries and cereals, confectioneries, soft drinks and canned and frozen dairy products) and non-food items such as ethanol. 71

PAGE 83

72 The US Sugar and HFCS Subsectors Figure 4.1, presents a flowchart of the US sugar and HFCS sweeteners' subsectors. As stated earlier, HFCS is available mainly in a liquid form and is an almost perfect substitute for liquid sugar, hence it is used in its entirety in the industrial sector. Sugarcane Impt. Impt. Expt. Raw Sugar ! * { By \ product Refined Sugar Sugar beet Industrial Demand T Consumer Demand product / Corn Wet Milling By product i £ HFCS Impt. Expt. Fig. 4.1. Flowchart of the US HFCS and Sugar Subsectors As evident from the Figure, useful byproducts are produced in the production of both sweeteners. However, the revenue generated from the sales of the byproducts as a proportion of the cost of the main input— sugarcane, sugarbeet and corn, respectively —are more substantial in the case of the production of HFCS.

PAGE 84

73 The US Sugar Subsector Focusing first on sugar, it should be noted that several studies have attempted to model the US sugar market (Haley 1998b; Atlas, 1996; Bernirschka, Koo and Lou, 1996; Marks 1993; Schmitz and Christian, 1993 ; Barros, 1992; Sudaryanto, 1987; Leu et aL 1987; Zietz and Valdez, 1986; Gemmill, 1976). Sudaryanto pointed out that many of the studies fail to distinguish between direct and indirect consumption. They omit variables explaining the derived demand characteristics of the consumption pattern. This study recognizes the point made by Sudaryanto and others (Haley, 1998b; Barros, 1992) on the need to disaggregate the market for sugars into derived and direct demand components. However, a slightly different approach was taken in separating the market demand components. For convenience, a distinction is made between crystallized sugar (CS) and liquid 16 sugar (LS) . In this study, crystallized (tabletop) sugar include sugars used both by households and those segments of the industrial sugar users, in which it is currently technically impossible to substitute sugar with liquid HFCS. The importance of this distinction will become evident later in the analysis. However, for the time being there is some evidence suggesting that at this juncture, it is in respect of the demand for liquid sugar that a structural shift in demand has taken place. Only the liquid HFCS competes with sugar. As pointed out earlier, liquid sugar and liquid HFCS are regarded as almost perfect substitutes (Marks, 1993; Barros, 1992). Further supporting evidence for this 16 Refined sugar is available in three main product categories: white granulated sugar, liquid sugar and specialty sugars. Granulated sugar is the most common form of sugar used in the households. Liquid sugar is a mixture of water and sugar that is preferred by some manufacturers and specialty sugars include icing sugar, brown or yellow sugars. Chemically, there is no difference between liquid sugar and crystallized (granulated and specialty).

PAGE 85

74 assertion exists when consideration is given to the nature of the penetration of HFCS in the sugar consuming industries. For example, Haley (1998b) points out that in the beverage industry, where a relatively large proportion of liquid sugar was used, HFCS has almost totally replaced the higher-priced sugar. Indeed, all major sweet drink manufacturers in the US use HFCS as their main caloric sweetener. Heuristically, one can therefore envisage the US sugar market as exhibiting two distinct demand configurations in the early stages of the HFCS development—one inelastic and the other being relatively elastic. The inelastic curve represents the demand for crystallized sugar (CS) as shown in Figure 4.2a. The curve is assumed to be inelastic due to the absence of any close substitute. It should be noted however, that the exclusion of the non-caloric sweeteners from this model implicitly assumes no formidable Price Price Fig. 4.2a Quantity P k Fig. 4.2b Quantity Fig. 4.2. Hypothetical US demand for crystallized sugar (CS) and liquid sugar (LS), respectively

PAGE 86

75 challenge by that industry to this segment of the market. The second demand curve represents the demand for liquid sugar (LS). As argued earlier, it is liquid sugar for which HFCS is almost a perfect substitute (Figure 4.2b). The presence of this substitute (HFCS) causes the demand curve for LS to become relatively elastic (rotating to the right as shown in Fig. 4.2b) reflecting the fact that manufacturers are much more apt than households to switch to a lower cost substitute because of the need to remain competitive . This characterization of the demand for sugar although novel, is nonetheless anchored by evidence from the literature. For instance, Sudaryanto (1987) in partitioning the US sweeteners market into a direct and derived demand for sugar found that the elasticity of the derived demand was far greater than that for direct consumption. His explanation was that since the consumption of final uses (direct) had approached saturation, the responsiveness to a change in the price of sugar would be minimal (inelastic). On the other hand, the demand for sugar in the food and beverage industry was more elastic since many more substitution possibilities were available. The Sudaryanto explanation of the differences in sugar demand configuration is somewhat similar to that advanced by Thomas (1985). Specifically, Thomas noted that the US market favored the development of the HFCS industry, since there existed high price elasticity of demand for sugar among industrial sweetener users. These users, he suggests, are far more ready than households to substitute one sweetener for another, 17 Economic theory holds that the price elasticity of demand for any good depends on many economic, social, and psychological factors. Included among the economic factors are the availability of close substitutes and necessity of a good. In general, necessities tend to have inelastic demand whereas goods with close substitutes tend to have a more elastic demand (Mankiw, 1997)

PAGE 87

76 given favorable price changes, and the technical characteristics of the individual sweeteners. Horizontally summing the two demand curves, CS and LS, in Figures 4.2a and 4.2b, gives us Figure 4.3, which represents the total market demand for refined sugar in the US. Of significance in our model is the kinked nature of the resulting demand curve ake. Point k on the curve, which represents the kink, can be interpreted as follows: At prices above pK there is no demand for liquid sugar~i.e. it is completely substituted for by HFCS. Below the kink however, there is both a demand for crystallized sugar (CS) and liquid Sugar (LS), the latter in competition with HFCS. Price P k Quantity Fig. 4.3. Hypothetical US aggregate demand for sugar In Figure 4.4, the US domestic sugar supply curve (S s ) is introduced and the established US domestic support (loan) price, P T , set at approximately US 23 cents per pound refined sugar. As drawn, the Figure indicates a situation in which there still

PAGE 88

77 remains a market demand for liquid sugar. Thus, Xi and X3 indicate the quantities of sugar produced and consumed, respectively. As such X3 X| represents the US tariff-rate quota (TRQ), which cannot fall below 1.14 MMT. X2 represents the quantity of CS demanded and X3X2 is the quantity of LS demanded. It should be noted that the quantity of LS demanded does not represent the total demand for liquid sweetener that would include the amount being satisfied by HFCS. Finally, Pw gives an indication of the Price ySs Pt i\ Pw 1 r 1 1 1 1 1 1 1 \ \ \ \ X e \ \b Xi X2 X3 Quantity Fig. 4.4. Hypothetical US supply and demand for sugar world market price in relation to the support price. Thus, the US sugar market might be represented by the following equations: Demand for CS: Dcs = f(Ps, Pf, Z) (4. 1 ) Demand for LS: Dls = f(Ps, Ph(.),Qh{.),Pf,Z) (4.2)

PAGE 89

78 Demand for Sugar: Ds = Dcs + Dis (4.3) Supply of Sugar: Ss = ScQ+SbQ (4 4) Import Demand: Id = DsSs > 1 .256tonnes Equilibrium Condition: Ds = Ss + Id (A (\\ where: Ps domestic price of sugar in the US Pf index of sugar and HFCS containing products Ph price of HFCS Qh quantity of HFCS supplied Sc &S b supply of cane and beet sugar, respectively Z a vector of other factors The US HFCS Subsector Turning attention to the HFCS subsector, the demand curve for HFCS is conceptualized, as shown in Figure 4.5. Such a demand configuration is based on the following grounds: 1) the observation that although HFCS is almost a perfect substitute for liquid sugar (LS) it has always been sold at a discount to LS on a sweetness equivalency basis (Fig. 2.3); and 2) the dual nature for the commodity, as discussed in chapter one, which suggests that HFCS can only be substituted for sugar within a certain range of the sugar demand function. It should be pointed out that above the controlled price for sugar, P T , the particular shape assumed by the demand curve is very ambiguous. This is due to the fact that the price of HFCS has never exceeded that of sugar or even come close to it. If one is willing to assume the conclusion reached by Tanyeri-Abur

PAGE 90

79 (1990) that if the price of HFCS should ever exceed that of sugar the industries would revert to using sugar completely, then this would suggest that the demand curve kinks at or just above the controlled price and become highly or even infinitely elastic. However, if one were to accept the arguments put forward by Marks (1993, p. 80 ) that "a reversal of the time path of the real US sugar prices would hardly likely bring about the same substitution pattern in reverse" then one might conceive of the portion of the demand curve above the controlled sugar price, tapers off as shown in Figure 4.5. Price Pt \ \ \ HFCS demand Quantity Fig. 4.5. Hypothetical US demand for HFCS Several factors could account for a shift in demand for HFCS. As discussed in Chapter 3 in the section on derived demand, the demand for an input depends considerably on the demand for the final good, hence any change in the demand for the latter will cause the derived demand for the input to also shift. Thus, a shift in the demand for HFCS could come about owing to any or a combination of the following: 1) changes in the prices of

PAGE 91

the other inputs used in producing the final good, especially the price of sugar; 2) an increase in the number of products using HFCS; 3) seasonality ; 4) changes in the price of the final commodity; and 5) an improvement in technology which improves the substitutability of HFCS for sugar. Consequently, the derived demand for HFCS can be represented by the following general specification: Demand for HFCS: Dh = / (Ph, Ps, Pf, T, Z) (4.7) where: Dh = demand for HFCS P H = price of HFCS Ps = domestic price of sugar in the US Pf = price index of sugar and HFCS containing products T = technology (trend variable) Z = a vector of other factors With respect to the supply, it is important to note that the consensus of industry experts is that the HFCS industry operates in an oligopolistic framework (Polopolus and Alvarez, 1991, Thomas, 1985). A consequence of this is that even if the exact profit function of such an industry were available it would not be possible to obtain the input supply function of the industry by way of the duality theory, since the underlying assumption of the theory is the existence of a perfectly competitive framework. Moreover, in light of the prevailing circumstances under which the HFCS is produced, notably a kind of captive market resulting from the restrictive policies of the US sugar program and the implicit floor price of sugar, it is argued that there is no supply function for the industry but rather a reaction function. This reaction function can be viewed in the context-although not identical—of the Stackelberg model in which one producer of the

PAGE 92

81 substitutes has market intelligence and knows the reaction function of his competitor but the other is naive. Consequently, the producer with the market intelligence take advantage of his competitor by supplying an amount at a price which will maximize his profit at the expense of his competitor. Here, the HFCS producers could be regarded as the sophisticated suppliers, while those supplying sugar are regarded as naive. Taking this analogy a step further, as in the case of the Stackelberg model in which it is possible for one competitor to force another out of the market by sufficiently differentiating the product, in a similar manner such an opportunity exist for HFCS to force liquid sugar out of the market. More specifically, it is assumed that the profit maximizing objective function of the HFCS suppliers can be represented as follows: n = (P h -C h )*Dh(.) (4.8) where n represents the maximum industry profit; P H is the price of HFCS; C H is the cost of producing HFCS and D H is the demand for HFCS as defined in equation 4.7. Assuming other factors remain constant, the reaction function of the HFCS suppliers is obtained by differentiating the profit function with respect to the price of HFCS, setting it equal to zero and solving for the price of HFCS. The reaction function will therefore reflect the best price to charge for HFCS given the price of sugar and the other exogenous factors that affect profit of the HFCS industry. Within the framework of the above conceptualization, it is assumed that: 1) the demanders of HFCS are price takers; and 2) that the marginal cost of producing the HFCS remains constant over the entire range of output. The first assumption seems plausible since in comparison to the number of firms producing HFCS there are large

PAGE 93

82 number of firms that use HFCS as input into their production process. With respect to the second assumption , this again seems plausible given that, the industry is highly capital intensive and that there currently exist a considerable amount of excess capacity within the HFCS industry . In addition, the amount of corn used in the industry is less than 1 0% of the overall US corn production, and consequently a moderate increase in demand for HFCS is not expected to have any major impact on the market price of corn. Finally, the industry is highly capital intensive and as such one might expect that the cost of an additional unit energy will remain fairly constant. On this basis, and in the absence of concrete cost of production data, the short-run marginal cost curve of the industry (the sum of the marginal cost over the variable cost) is conceptualized as shown in Figure 4.6. The curve is drawn showing a section which is nearly perfectly elastic (constant marginal cost) over some quantity range and becoming vertical as output capacity is reached. Price SMC, HI SMC, Quantity Fig. 4.6. Shift in hypothetical HFCS short-run marginal cost curve In Figure 4.6, SMC H i represents the industry short-run marginal cost function in period one and SMCh2 is the short-run marginal cost curve in period two. As drawn the shift in the marginal cost curve represents an overall reduction in the cost of production.

PAGE 94

83 The shut down price is represented by P . The main factors which appear to influence the marginal cost of producing HFCS are: 1) the net price of corn (price of corn less the price of the byproducts); 2) the price of enzymes used in the conversion process; 3) the cost of energy and capital; 4) labor cost; and 5) improvement in technology. The hypothetical US HFCS market, as depicted in Figure 4.7, is obtained by combining Figures 4.5 and 4.6. This representation is consistent with the view of industry experts that the producers operate within a basically oligopolistic market structure, Fig. 4.7. Hypothetical US HFCS short-run marginal cost and demand configuration implying among other things that demand and supply of the product do not determine the pricing of the commodity. Moreover, because of problem with storage, production closely matches that of the quantity demanded at the established price (See Chapter 2). In Figure 4.7, it is assumed that the general pricing policy followed is that discussed earlier, whereby the producers of HFCS having determined the best price for the commodity, based on their reaction function and knowledge of the sugar subsector, supply a quantity Price HFCS demand QhI Qh2 Quantity (sugar equiv. basis)

PAGE 95

84 of the product which closely matches that demanded. This pricing strategy implies, among other things, that the quantity supplied and demanded does not occur at the intersection of the industry short-run marginal cost curve and the demand curve, but rather at the point where the demand curve intersect with the established price. The significance of this as far as estimation is concerned is that price and quantity observations for the HFCS industry traces out the demand for HFCS. To the extent that the prices received by the producers of HFCS is above the average cost of production, they will be receiving economic rent and are in a position to adjust price of HFCS within a certain range, in order to stave off the challenge from sugar in order to maintain or increase market shares. Based on the above line of argument, it is therefore theoretically possible to estimate the derived demand function for HFCS using shifters and the observed price and quantity data. This is done in the next chapter. With respect to the long-term prospect for the supply of HFCS, there is consensus that the industry is expected to benefit much more in the future than the sugar industry owing to the fact that there exist both internal as well as substantial external economies of scale (Marks, 1993; Thomas, 1985). Economic theory holds that in cases where there are internal economies, one can expect a reduction in the average cost of the particular firm as it increases its output. It also suggests that in the case of external economies of scale the average cost of the typical firm declines as the output of the industry expands (Lindert and Pugel ,1996). These authors noted that external scale economies can come about in a variety of ways, including: 1 ) specialized services for the industry; 2) the spill over effects of new knowledge about the product and production

PAGE 96

85 technology; and 3) specialized machinery. They further suggest that if scale economies are substantial over a large range of output, then a likely result is that a few firms will expand their operation in order to reap the scale economies, giving rise to oligopoly market structure. As it is well known, firms in an oligopoly can influence prices, can earn economic profit and can vary the level of competition. More importantly, the existence of scale economies has been shown by authors such as Krugman (1979) to provide as solid a basis for trade as is comparative advantage. This line of argument forms the basis of the New International Trade Theories (NITTs). Much of the above is true for the HFCS industry, which has benefited substantially from path breaking gains in biotechnology and processes such as fractionation technology. Indeed, there is the view that the HFCS industry represents the successful marriage of two industries: the wet corn processors and the enzymes industry. In light of the above and the feet that the industry is still relatively young and continues to benefit from technological improvements, the study assumed a long-run HFCS industry average cost reflecting external economies as depicted in Figure 4.8 . Noticeable is the shape of the cost curve (LRAC) implying possible gains from increasing return to scale. The figure also shows two hypothetical demand curves Dus and Dfta , reflecting the possible increase in the demand for HFCS as a result of the formation of a free trade area, emphasizing the possible gains to be made from the lowering of the average cost. The assumption of the long run shape of the HFCS industry cost function forms a crucial part of the subsequent analysis of the trade implications within the framework of the proposed FTAA.

PAGE 97

86 Price LRAC Quantity Fig. 4.8. Hypothetical HFCS industry average costs, reflecting external economies The Interrelationship between the US Sugar and HFCS Markets As indicated earlier there is a unique relationship between the US sugar and HFCS markets. Sugar can substitute in all cases for HFCS but HFCS can only replace sugar in a selected number of uses—albeit in such cases it has proven to be quite an effective substitute (Barros, 1992). The implication is that it is only one segment of the demand curve for sugar that is affected by changes in the demand and supply conditions within the HFCS subsector. As indicated earlier in relation to Figure 4.3, the kink in the sugar demand curve is of importance since it indicates the price above which there would be no demand for liquid sugar. Several factors affect the positioning of the kink. This point is not stationary but shifts in relation to changes to these factors. The primary factors determining the positioning of the kink are: 1) the availability of other substitutes; 2) the price of HFCS which itself is influenced by several factors including the price of

PAGE 98

corn, the prices of the by-products, the price of domestic sugar; and 3) the supply behavior of HFCS. These three factors are discussed below in highlighting some of the dynamics of the sweetener market. First, assuming other factors remain constant, it can be expected that an increase in the number of available substitutes and the closer the substitutes on a per unit sweetness equivalency basis the lower will be the positioning of the kink. This follows on the assumption that there does not exist any switching cost for the manufacturers (user of sweeteners) in utilizing one input over another. This latter assumption however is not entirely accurate since such costs do exist in one form or the other. However, the general principle applies in cases where such costs are negligible compared with the anticipated increase in profits resulting from the change. Second, and of greater importance is the relative price movement. For instance a decrease in the price of HFCS is expected to cause the kink in the sugar demand curve to shift downward, from ke to k 'e ', implying a lower level trigger price that sugar would have to be sold in order for it to become an effective HFCS substitute (Figures 4.9a and 4.9b). The converse situation would also hold, that is, an increase in the price of HFCS relative to the price of sugar shifts the kink, ke, upwards, implying an increase in the trigger price at which sugar becomes an effective substitute for HFCS. These movements postulated are consistent with economic theory which holds that a reduction in the price of one commodity that is a close substitute for another commodity will cause a downward shift in the demand for the substitute commodity. Thus, the change in the relative price of HFCS only caused that segment of the sugar demand, (below the point of the kink in Figure 4.9a, which is ke) to shift, since it is this segment which represents the demand for

PAGE 99

88 liquid sugar, the HFCS substitute. It should also be noted that, other factors remaining constant, the movement of the segment of the curve ke occurs along the dotted portion of the demand curve akb as shown in Figure 4.9a for the case of a shift from ake to ak'e '. The nature of this movement has important trade implications. For one thing, it implies that lowering the US support price for sugar (P T ) would induce only a slight increase in the quantity of crystallized sugar demanded if the final support price is still above the kink. This postulated movement is contrary to popular view of a substantial increase in sugar demand associated with sugar price reduction in the US (Atlas, 1996; Schmitz and Christian, 1993; Sudaryanto, 1987). This is unlikely, unless the price of sugar is lowered Fig. 4.9. Hypothetical supply and demand for sugar, and hypothetical demand for HFCS, respectively

PAGE 100

89 substantially so that it falls well below the point of any kink in the demand curve. And, as noted earlier, the location of the point of the kink is a function of the dynamics of the rapid economic and technological changes occurring within the HFCS industry. The third factor mentioned above which influences the positioning of the kink is the supply behavior of HFCS. This factor is closely related to the previous two factors discussed. Marks (1993) in alluding to the pattern of the substitution between HFCS and sugar asserts that it is a function of both the demand for sweeteners and the supply behavior of HFCS. Elaborating on the latter factor, he postulated that a reversal of the time path of the real US sugar prices would hardly likely bring about the same substitution pattern in reverse. His claim for the likely failure of the reverse pattern of substitution to occur is based on the view that "the HFCS industry is in place with hundreds of millions of dollars of sunk costs to protect and the advantage of learning-bydoing and other scale economies" (p. 80). The implication of his assertion is that due to the advancement made in the technology of producing HFCS, the benefit of product development and the need to protect investment, the shifts in the sugar demand could be structural. Moreover, on the demand side there also exist switching costs associated with changing from one form of sweetener to the other. For example, in switching from liquid sugar to HFCS certain costs would have been incurred both in modifying recipes and in putting in place certain physical infrastructure such as storage facilities. While these factors do not guarantee complacency on the part of the HFCS supplier, they imply a lowering of the kink and that the price of sugar would have to fall substantially to bring about a reversal of the current trend of sweetener usage.

PAGE 101

90 Related to the issue of the supply behavior of HFCS and the kink is the structure of the HFCS industry. Specifically, the degree of market power which exists in the industry can influence the positioning of the kink. In general, other factors remaining constant, it can be assumed that the less the intensity of competition among the HFCS producers (the higher the degree of market power among the producers) the greater will be their influence on the positioning of the kink. This is evident since it implies a greater probability that the producers of HFCS will be able to influence the price of HFCS covertly, which in turn will influence the positioning of the kink. Trade Implications of Changes in the US Sugar Program within an FTAA Given the conceptual model developed for the US caloric sweeteners industry the question now addressed is "What are the implications of changes in the US sugar program within the context of an FTAA, for the rest of the sugar producing members, in particular those of the CARICOM?" This question attains some primacy based on the objectives of this study. In this section, an attempt to address this matter using the conceptual model developed. The analysis starts with a brief review of the concept of a free trade area (FT A) as background to the subsequent analysis. Free Trade Area In the literature, a free trade area in the strictest sense is defined as an association of countries that have agreed to remove all trade barriers among themselves, but retain their individual national barriers against trade with third countries. In such an area it would still be necessary to have custom officials at the borders to ensure that goods from third countries do not enter through lower barrier members and trade within the higher

PAGE 102

barrier members. A Customs Union on the other hand, is where, in addition to the removal of all barriers to trade among the members, there is the adoption of common set of external barriers thus avoiding the need for custom inspection at the internal borders (Lindert and Pugel, 1 996). However, in practice it is not uncommon to find a hybrid of the two integration forms such as is the case of NAFTA, where most but not all of the barriers to trade are removed and there is an agreement on phasing and harmonization of barriers to third countries. While details of the specific form that the proposed FTAA will take are not yet available, one can nonetheless advance a number of scenarios and examine the consequences for the sugar and HFCS industries. In advancing these scenarios cognizance is take of the changes to the US sugar program as a result of the FAIR 1996 Act. In particular, note is taken of the attempt by the US authority to make the program much more market oriented by the introduction of the recourse loan policy, which essentially weakens the domestic guaranteed price. At the same time, the elimination of the no-net-cost provision of the program makes the drawing of any definite conclusion with regard to future state of the program ambivalent. In this regard consideration is given to the view of Schmitz and Polopolus (forthcoming) that the changes to the program as reflective of the FAIR 1996 are minor. However, as the corn refiners seek to expand their markets overseas, in light of: 1) the importance of sugar to the prospective members of the FTAA; 2) developments within the WTO/GATT negotiations: 3) internal domestic pressure from other commodity groups for which the changes under FAIR were more severe; and 4) the weakened (although still powerful) US sugar lobby it is postulated that the current US sugar program will need to undergo substantial changes within the FTAA framework. In this connection, the following

PAGE 103

92 analysis which is predicated on the foregoing conceptual model of the US sweetener market, considers two scenarios: 1) a partial liberalization; and 2) a complete liberalization of the US sugar program. Scenario IPartial Liberalization of the US Sugar Program Trade implications In view of the sensitive nature of sugar and the still relatively strong sugar lobby within the US, one possible scenario is a partial liberalization of the US, one possible scenario sugar program. It is conceivable that within an FTAA negotiation framework the US could offer initially, to reduce its current level of support to domestic sugar producers with an agreement that other member countries of the FTAA open up their markets to trade in HFCS . For example, the US could propose lowering the domestic support price to say 1 5 cents per pound—a price most experts feel would be the true world free trade price. The effect of this decision is reflected in Figure 4.10a in terms of our conceptual model as a fall in the support price from Pji. Here, it is assumed that the initial situation which existed in the US market was such that at the current support price there were no demand for liquid sugar i.e. substitution had been completed. This is reflected by the fact that the kink is placed below the support price Pti. With the lowering of the support price, two cases are possible. The first case is where the new support price Pt2 is at or above the kink in the US sugar demand curve (Dsi). For convenience it will be assumed initially 18 A similar agreement was negotiated within the framework of the NAFTA Agreement between the US and Mexico whereby provisions were made for the phased opening up of the US sugar market to excess supply from Mexico over a 15-year period, while agreeing that Mexico's import tariff on US exported

PAGE 104

93 that the price ratio of HFCS to sugar remains constant. This assumption is made to prevent the shifting of the kinked portion of the sugar demand curve as discussed earlier. In this case it can be seen that such an outcome would not affect the US demand for liquid sugar and hence the demand for HFCS. In other words, any changes in the US sugar program which result in the new price still being at or above the original kink will not affect the US demand for liquid sugar, as long as the price for HFCS remains below the domestic price of sugar. This is so because the price decline, although it increases the Price Price Price Fig. 4.10a. US sugar market Fig. 4.10b. Trade diagram Fig. 4.10c. ROW Fig. 4.10. Trade implications under partial liberalization of US sugar program HFCS would be no more than 15% and would be reduced to zero over a 10-year period (American Sugar Alliance, 1998).

PAGE 105

quantity of sugar demanded, from X2 to X3 ^ and the import quota , it causes only a movement along the inelastic portion of the sugar demand curve. This implies only a slight increase in the demand for crystallized sugar (CS). Depending on where the kink exists in relation to Pti, such a move could represent an appreciable fall in the support price. The second case considered is where the new support price falls below the original kink in Figure 4.10a--i.e. a movement from P T i to P T 3Here it would appear that the new support price of sugar makes it competitive with HFCS and so there is an effective demand for some liquid sugar. Other factors remaining constant, the consumption of sugar (both LS and CS) appears to increase substantially from X2 to Xs . This would certainly be the case, all other factors remaining constant. However, one needs to consider the changes, which would have been occurring almost simultaneously within the HFCS industry. As the hypothetical model of the long-run situation in the HFCS industry shows (Figure 4.1 1), as a consequence of the external economies of scale there are potential gains to be made as a result of the increased demand resulting from the opening up of the HFCS industry to international trade. Furthermore, to the extent that the industry retains some of its market power in the expanded market, producers will be able to obtain economic rent as shown by the hatched area in Figure 4. 1 1 . As such gains are realized, it would result in a lowering of HFCS, which would result in lowering of the kinked portion of the US sugar demand curve— i.e. from Dsi to Ds2 in Figure 4.10a.(see earlier discussion on page 91). To the extent that the new kink D S 2 is below P T3 , (as shown) the situation is very much the same to that examined in the earlier case—i.e. the change in the US sugar

PAGE 106

95 policy having no effect on the US demand for liquid sugar. Consumption of CS increased marginally from X2 to X4. If instead the new price is below the kink in Ds2, then it could imply a sharing of the market between HFCS and liquid sugar. To complete the analysis we invoke the large country trade assumption and construct the usual trade-diagram and rest of the world (ROW) exporters, Figures 4. 10b and 4.10c, respectively. Assuming that the increased demand for HFCS, resulting from the opening of the industry to trade within the FTAA, works mainly through the demand for sugar in the rest of the world (Figure 4.10c), this development can be modeled as a combination of two effects. Firstly, the presence of a substitute for liquid sugar in the market causes the demand for sugar in the ROW to become more elastic. This is modeled as a slight rotation of the demand curve. Secondly, as the price of HFCS declines and Price P2 C L Pi LRAC Y, Y 2 Quantity Fig. 4.11. Hypothetical HFCS industry long-run demand and average cost reflecting external economies non-US users make the switch, the demand for HFCS gains at the expense of sugar. This results in a leftward shift in the rest of the world demand for sugar, from RWDsi to RWDs2. In other words, the increased consumption of HFCS in the

PAGE 107

rest of the world will be at the expense of sugar. This leftward shift in the ROW demand curve is explained by assuming that as the income of the members of the FTAA improves, as a result of integration, there is for example a decrease in home baking and an increase in eating outside of the home. Since less food is prepared at home, and since it is more in the away-from-home food that there is a demand for HFCS, an indirect substitution of HFCS for sugar occurs. Assuming further that the ROW sugar supply remains the same-given the limited incentive in the US market resulting from the change in policy—the net effect would be a rightward shift of the excess sugar supply function in Figure 4.10b, from ES, to ES 2 , implying a fall in the world market price for sugar, from P wl to P w2 . The implication is that the partial liberalizing of the US sugar program within the context of an FTAA, could in fact depress the world market price of sugar. This conclusion differs from previous studies of the likely effects of a partial or full liberalization of the US sugar policy which suggested that such action would increase the world sugar price (Atlas, 1996; Marks, 1993; Schmitz and Christian, 1993). Welfare implications The welfare implications of the effect of a change in the US sugar policy within scenario I, can be deduced from Figures 4. 10 and 4. 1 1 . Notably, the producers of sugar in the US would lose, while the producers of HFCS would gain substantially. The producers of sugar in the US would lose as a result of lost revenue or income owing to eut back in output and a reduction in the domestic price of sugar. The producers of HFCS however, would gain from the lower cost of production and the increased market access within the framework of the FTAA. Furthermore, assuming that such industries manage to maintain

PAGE 108

97 some of their market power, then their gains would be even larger than that associated with the traditional producer surplus, since they would be able to obtain economic rent. In the case of the sugar exporting countries in general, the producers would lose as the cheaper HFCS displaces some of their domestic market and with little hope of increased access to the US sugar market. Moreover, revenues from sales on the world market would decline as a result of the fall in price. Consumers on the other hand would in general benefit from the lower all round prices. The consumers in the US would gain considerably as both the prices of HFCS and sugar fall. Contrary to the expectation from conventional trade theory which implies that the consumers in the exporting countries are usually made worse off as a result of the opening of the country to trade, in this case the opposite occurs as the users of HFCS in the US will benefit from the reduction in prices associated with external economies of scale. Within the context of the FTAA, the major losers would be the relatively high cost sugar industries of the small CARICOM countries. Firstly, they would be losing their US import quota rent. Secondly, due to the permanent displacement of a portion of the sugar market, it is likely that there would be a much smaller US market for the sugar produced by these countries due to the fact that the cost of production within these countries exceeds the current US domestic support price of approximately 1 8 cents per pound of raw sugar. For example, production cost in Jamaica is reported to be US 33 cents per lb. (Caribbean Update, 1998). Also of consequence, is the fact that three of the world's most efficient sugar producers; Brazil, Colombia and Guatemala would be included as members of the FTAA and the region as a whole is a net supplier of sugar.

PAGE 109

98 Thus, these low cost sugar-producing countries could easily supply any additional import demand that might become available. At present, Guatemala supplies sugar~in addition to its TRQ allocation~at the world market price to the US. This sugar is refined and reexported as sugar or in sugar products~an indication of the competitiveness of Guatemala's sugar industry. Thirdly, as was mentioned in chapter one in the section outlining the problem, there is the likelihood that the CARICOM countries in becoming members of the FTAA may have to give up their lucrative preferential EU/ACP sugar quota. This situation comes about in the wake of a possible violation of the terms and 19 conditions of their Lome Convention . If indeed this were the situation, the consequence for the CARICOM sugar industry would be catastrophic. Finally, the governments of the countries would lose revenue that they would have obtained from tariff on sugar imports. Scenario IIComplete Liberalization of the US Sugar Program Trade implications A second but more unlikely scenario to be considered is the complete liberalization of the US sugar program. Figure 4.12 represents the usual large country trade analysis and is similar to Figure 4. 1 0. A major difference here however, is an elimination of import restrictions in the US sugar market. While the exact outcome is somewhat ambiguous, the analysis suggests that the world market price of sugar would remain more or less at the current level~i.e. no substantial increase and a strong 19 Lome Convention is a contractual non-reciprocal trading arrangement between African, Caribbean and Pacific (ACP) countries and The European Union (EU). Prices received by ACP members for export of sugar to the EU are on average three times that of the world market price. However, Article 1 74 Sec. 2a of

PAGE 110

99 possibility of a slight decrease. The actual analysis leading up to this result is as follows: With the opening of the US sugar market under an FTAA the world market price of sugar rises from the pre-liberalization price of P wl to P w2 ? me free trade market price (Fig. 4. 12a). At this juncture, sugar now competes directly with HFCS with the potential to regain some of the sweeteners market share. This is shown in Figure 4.12a as the potential for the US consumption of sugar to increase from X2 to X3, which would imply that both crystallized and liquid sugar are consumed. However, the increased demand for HFCS in the non-US countries (ROW) and the realization of industry external economies of scale cause the price of HFCS to fall (Fig. 4.1 1). This, as discussed earlier, results in a downward shift in the US sugar demand curve from Dsi to Ds2 , with the accompanying shift in the excess sugar demand curve, EDi to ED2, in Figure 4.12b. At the same time, the increase in demand for HFCS displaces some of the sugar demand in the non-US countries (ROW) causing their demand function to slightly rotate counterclockwise while shifting to the left, from RWDsi to RWDs2 , as previously discussed. The increased opportunity to export sugar to the US also causes some of the more efficient sugar producers in the FTAA to increase their supply of sugar, resulting in the ROW sugar supply curve shifting to the right, from RWSl to RWS2. The combination of the leftward shift in the demand curve and the rightward shift in the supply curve in the ROW result in a substantial rightward shift in the excess sugar supply curve, from ESi to ES2 in Figure 4.12b. A new free trade price is established resulting from the intersection of ED2 and ES2 . As drawn in Figure 4.12b this new price Pw3 is the same as Pwi , the preliberalized price. While this might not be the exact outcome, the point to be made is that the Agreement states that in order to be eligible for the benefits developing countries are obligated to provide no less favorable treatment to the European Union than they provide to any developed country.

PAGE 111

100 owing to the presence of HFCS in the market, and the fact there is a strong possibility that it can compete with sugar at world market prices, there is no guarantee that the elimination of the US sugar program will necessarily result in an appreciable rise in the world market price of sugar . The exact outcome will be determined, among other things, by the extent to which the importing countries open up their market and the degree of the penetration achieved and sustained by HFCS. Additionally, it will depend on the long-run ROW export supply price elasticity. Marks (1993) noted that in general the lower the ratio of net exports to production, the higher should be the elasticity of export supply. Price Price Price XI X2 X3 X4 Quantity Quantity Quantity Fig. 4.12a. US Sugar Market Fig. 4. 1 2b. Trade Diagram Fig. 4. 1 2c. ROW Sugar Market Fig. 4.12. Trade implications under completely liberalized US sugar program

PAGE 112

101 Using his estimate of 13.12 as a rough guide, the implication is that the opening up of the US market should not seriously affect the world market price since a 100 % increase in the quantity imported by the US would only result in an approximately 7.6 % increase in the world market price, ceteris paribus. Figure 4.12a also indicates a final outcome in which, it is possible that both liquid sugar and HFCS are consumed with sugar regaining some of the market share that was lost to HFCS. One can envisage a situation similar to that which exists in Canada, where HFCS is still consumed in an environment where there is no special protection given to HFCS and sugar prices reflect those of world market prices. Thomas (1985) points out that in spite of such a situation, HFCS consumption captured approximately one-third of the sweeteners market in Canada. Welfare implications The welfare effects of scenario II are similar to those that occurred in the partial liberalization case under scenario I, although the magnitude of the benefits and losses will differ considerably. Noticeably, the sugar producers in the US are expected to experience considerable loss, while the gains to the HFCS producers would be less than in the previous case. This latter situation will result since HFCS can be expected to lose some of the market share as it competes with liquid sugar. The situation with regards to producer in the rest of the world, including the FTAA is not clear, but the indications are that they would again suffer losses or at best be made no worse off than previously. There would however, be a noticeable redistribution of the benefits and losses among the producers, with the relatively high cost CARICOM sugar producers being unfavorably disposed.

PAGE 113

102 The main beneficiaries would be the consumers in both the US and the rest of the world. The global benefits could be expected to exceed the losses.

PAGE 114

CHAPTER 5 EMPIRICAL MODEL OF THE US DEMAND FOR HFCS The preceding chapter outlines the conceptual model of the US sugar and HFCS subsectors. This chapter focuses on the estimation of the derived demand for HFCS. The first part deals with the specification of the model and gives a description of the variables and data used. In the latter sections the results of the demand estimation and the model implications are discussed. Empirical Model In keeping with earlier discussion of a derived demand function (Chapter 3) it is postulated that the demand for HFCS is a function of the price of HFCS, the prices of substitutes, and the prices of other inputs used in the production of the final HFCScontaining goods. The total amounts of HFCS-42 and HFCS-55 utilized (millions of short tons dry weight equivalent) were used to represent the quantity of HFCS demanded. This amount reflects the total US domestic disappearance—total supply less exports and stock of HFCS. The HFCS-42 price (cents per pound dry weigh) was used as a proxy for the price of HFCS. Using this series was necessary in light of the incompleteness in HFCS55 price series. This should not pose a serious problem, since both syrup prices tend to be highly correlated. The US wholesale beet sugar price series (cents per pound) was chosen to represent the price of substitutes. Wholesale refined sugar prices rather than the wholesale raw sugar prices were used, since the former is more readily comparable to the 103

PAGE 115

104 price of HFCS on a sugar equivalency basis. While other corn sweeteners such as dextrose and glucose can be substituted for HFCS, in general owing to their much higher price per unit of sweeteners there is limited substitution amongst these products. For reasons discussed earlier, the non caloric (artificial) sweeteners were excluded from the analysis. Consequently, only the price of sugar was included in the equation to capture the effects of substitutes. With respect to the final HFCS-containing products, only the price of soft drinks was used. Although there are several such final products spanning several industries, the models were limited to only the price of soft drinks, since this product utilizes the bulk of the HFCS produced. In addition, sufficient information was not available to compute a weighted index across the other products. Given the high degree of pair-wise correlation among these indices, inclusion of more than one of these indices in the equation could give rise to multicollinearity problems. While the soft drink index is not completely representative of the range of final products, it was used as a proxy for those unrepresented final products. The final variable included in the derived demand equation is the price of flour. Flour was used since a fair amount of HFCS is utilized in the baking industry. The inclusion of this index represents a proxy for the other inputs used in the production of final HFCS-sugar containing products. Finally, all prices in the equation are expressed in real terms using the GDP deflator. The final model is expressed with a double log specification representing the derived demand equation: LQhi = Po + PiLRPh, + p 2 LRP s , + psLRPot + p 4 LRP Ft + e, (5.1)

PAGE 116

105 where LQ Ht is the log of the annual quantity of HFCS used in period t; LRP Ht is the log of the real price of HFCS; LRP St is the log of the real price of refined sugar; LRPd, is the log of the real price of soft drinks; LRP F , is the log of the real price of flour; and d is the error term. Of major concern is the dynamic path of adjustments of the coefficients over time. As discussed in Chapter 3, equation (5.1) can be expressed in the state-space framework, which then allows for estimation, prediction and signal extraction to be carried out using Kalman filters and a smoothing algorithm. In the state-space framework and allowing for a stochastic trend and timevarying parameters equation (5.1) can be represented as follows: LQ Ht = \k + PitLRPHt + 02tLRP s , + pBtLRPo, + p 4 tLRP F « + e, (5.2a) u t = Um + Ym + n, (5.2b) Y« = Yt-i + & (5.2c) Pit = Pi(t-i) + co it fori =1,2,3,4 (5.2d) where, ji, is the stochastic trend, and et, r| t , % t and tOi t are all disturbance terms assumed to be distributed normally and independently of each other, with zero mean and constant variance. Equation (5.2a) is the measurement equation and equations (5.2b 5.2d) are the transition equations. In matrix notation the above set of equations could also be represented as: Y, = (1 0 Z,')a, + e, (5.3a)

PAGE 117

106 a, ft "1 1 0 0 0 0" *7. Yx 0 10 0 0 0 A t-\ Pn 0 0 10 0 0 Aa-i) + P* 0 0 0 10 0 Put-i) ">* P* 0 0 0 0 10 A J 0 0 0 0 0 1 Pm-\)_ [A] [B] [C] [D] (5.3b) where Zt represents in this case a 4x1 vector of explanatory variables; the a t is the state vector to be estimated; p t denote the stochastic trend as defined in (5.2b) above, and p t s are the timevarying coefficients included as part of the state vector; and Y corresponds to the vector of LQ t . Equations (5.3a & 5.3b) are the measurement and transition equations, respectively. The co variance matrix of the disturbance in (5.3b) is o E 2 Q where Q= diag {0*,&z,&L>&L>°L>&L } and
PAGE 118

107 linear form or a suitable approximation of a linear form and then test such assumptions using test procedures such as the Ramsey RESET or the Box-Cox. The latter approach was used to test the linear specification versus the log-linear specification. The results of the procedure showed that the likelihood function was maximized with the value of X = .01. This indicated that the log-linear function was superior to the straight linear form and provided the criterion for choosing this particular functional form to represent the derived demand A second issue relates to the exogenity of the independent variables. This issue becomes important because it assist in determining the appropriate estimating technique to be used and hence the validity of the results. For example, in applying OLS to estimate a single equation, the assumption is made that all the independent variables are exogenous to the system (Darnell, 1994). If such is not the case then it poses an estimation problem since, in this case, the disturbance term in the equation would become a determinant of both the price and quantity. In other words, a right-hand side variable is correlated with the disturbance. In standard econometrics this can be handled in one of several ways, including the use of instrumental variables, two stage least squares (2SLS) or obtaining the reduced form of the price variable. Similarly to the case of the OLS regression, one of the requirements for utilizing the state-space model with a single equation is that the independent variables are exogenous to the system being estimated (Harvey, 1987; Engle and Watson, 1987; Darnell, 1994). Owing to concerns surrounding the price of HFCS, a test of exogeneity (Granger causality test) was performed on the quantity of HFCS demanded and the price of HFCS. The results confirm the suspicion that the HFCS price was not exogenous. The null

PAGE 119

108 hypothesis that the quantity of HFCS demanded did not influence the price of HFCS could not be rejected. This could be interpreted that although the users of HFCS are price takers, there is a feedback mechanism, whereby the quantity demanded influences the price set by the suppliers. Viewed in this way, it meant that even though the price of HFCS is not operating within a perfectly competitive framework and does not adjust to ensure that the market is cleared within each period, it is nonetheless responsive to the level of excess demand, which is given by the difference between the actual quantity demanded and the actual quantity supplied. In order to address these issues and to facilitate use of Kalman filtering, it was necessary to obtain a reduced form of the price variable. This meant specifying an approximation of the way in which prices of HFCS were formulated in terms of a set of exogenous variables that are consistent with our earlier notion of the HFCS suppliers' response function(discussed in the previous chapter). This was done by first regressing the price of HFCS on all of the exogenous variables in the system. In a sense the approach adopted can be viewed as carrying out the first stage of a two stage least squares (2SLS) in which the endogenous variable is regressed on all of the exogenous variable in the system. The uncorrelated estimate of the price of HFCS could then be used instead of the correlated price variable in the second stage. In view of the profit maximization objective assumed for the supplier and the notion of a reaction function, three additional explanatory variables were introduced in carrying out the regression. This gave rise to the following price specification as our working hypothesis with regard to price formation: LRP-Ht = + iLRP C t + <1>2LRPe, + foLRI, + ^LRPs, + fcLRPo, + 6 LRP F , + v t (5.4)

PAGE 120

109 where, LRP H t is as defined earlier; LRP Ct is the log of the real price of corn; LRPet is the log of the real price of energy; LRI t is the log of the real long term interest rates; v t is the usual error term; and the remaining variables are as defined in the demand equation (5.1). The real price of corn (LRPct) was included since it constitutes the main input used in the production of HFCS. The price series for the yellow dent corn was chosen since this represents the variety most commonly used in wet-milling process. Consideration was given to using the net cost of corn starch, (i.e., the price of the starch after the returns from the major byproducts have been netted out) rather than the gross price of the corn. However, econometrics experiments with both variables showed that the price of corn was the better predictor of the price of HFCS. The cost of energy (LRPeO was included because of the highly capital intensive nature of the industry. Energy cost is the second highest variable cost used in the production of the product. Finally, long-term real interest rate (LRI t ) was included to reflect the cost of borrowed capital. This was computed as the difference between the long term nominal interest rate and the inflation rate. Equation (5.4) was estimated by OLS and the predicted value of the dependent variable LRPHt* was obtained and used as a proxy for the original HFCS price variable. The respecified derived demand equation which utilized the Kalman filter is: LQ Ht = H*+ PitLRP Ht * + p2,LRP s , + p3tLRP SD t + p4tLRP Ft + e, (5.5) A third issue that arose had to do with supplying the values for the hyperparameters and the initial values of the state vector within the framework of the state-space representation (See Chapter 3, section on Kalman filter). In situations in which the initial values of the state vector are not provided by the researcher, the first k (the number of elements in the state vector) observations are regressed using OLS and the

PAGE 121

110 parameters obtained are used as the initial values. In the case of the hyperparameters the identity matrix is usually assumed if the initial information is not provided. In the current exercise the information for the initial transitional covariance matrix was supplied by setting it equal to the covariance matrix obtained from estimating equation (5.5) by means of OLS. Once the initial values have been supplied, the Kalman filter generates the likelihood function which is evaluated in terms of the prediction errors. The function is maximized numerically with respect to the relative variances of the state vector. Subsequently, a smoothing algorithm is applied to obtain the optimal (minimum mean square) estimates of the state vector. In this case the fixed-interval smoother was used. Description of Data Annual data covering the period 1977 to 1998 were used in the estimation of equations (5.4) and (5.5). The choice of the period was limited by the fact that the HFCS industry began operations in the early 1970s with commercial production of HFCS-55 commencing in 1977. Data on the prices of sugar, HFCS, corn and corn starch, and the quantities of HFCS were obtained from various issues of the United States Department of Agriculture (USDA), Economic Research Services (ERS), Commodity Economic Division, Sugar and Sweeteners Situation and Outlook publications. The consumer price index for carbonated drinks, the price index for flour, the index for energy (electric power and natural gas utilities) and the GDP price deflator were all obtained from the Bureau of Labor Statistics (BLS). Price index for flour and the index for energy were taken from the BLS producer price index revision-current series. As indicated earlier, the real long term interest rate was calculated as the difference between the 30-year (Moody's) corporate bond and the rate of inflation

PAGE 122

Ill (obtained from the GDP index). Information on the interest rates was obtained from the Economic Report of the President (1997). Estimation Results The remainder of this chapter discusses the empirical results achieved using the two-step estimation process, with equations 5.4 and equation 5.5. First, for simplicity equation 5.4 was estimated as a double log using OLS. However, the diagnostic results showed that it had a goodness of fit value, as reflected by the adjusted R 2 , of 0.63 with two of explanatory variables being significant at the 5% level. Moreover, the DurbinWatson value was relatively low (1 .09) indicating a possibility of the existence of serial correlation among the error terms. The equation was re-estimated, assuming the existence of first-order serial correlation, and the maximum likelihood technique was employed. The new equation resulted in an adjusted R 2 of 0.79 and the DurbinWatson value increasing to 1 .72. The prices of corn, sugar and the final good were all shown to have significant impacts on the pricing of HFCS in a manner consistent with a priori expectation. The results of both equations are shown in Table 5.1 as equations (5.4a) and (5.4b), respectively. With the information obtained from equation (5.4b), in particular the estimates of price of HFCS(LRP H t*), the Kalman filter was used to estimate equation (5.5). The values of the most recent set of estimates of the parameters together with their pvalues are presented in Table 5.2 20 . This Table shows that the trend (u, t ), price of HFCS and the Since the HFCS industry is one of imperfect competition the observation at any point in time t lies on the demand curve corresponding to the price at which the MC = MR. The use of time series data to estimate the derived demand curve implicitly assumes that all such points lie on a given demand curve. However, in

PAGE 123

112 Table 5.1. Estimation of Price Equation Description Equation 5.4a-OLS Estimates Equation 5.4b~Maximum Liklihood Estimates Parameters Variables Coef. Std. Error pvalue Coef. Std. Error pvalue 4b Intercept 0.7395 1.4951 0.628 0.2797 0.9267 0.763 i Price of Corn (LRPq) 0.2873 0.4150 0.499 0.6347 0.3310 0.055 Price of Electricity(LRP Et ) -2.0969 0.9381 0.041 -1.0939 0.8579 0.202 Interest Rate (LRU -0.4680 0.5829 0.435 -0.2425 0.4541 0.593 Price of Sugar (LRP St ) 0.5301 0.4065 0.212 0.5772 0.2522 0.022 Price of Soft Drink (LRPc) -1.2976 1.0847 0.250 -1.1964 0.8114 0.140 Price of Flour (LRP Ft ) 6.1545 2.3772 0.021 3.5046 1.8791 0.062 Adjusted R 2 0.62 0.79 Durbin-Watson Statistic 1.09 1.73 rho 0.70 prices of the soft drinks (final goods) were significant at the 10% level. The signs on the trend and HFCS price variable were consistent with expectations. In the case of the other three variables no a priori assigning of the signs was possible given the nature of the derived demand curve and the fact that the output and substitution effects do not Table 5.2 Final Estimates Using Kalman Filter Over the Period 1977-98 Parameter Variable Coefficient Standard Error pvalue Absolute Beta Coefficient M< Intercept 7.97624 4.47635 0.075 Pit Price of HFCS (LRP») -1.63964 0.88958 0.065 2.26979 Pa Price of Sugar (LRPst) -0.43357 1.30173 0.739 0.87828 P* Price of Flour (LRPpt) -0.98789 1.82528 0.588 2.80602 P« Price of Soft Drink (LRPq,) 5.34171 2.77963 0.055 23.10574 view of the dynamic nature of the HFCS industry this is highly unlikely. Consequently, caution should be exercise with interpreting the final estimates in Table 5.2 as representative of the derived demand curve.

PAGE 124

113 necessarily act in unison. Since the double log formulation was used, the values of the coefficients also represent the elasticities. The estimate of the own-price elasticity of HFCS was -1 .64 implying that the demand for the commodity was relatively elastic. In the case of sugar the cross price elasticity was inelastic but not significant at any reasonable level of confidence. While the price of flour was non significant at the 10% level the price of soft drink was significant at this level. The last column of Table 5.2 shows the absolute values of the Beta (standardized) coefficients. These values give an indication of the relative importance of the independent variables in the model and allow for direct comparison (Pindyck and Rubinfeld, 1991). Thus, with respect to the price of HFCS the value implies that a one standard deviation change in the price of HFCS leads to a 2.27 standard deviation change in the quantity of HFCS demanded. The values also reveal that among the explanatory variables chosen the price of soft drink was the most important in explaining the relative variation in the quantity of HFCS demanded. Attention is now turn to an examination of the implications of these estimates on the time path adjustments for the selected parameter of the trend, own-price elasticity of HFCS, and the cross-price elasticity with respect to sugar. Model Implications The state-space and the Kalman filter allow for the modeling of time-varying parameters and for stochastic movements in the underlying trend variable. Addressing an observation made by Engle and Watson (1987) that timevarying parameters when used in most studies are typically interpreted as tests for the stability equation, the emphasis here is on the economic importance of such variations. As such the concern is not so much with whether the estimated parameters are stable and represent long-term

PAGE 125

114 relationship, but rather whether any important economic interpretation can be associated with their movements through time and the implications of such movements. In this regard, discussion is therefore focused on the trend, own price elasticity of HFCS and cross-price elasticity of HFCS with respect to sugar. To assist with the interpretation the respective values of the parameters are plotted against time, ignoring the first few observations which were used to derive the starting values. Thus, Figures 5.1, 5.2 and 5.3 show time paths of adjustments for the stochastic trend (n,), own price elasticity (p h ), and the cross-price elasticity (p 2t ) for the period 1982-1998, using the smoothed estimates (Appendix). In the case of Figure 5.1 the graph is indexed on the coefficient for 1982. Intercept Adjustment Over Time (m ) The dynamic path of the underlying trend in the utilization of HFCS is shown in Figure 5.1. The path is nearly linear in the earlier stages, followed by an exponential growth from the late 1 980s until about 1 996 and then a slow down in the growth pattern in subsequent years. The observed pattern is very much in keeping with what is known about the industry and support the hypotheses advanced in Chapter 1. HFCS was commercialized in the 1970s and despite being sold at a discount to sugar its adoption by many of the users was gradual. The early slow rate of adoption was due in part to the general skepticism which would have surrounded the use of a new product, as well as the existence of switching costs —the cost to the users associated with developing new recipes and modifying existing ones. A period was needed for users to learn more about the product, to determine its compatibility with the manufacturing process, and to formulate new products as well as to assess consumers' response.

PAGE 126

115 Years Fig. 5.1. Dynamic path of HFCS intercept overtime, 1982-98 The exponential growth beginning toward the end of the eighties and continuing to about 1996 is consistent with the fact that by the late 1980s all the major soft drink producers had switched from using liquid sugar to HFCS. In addition, it also reflects changes in productivity resulting from technical progress and increased capital stock among the users of the product. It is well known that the producers of HFCS work in concert with the users to modify the product so as to improve its acceptability. With the general up turn in the state of the US economy beginning around 1992 there has been an increase in the HFCS demanders (increased capital stock) resulting from an increase in the demand for final HFCS-containing commodities. Figure 5.1 reveals however, that

PAGE 127

116 since around 1996, there appears to have been a slow down and a leveling off in the growth pattern of the demand for the product. This finding is important from two perspectives. First, it corroborates the suggestion made by some industry experts that within the framework of the current technology, the extent of the substitution between HFCS and liquid sugar has been completed. Secondly, the finding has wider implications with regard to the future of the HFCS industry and its expansion within a wider trading environment occasioned by the proposed formation of the FTAA. It also support the earlier conjecture with regard to possible rent seeking behavior on the part of the suppliers within a much more liberalized trading environment. This latter issue is discussed in greater detail in the next chapter. Own-Price Elasticity (p L ,) Figure 5.2 shows the time path adjustments for the own price elasticity of HFCS. In particular, the evidence suggests that the elasticity of the demand has been becoming less elastic at a decreasing rate. In other words, the demand for the product although still elastic, has become relatively less elastic changing from -2.2 to -1.65. Again, this seems plausible given the developments that have taken place on the supply side in differentiating the product from liquid sugar and making it more user-friendly. Recalling the sugar and HFCS competition within the framework of the Stackelberg model (Chapter 4), the observed pattern would be consistent with the theory which suggests that by differentiating a product to such an extent where it is perceived as completely different from its competition, it is possible to drive the competitor out of the market. This has been the case to some extent with the HFCS-liquid sugar competition. Judging from statements made by some of the industry experts, HFCS is now considered to be a

PAGE 128

117 superior input to sugar in some applications such as in the coloring of certain baked products. Moreover, as pointed out by Marks (1993) many of the users, having been convinced of the attributes of HFCS and its favorable long-term pricing Years Fig. 5.2. Dynamic path of adjustment in HFCS own-price elasticity, 1982-98 prospects relative to sugar, have made investments in the necessary machinery and equipment for using and storing HFCS. Consequently, the responsiveness to a change in the price of the product is expected to fall since the importance of the product has increased and the degree of liquid-sugar as a substitute should have declined. Finally, the decline in the elasticity is consistent with our earlier suggestion of a possible maturity with regard to substitution possibilities. This is so since if one were to accept the

PAGE 129

118 argument, then further reduction in the price of HFCS should have less and less impact given the limitation of technology. Notwithstanding the above, the fact that the demand for the product is still elastic is consistent with the observed recent pricing of the product, whereby the price has been steadily declining since 1995. If the result had indicated that the demand was inelastic, then owing to the constancy of the elasticity associated with double log model, it would imply that the suppliers would be better off increasing their price in order to maximize their profit. However, with an elastic demand expenditures can be increased and market shares retained by reducing the price of the product. This is what has been taking place in the industry and could help to explain why the price of HFCS has been decreasing in recent years. Cross-Price Elasticity (p^ Table 5.2 reveals that the coefficient for the price of sugar was not significantly different from zero. The estimated coefficient implies that currently the effect of a change in the price of sugar is having no impact on the quantity of HFCS demanded. Although this result might seem somewhat surprising, it is nonetheless plausible if one were to accept the argument that the suppliers of HFCS have managed to differentiate the product vis-avis liquid sugar. And, that the extent of the possible substitution between HFCS and liquid sugar, within the confines of current technology, is complete. Given current restrictions on the price of sugar, the advances in technology which have facilitated the substantial reduction in the cost of production of HFCS, and the extent of the product development, HFCS seems to be operating in its own market as a separate and distinct product from liquid sugar. As a consequence, a change in the price of sugar is having

PAGE 130

119 little or no impact on the quantity HFCS being used. It now seems plausible that a considerable reduction in the price of sugar would be needed to impact on the quantity of HFCS currently being demanded Years Fig. 5.3. Dynamic path of adjustment of cross-price elasticity of HFCS with respect to sugar, 1982-98 This latter view is also consistent with the dynamic path observed in relation to the cross-price elasticity as shown in Figure 5.3. Figure 5.3 shows clearly that the adjustments over time has been in a manner such that the impact that a change in the price of sugar was having on the demand for HFCS has been declining. If so, the extent of the substitution of HFCS for sugar has been largely completed and the use of sugar

PAGE 131

120 within such industries can be view more as a complement (joint inputs) 21 rather than substitute commodity. The declining degree of substitutability is perfectly consistent with a product that eventually established its role as a reliable and safe alternative to liquid sugar. 21 Recall that for some industrial uses HFCS is inappropriate and sugar must be used. It is in this broad sense that the two products might be viewed as complements in the production of a representative industrial product, given current prices and level of technology.

PAGE 132

CHAPTER 6 SUMMARY, POLICY RECOMMENDATIONS AND SUGGESTIONS FOR FURTHER RESEARCH Summary and Conclusions This dissertation had two general objectives. The first objective was to qualitatively assess the potential economic impact that likely changes to the US sugar program, and an expanded High Fructose Corn Syrup (HFCS) industry, might have on the CARICOM sugar industry within the framework of the proposed Free Trade Area of the Americas (FTAA). The second objective was to provide quantitative estimates for the US HFCS demand configuration that can be used in subsequent modeling exercises. In chapter 1 the problem to be addressed as well as the hypotheses and specific objectives were elaborated. It was noted that the issues were complex and involved many players and institutions operating within a changing global policy environment. Only a subset of the problem was addressed in this study. The approach to the study was outlined. It involved developing a conceptual model of the US sugar and HFCS markets and assessing on the basis of this model the trade and welfare implications resulting from likely changes to the US sugar program and an expanded HFCS industry within the context of the proposed FTAA. Subsequently, the derived demand for HFCS in the US was estimated and used to explain likely concerns of the CARICOM sugar producers. The analysis began in Chapter 2 with the development of a detailed background on some of the key players and institutions involved. The complex nature of the problem 121

PAGE 133

122 on hand within a rapidly changing international agricultural policy arena was highlighted. Initiatives such as the WTO and FTAA were explored within the context of trade liberalization and an improvement of the global/regional benefits as well as their implied costs for some sectors. The CARICOM sugar industry was identified as one likely to suffer adverse consequences of the rapid changes occurring in the international agricultural policy arena. It was noted in chapter 2 that the proposed FTAA comprises a wide diversity of countries-some less developed such as those in the Caribbean, others developing such as Brazil and Chile and others which are developed such as Canada and the US. A likely consequence of this is that while the overall benefits to the region will improve as a result of integration, some countries will benefit more than others. Indeed, given the features of some of the smaller CARICOM countries, they could be among the losers~at least in the short-run. A detailed review was made of the trends in the world, and Western Hemisphere sugar and HFCS prices, production, consumption and trade, and the importance of the sugar industry to the CARICOM sugar producing countries discussed. In the 1 980s, there have been major structural changes in the world sugar market as a consequence of HFCS. Developing countries now dominated the import trade and their lower income, and higher price elasticities suggest fewer fluctuations in world market price (Hannah and Spence, 1997). A further implication is that the current relatively low world market prices of sugar can be expected to continue into the future. This change in the status of the market is attributed largely to the developed countries implementing policies, which fostered the growth of their domestic sweetener industry. A consequence has been the

PAGE 134

123 growth of large HFCS industries in the US, Japan and the EU. The US is the largest sugar importing country within the hemisphere and the world lowest-cost producer of sweeteners because of its highly efficient HFCS industry and relative abundance of corn. Currently, the US produces almost 75% of the world supply of HFCS. Even so, production capacities remain underutilized. Since 1980s there has been a gradual penetration of HFCS into developing countries, a trend which is likely to continue (Hannah and Spence, 1997). There exist a real possibility that the product could pose a formidable challenge to higher cost sugar markets in the region. The sugar industry of the C ARICOM sugar producing countries is of great importance to the subregion. It contributes to foreign exchange, employment and domestic activities and plays a vital role in the development of the countries through the maintenance and upgrading of rural infrastructure, health, recreational and education facilities. In Chapter 3, studies of the US sugar market that took into consideration HFCS in their analysis were reviewed and important issues with respect to time series analysis discussed. All of the studies used time series analysis, and there are always concerns when time series data are used to estimate economic parameters of interest. The impact of HFCS was considered indirectly through the substitution effect in previous studies and none included HFCS directly into their trade analysis. In general, all studies concluded that the world market price of sugar would increase if the US partially or completely liberalized its sugar policy. In the final sections of the chapter a description of the statespace methodology and the Kalman filter was presented This framework was used to estimate the derived demand for HFCS. Advantages of this approach were discussed and

PAGE 135

124 a justification given for use of the methodology in this study. This methodology: 1) allows for modeling of timevarying parameters, if the situation so warrants and provides estimates at any point within the sample permitting one to estimate the path parameters have taken over time; 2) deals fairly easily with nonstationary time series ; and 3) allows modeling to be done in level versus differenced form which facilitates interpretation. The flexibility of the state-space model makes it appropriate for estimation in situations, such as in the case of derived demand for HFCS, where there are indications that the industry is still evolving. The chapter was concluded with a brief theoretical discussion of the concept of a derived demand. A major part of the dissertation was concerned with constructing the conceptual model of the HFCS and sugar markets used in assessing the implications. This was done in Chapter 4. For convenience, the US demand for sugar was separated into the demand for crystallized sugar (CS) and liquid sugar (LS). The CS represents the demand for tabletop sugar plus sugar used in the industrial sector in processes in which it was not technically feasible to use HFCS. The LS represents the demand for sugar for which HFCS provides an almost perfect substitute. This kind of classification results in a kinked demand for sugar. At prices above the kink there is no effective demand for liquid sugar. The model is essentially static and therefore only provides a first approximation of the likely occurrence. Dynamics were inserted through a comparative static framework with regard to the interrelationship between the sugar and HFCS price. In modeling the HFCS market, the imperfectly competitive framework within which the industry operates was recognized. Consequently, it was argued that a supply function for HFCS did not exist and that the supply could be better explained in terms of a reaction function, somewhat

PAGE 136

125 similar to the situation in the Stackelberg model in which one player was sophisticated (HFCS producers) and the other was naive. The users of HFCS were assumed to be price takers. Suppliers established the best-response price, with consideration to the situation in the sugar market and their objectives of maximizing profit and/or maintaining market shares. Industry experts believe the long-term prospects of the HFCS industry appear more favorable than its sugar substitute. This belief played a central role in modeling the long-run situation in the HFCS industry. This view stemmed from the fact that the product is the outcome of biotechnology with prospects of further significant improvements in technologies, compared with sugar that is essentially a chemical product. A likely implication is that the cost of producing HFCS could decrease considerably as external economies are realized, providing real justification on the part of the corn refiners to actively seek expansion of the market for HFCS products. Theoretical Findings The impact of a partial liberalization of the US sugar program and a complete liberalization of the program within the framework of the FTAA were also considered in Chapter 4. The conceptual trade model utilized the typical large country trade assumptions and assumed that the effect of an expanded HFCS industry could be captured through its impact on the excess sugar supply and demand curves. The results of the first investigation revealed the importance of the kink in the sugar demand configuration. If in a situation of partial liberalization of the US sugar program, the US domestic support price were to be reduced to a level at which it was still above the kink in the sugar demand then, there would be an increase in the US demand for sugar and most likely an increase in the levels of imports. However, the increase in the demand for

PAGE 137

126 sugar would be marginal, since it would not affect the demand for HFCS~i.e., the increase in the demand would only be in relation to crystallized sugar. If on the other hand, the price were lowered to a level that initially fell below the kink, then this would set off a chain of events. Given declining cost in the production of HFCS and the current view that the HFCS producers earn economic rent, then to maintain market shares the price of HFCS could be lowered to such an extent to cause the effective demand for liquid sugar to fall below the kink. Any increase in demand for sugar in the US market would again have to be accommodated under the demand for tabletop (CS) sugar. Implications are that structural changes have occurred within the US market and sugar demand in the future will be limited further. When the trade considerations are imposed on the model, a somewhat startling result is obtained. The model suggests that within the context of a partial liberalizing of the US sugar program and the prospects of HFCS penetrating overseas markets, the world market price of sugar could fall rather than increase. This is due mainly as a result of a portion of the rest of the world (ROW) domestic sugar demand being displaced with HFCS and surpluses are forced on the world market. This revelation represents a radical departure from the general findings of most studies. The discrepancy is due to a failure of these studies to directly take into consideration the prospect of HFCS entering into trade in their analysis. For example, HFCS could easily displace a substantial quantity of the sugar used in Mexico's soft drink industry, with consequences of an increase in the world surplus of sugar. The welfare analysis associated with the first scenario suggests that whereas consumers in general and the industrial users of sweeteners, in particular would benefit a

PAGE 138

127 great deal, the higher cost sugar producers would suffer losses. Sugar producers in the US would lose as a result of the decline in the support price. However, the HFCS producers would gain as a result of a combination of increased sales and further decline in production costs. The CARICOM producers with their relatively high cost of production would be negatively impacted in competition with some of the more efficient sugar producers, such as Brazil and Guatemala, for a share of the US market. The losses they would experience could come as a result of a: 1) loss of US import quota rent; 2) possible loss of lucrative preferential EU/ACP sugar quotas; 3 ) loss of a portion of their domestic market; and 4) loss in earnings if the anticipated fall in world market price prevails. The theoretical findings in relation to the second scenario, the full liberalization of the US sugar program, proved inconclusive. They did suggest however, that the prospect of the world sugar price increasing substantially as a result of this decision was highly unlikely. At best one could expect a slight to moderate increase in the world price of sugar. Again, these results differ from those of most of the sugar studies reviewed, which indicated an appreciable rise in the world market price of sugar as result of such action. This apparent discrepancy is due to the failure of such studies to take into consideration the potential impact of trade in HFCS in a much freer environment. The theoretical model also suggests that with the complete liberalization of the US sugar program, relatively cheap imported sugar might be able to reclaim some of its lost market shares. Again, this conclusion rests on speculation, since it is known that the HFCS product is being differentiated to the extent that it is being regarded in some circles as a superior product to sugar. If indeed this were the situation, it would imply that the price of sugar would

PAGE 139

128 have to be at a considerable discount to HFCS to cause a reversal in the current sweetener pattern in the US. The welfare implications of the full liberalization scenario were somewhat similar to that of the partial liberalization, with consumers in general benefiting and the relatively high cost sugar producers being the losers. Producers of HFCS again are expected to benefit although to a lesser extent than in the case in which there is only a partial liberalization of the program. And, as in the previous scenario the relatively high cost CARICOM sugar producers would be the major losers. Empirical Findings The estimation of the US derived demand for HFCS was the only aspect of the conceptual model investigated empirically and was the subject of Chapter 5. The ongoing product development that has occurred in the market for HFCS, suggests that the coefficients of the derived demand should be estimated within a framework that allows for the estimation of timevarying parameters. The particular framework chosen was the state-space model and the Kalman filters which permits one to observe the dynamic path of adjustments taken by the varying parameter. In applying the technique it was necessary to utilize a two-step approach. The first step essentially involved placing the model in the state-space form and ensuring that all the explanatory variables were exogenous. The second step involved carrying out the actual estimation of the equation using the Kalman filter. The double log form was chosen to represent the form of the derived demand equation so the estimated parameters are the respective elasticities. In keeping with the focus of the dissertation, attention was only given to the dynamic path of adjustments with respect to the intercept, and the ownand cross-price elasticities.

PAGE 140

129 The empirical results clearly showed that structural drift has been an important component of the derived demand for HFCS over time. The intercept coefficient was significant, and indicated an asymptotic growth pattern which supported the view of some industry experts that the industry is now in a mature stage and has reached the limit of its technological applicability to various uses as far as the US market is concerned. The implications for the sugar interest and the CARICOM sugar producers suggest that producers of the product, barring any significant break-through in the crystallized market, now face a problem of actively seeking external markets for their product. This becomes even more crucial with the additional gains to be had from the realization of external economies within such an expanded market. All this could spark a further round of rent seeking on the part of the corn refiners as they actively seek to expand their market, with fewer barriers to trade. It also implies a further weakening of the US sugar-corn coalition lobby as the corn processors appear no longer interested in the protection offered by the sugar program at the expense of penetrating overseas markets. For the sugar producers in the US, the findings represent somewhat of a comfort to the extent that there might not be any further erosion of the market shares in the US sweeteners within the confines of the current technology. In this regard, it allows one to get an indication of the likely size of the sugar market in the US and by extension the size of the sugar import market. However, the tabletop (crystallized sugar) market still represents a substantial portion of the sweetener's market and could result in more aggressive research on the part of the HFCS producers to perfect their effort to produce a competitive crystallized HFCS product. With respect to the non-US sugar producers, such as CARICOM, the main concern of the findings and its implications will be the

PAGE 141

130 extent to which HFCS could penetrate the non-US market and displaces a portion of the existing sugar trade levels. As in the case of the intercept coefficient, the own-price elasticity was significant and elastic. The dynamic path of adjustments showed that the absolute elasticity had been decreasing over time, implying that the demand for the commodity has become relatively less elastic. This finding is consistent with the views of some industry experts that in light of the considerable amount of HFCS product development over the years, the product is establishing its own market and is being regarded in certain uses as a superior product to liquid sugar. An implication of this is that the shift away from liquid sugar which has occurred in the sweetener market in some uses, might be structural and that a simple reduction in the price of sugar might not cause a major reversal of the current pattern of sweetener used in the US. Unlike the other two coefficients considered, the cross-price elasticity of HFCS with respect to sugar was non significant. This may suggest that at present the change in the sugar price no longer has a significant impact on the demand for sugar. This however does not necessarily imply that at an earlier time period the impact of the sugar variable was not significant. Again this finding is consistent with the view that the HFCS is establishing its own market. The dynamic path of adjustments of this parameter estimate showed that the relative cross-price elasticity had decreased considerably over time. This suggests that over time the impact of a one-percent change in the price of sugar is having a smaller effect on the relative quantity of HFCS demanded. The above findings make the US sweetener market and how it will react less predictable with the liberalization of sugar and with the price of sugar falling to a level

PAGE 142

131 that is within the vicinity of the price of HFCS. The empirical findings therefore lend supports to some of the earlier findings of the conceptual model, with regard to uncertainty surrounding the possible outcome in the event that the US completely abolishes its sugar program. Policy Implications and Recommendations In general, the findings of this study confirm inter alia, our hypotheses (See Chapter 1). In particular, they supported the hypothesis that unqualified membership of CARICOM countries in the proposed FTAA, could in the short to medium term have potentially negative economic impacts on the subregion's sugar industry. Also, that an expanded US HFCS industry could pose a threat to the subregional sugar industry. It should be noted however, that the study did not specifically consider the competitiveness of the CARICOM sugar industry vis-a-vis other sugar industries in the region. This was not undertaken since it is generally accepted that despite the best efforts to modernize the CARICOM sugar industry it can never be as competitive as segments of the US sugar industry or those in countries such as Columbia, Brazil and Guatemala. As pointed out by Kennedy et al (1998), the limitation of land and other resources are severe constraints in CARICOM countries. A study conducted by LMC and reported by Haley (1998a) suggests the field cost of production in the US is comparable to that of some of the CARICOM countries. However, the US enjoys in some areas as much as a 40% advantage over non-US countries factory costs. There are however, significant areas in which the CARICOM sugar industry can improve its productivity and efficiency considerably (Caribbean Update, 1999, James, (personal interview, 1999)).

PAGE 143

132 As pointed out in Chapter 2 the industry is of paramount importance to the maintenance of social and economic stability. It provides valuable foreign exchange for which the shadow price exceeds the official rate and the earnings used to purchase valuable capital goods that the subregion must acquire from outside sources. The industry has a history dating back to the seventeen-century when the subregion produced the bulk of the world output (Hannah and Spence, 1997). A consequence of this tradition is that the subregion has acquired knowledge and skill in the cultivation and processing of the crop. Moreover, a substantial amount of investment has been amassed in the industry and fixity of these assets make it difficult to shift resources without creating serious social and economic dislocation (Haraksingh, 1997). The importance of the sugar industry to the CARICOM countries and its limitations in terms of competitiveness must therefore be taken into consideration in any discussion about integrating the subregion into a wider hemispheric free trade area. Negotiations cannot be based solely on overall efficiency and competitiveness. Rather, such considerations will have to be based on objectives such as those considered in the political economy literature. Such a body of literature (Helpman, 1995; Rausser et al, 1982) emphasizes distributional considerations as a means of explaining among other things, various trade polices of countries. This body of literature for example, views trade policy as a device for income transfers to preferred interest groups in a society or group of societies. Although this theme and approach was not explored in the main body of the dissertation, it is clear that this dissertation has a political economy aspect. As pointed out in Chapter 1 , a dilemma facing CARICOM is whether to integrate with the Americas within the framework of the FTAA or with the EU. The dilemma

PAGE 144

133 stems from the fact that the CARICOM sugar producing countries run the risk of losing their lucrative sugar market in the EU should they proceed with the decision to become part of the proposed FTAA. While it has been argued (Gonzales, 1995) that the benefits of participating in the EU/ACP program have not brought about the expected level of development to CARICOM and its sugar industry, the benefits realized are quite substantial. Also, while it is true that the accelerated pace of globalization and changes to the EU policy are reducing the magnitude of such benefits, their contribution to the social and economic stability is still of consequence. Moreover, such benefits when viewed in the context of the findings of this study assume an even greater level of importance. As the findings suggest, the CARICOM countries within the framework of an FTAA would be hard pressed to compete with the more efficient sugar producers for the US market. Furthemore, such competition would be carried out for a sugar market that has undergone considerable structural changes resulting in a decrease in its effective size. The situation becomes even more uncertain when cognizance is taken of the fact that in accordance with the provision of the NAFTA agreement, the US and Mexico's sugar markets are to become fully integrated by the year 2008. In addition, there is the potential threat from the possible penetration of HFCS in non-US markets in the region that would make it unlikely for the CARICOM sugar producers to successfully compete in those markets. For CARICOM, the situation could worsen especially if the domestic CARICOM sugar market also shrinks as a result of successful penetration by HFCS. Finally, even if the CARICOM countries were to modernize their industry to the extent possible and be guaranteed a portion of the regional market within the framework

PAGE 145

134 of the FTAA, it is clear that the price that they would receive would be less than that which they are likely to receive in the EU market. This is so because there are several low cost sweetener (sugar and HFCS) producers in the region, whereas in the EU the cost of beet production is relatively high. Consequent on the above, the CARICOM sugar producing countries need to push ahead with their current restructuring aimed at streamlining the industry to take advantage of production economies where such possibilities exist. This can be achieved to some extent by closing down unnecessary mills and cutting back production carried out on land deemed marginal for sugar cane production. In an interview with Mr. James, the chairman of the Sugar Association of the Caribbean (SAC), it was pointed out that this was an area in which there could be improvement. In some countries there are too many mills which are not being operated at full capacity. Of greater importance, the Lome/FTAA trade dilemma must be resolved through a tripartite discussion throughout the various stages of the new Lome and FTAA negotiation. It is clear that for the CARICOM sugar industry to survive, the retention of its EU/ACP sugar quota is essential. And, while undoubtedly the overall benefits to the region resulting from the formation of an FTAA will be increased, it must be borne in mind that the benefits and costs will be distributed disproportionately. Furthermore, the FTAA represents a group of countries at various stages of the development continuum. On this point Balassa (1969) states quite clearly that regional agreements which involve countries with dissimilar development levels usually offer fewer benefits to some members, since differences in productivity and efficiency are usually of major importance. It is known that the benefits of such agreements are usually greater where the

PAGE 146

135 trade flows are mutually complementary. While to some extent there is some complementarity of trade flows within the proposed FTAA, the majority of the prospective members are major producers and exporters of sweeteners. Any negotiated agreement to which CARICOM is a part should therefore include special provisions for the CARICOM sugar industry and accommodate the sugar arrangement embodied in the EU/ACP sugar protocol. The HFCS in the future, whether within an FTAA framework or not, is going to pose a formidable challenge to sugar as the dynamics of the sweetener market continues to unfold. Consequently, to the extent possible a third recommendation is that the subregional sugar industry must pursue an active program aimed at diversifying its end product as well as the uses of sugar cane. Limitations of the Study and Suggestions for Further Research The research undertaken centered on assessing the dynamics of the US sugar and HFCS sugar markets and on drawing of implications for the CARICOM sugar producers within the framework of the proposed FTAA. The following limitations of this study suggest areas for future investigation. First, although some attempts were made to explore the dynamics of the situation in the theoretical model and more so in the empirical estimation of the demand curve, the setting was essentially static. As pointed out elsewhere, static models can only provide useful summary statistics and indicate the direction of the outcome. Thus, while being a useful first step in an attempt to understand complex issues, fully dynamic models are required to capture the full effects and better represent the reality of the assumed changes. A Computable General Equilibrium (CGE) model which takes into consideration some of the issues raised in this study might offer a

PAGE 147

136 worthwhile approach in the future as more data become available. Consideration might also be given to developing a dynamic partial equilibrium net trade sugar and HFCS simulation model for the FTAA countries and the rest of the world (ROW). This study focused on only the sweetener sector. The inclusion of other sectors in an analysis could serve to better improve the understanding of the issues confronted. Such an approach requires additional data and time not available within the confines of this study. Only one aspect of the theoretical model was investigated empirically. Although this is in keeping with the specific objectives identified in chapter 1, a more complete model which facilitates testing of some of the theoretical findings is needed. This study did not estimate the kink in the sugar demand curve that played an important role in modeling the impact of the likely changes of the US sugar program. This too needs to be studied. The location of this kink is a function of the dynamics of the rapid economic and technological changes occurring with the sweetener market and should be studied in a dynamic framework. A game theoretical analysis of the behavior of the HFCS suppliers, may provide a clearer understanding of how suppliers might respond to a freer trading environment. A third shortcoming had to do with the exclusion of artificial sweeteners-noncaloric sweeteners such as saccharin, aspartame, acesulfame-K. The rationale given for their exclusion was that there is a certain amount of health related issues surrounding the use of these products. While to some extent this is true, within recent years various health authorities have cleared most of the products. Such sweeteners are increasingly being used in the so-called diet foods and soft drinks and are usually 200 to 300 times sweeter than sugar. While these sweeteners have not yet made significant inroads into the

PAGE 148

137 traditional sweetener market (Hannah and Spence, 1997), it is the view of these experts that as more research is undertaken with respect to these products, they will begin to seriously encroach on sucrose and HFCS, rather than remain confined to the diet market. At such time it will not be easy to exclude them from the analysis of the dynamics of the sweetener market. Finally, this analysis was severely hampered by the lack of data on the HFCS industry from two perspectives. Firstly, the industry is relatively new being in existence for less than three decades. A consequence is that there are only a limited number of annual sample observations available that severely constrain the methodology one can employ. Secondly, in view of the competitiveness of the sweetener market and the oligopolistic nature of the structure of the HFCS industry, important sensitive cost of production data and prices are proprietary and remain difficult if not impossible to access for public research purposes.

PAGE 149

APPENDIX STATE AND SMOOTEHED ESTIMATES STATE ESTIMATES YEAR INTERCEPT HFCS SUGAR FLOUR SOFT DRINKS 1977 2. 35060 -1. 15039 0 .54531 0. 36625 4. 37817 1978 2 . 35060 -1. 15039 0 .54531 0. 36625 4. 37817 1979 2. 35060 -1. 15039 0 .54531 0. 36625 4 . 37817 1980 2. 35060 -1. 15039 0 .54531 0. 36625 4. 37817 1981 2. 35060 -1. 15039 0 .54531 0. 36625 4. 37817 1982 1. 77204 -0. 75027 0 .31236 1. 01283 3. 21563 1983 2. 19728 -1. 04408 0 .47241 0. 69941 3. 84027 1984 2. 69823 -1. 58493 0 .87955 -0. 54212 5. 23126 1985 2. 93150 -1. 76100 1 .05752 -2. 34914 6. 68421 1986 2. 71913 -1. 89913 1 .26641 -2. 91761 6. 98483 1987 2. 71488 -1. 90004 1 .26872 -2. 92442 6. 98753 1988 3. 39063 -2. 30996 1 .45474 -2. 84288 6. 89725 1989 3. 54696 -3. 14583 2 .15802 -2. 66301 6. 80810 1990 0. 40229 -2. 37257 2 .45879 -4 . 31041 6. 26375 1991 0. 43293 -2. 36084 2 .43532 -4 . 19048 6. 18130 1992 1. 41812 -2. 33124 2 .12672 -4 . 10239 6. 47785 1993 1 . 94168 -1. 76743 1 .48854 -4 . 29419 6. 43354 1994 2. 58921 -1. 66569 1 .21797 -4 . 34319 6. 19456 1995 4. 72430 -1. 40368 0 .35185 -3. 49925 5. 94294 1996 7. 56107 -2. 34393 0 .32677 -1. 66712 5. 41541 1997 7. 50306 -1. 93869 -0 .01940 -1. 48805 5. 35799 1998 7. 97624 -1. 63964 -0 .43357 -0. 98789 5. 34171 138

PAGE 150

139 SMOOTHED ESTIMATES TEAR INTERCEPT HFCS SUGAR FLOUR SOFT DRINKS 1977 2. 35060 -1. 15039 0. 54531 0. 36625 4 . 37817 1978 2. 35060 -1. 15039 0. 54531 0. 36625 4 . 37817 1979 2. 35060 -1. 15039 0. 54531 0. 36625 4 . 37817 1980 2. 35060 -1. 15039 0. 54531 0. 36625 4. 37817 1981 2. 35060 -1. 15039 0. 54531 0. 36625 4 . 37817 1982 4 . 46769 -2. 22674 1. 10953 -2. 91161 6. 54083 1983 4 . 55585 -2. 14419 1. 01180 -2. 82031 6. 28829 1984 4 . 75492 -2. 09098 0. 90724 -2. 67334 6. 13299 1985 4. 97237 -2. 04130 0. 80051 -2. 51831 5. 99348 1986 5. 18549 -1. 99177 0. 69514 -2. 36385 5. 85107 1987 5. 40252 -1. 94281 0. 59014 -2. 23221 5. 74120 1988 5. 58922 -1. 89038 0. 49291 -2. 15580 5. 67193 1989 5. 76038 -1. 83814 0. 39991 -2. 08330 5. 60115 1990 5. 94295 -1. 77051 0. 28948 -2. 02052 5. 53505 1991 6. 32377 -1. 74007 0. 14978 -1. 85404 5. 50256 1992 6. 69173 -1. 70745 0. 01258 -1. 70599 5. 48071 1993 7. 04554 -1. 67621 -0. 11948 -1. 55618 5. 45034 1994 7. 38946 -1. 66450 -0. 23255 -1. 38316 5. 41328 1995 7. 71357 -1. 66559 -0. 32941 -1. 19581 5. 37920 1996 7. 93952 -1. 68225 -0. 38388 -1. 03380 5. 34817 1997 7. 96047 -1. 66338 -0. 40732 -1. 01011 5. 34521 1998 7. 97624 -1. 63964 -0. 43357 -0. 98789 5. 34171

PAGE 151

LIST OF REFERENCES American Sugar Alliance. Initiation of Section 302 Investigation on Mexican Practices Affecting HFCS, (Office of the United States Trade Representative) June 19, 1998. Anderson, B and J . Moore Optimal Filtering. Englewood Cliffs, NJ: Prentice Hall, 1979 Aoki, M. State Space Modeling of Time Series. 2 nd edn. New York: SpringerVerlag 1990. Atlas, M. "The Impact of Trade Liberalization by Major Importing and Exporting Countries on World Sugar Trade." Ph.D. Dissertation, University of Colorado, 1996. Balassa, B. The Theory of Economic Integration. London: George Allen & Unwin Ltd, 1969 Barros, A. R. "Sugar Prices and HFCS Consumption in the United States." Journal of Agricultural Economics, 43 (1992):64 -73. Bernal, R. The Integration of Small Economies in the Free Trade Area of the Americas. Paper prepared under the aegis of the Center for Strategic and International Studies (CSIS), Washington: Volume IX Study 1, February 1998. Bernirschka, M., W. Koo, and J. Lou. World Sugar Policy Simulation Model: Description and Computer Program Documentation. Fargo department of Agricultural Economics, Agricultural Experiment Station, North Dakota State University, 1996. Blanco, M. "ACP-EU Trade: Into the New Millennium." The Courier. ^(NovemberDecember 1997): 64-67. Borrell, B. and R. Duncan. "A Survey of World Policies," in The Economics and Politics of World Sugar Policies S. V. Marks and K. E. Maskus , eds., p. 15-48 Ann Arbor: The University of Michigan Press, 1993. Bryan, A. "Coping with the New Dynamics" in The Caribbean: New Dynamics in Trade and Political Economy, A. T. Bryan, ed., North-South Center, 1995. Buzzanell, P. U.S. Sugar Trade Relations with Latin America. Paper presented at the Third World Sugar and Sweeteners Conference, Sao Paulo, Brazil, October 1998. 140

PAGE 152

141 Caribbean Update. Kal Wagenheim: Maplewood. Vol 14 #11 Dec. 1998. Caribbean Update. Kal Wagenheim: Maplewood. Vol 14 #1 1 July 1999. Carman, F. "A Trend Projection of High Fructose Corn Syrup Substitution for Sugar." American Journal of Agricultural Economics, 64(1982):625-633. Cheung, Y. and K. Lai. "Finite-Sample Sizes of Johansen's Likelihood Ratio Tests for Cointegration." Oxford Bulletin of Econometrics and Statistics, 55(1993):3 13-28. Congress of US, The 1997 Economic Report of the President: Economic and Budget Outlook: Hearing Before the Joint Economic Committee, Congress of the US, One Hundred Fifth Congress, First Session. Washington DC: US GPO, 1997 Contrearas, A. "Agricultural Issues Relating to the 1999 World Trade Organization Multilateral Trade Negotiation". House Agriculture Committee Field Hearing, January 22, 1998. Cooley, T. and E. Prescott "An Adaptive Regression Model," International Economic Review, 14(1973):364-71. Darnell, A. A Dictionary of Econometrics. Aldershot: Edward Elgar, 1994 Engle, R. F. and M. Watson " A Time Domain Approach to Dynamic Factor Analysis and Mimic Models," Les Cahiers du Seminaire d' Econometric, 22,(1980): 109-22. Engle, R. F. and M.Watson "The Kalman Filter: Applications to Forecasting and Rational Expectations Model" In T.F. Bewley (ed.)Advances in Econometrics: Fifth World Congress. Cambridge: Cambridge University Press, 1(1987): 245-84. Flemming, J. Inflation, Oxford: Oxford University Press, 1976 Friedman, M. " Optimal Expectations and the Extreme Information Assumptions of Rational Expectations' Macro Models", Journal of Monetary Economics, 5 (1979): 23-41. Gemmill, G. The World Sugar Economy : An Econometric Analysis of Production and Policies. Michigan State University. Agricultural Economics Report No. 313, 1976. Gonzales, A. "Europe and the Caribbean: Toward a Post-Lome Strategy," in The Caribbean: New Dynamics in Trade and Political Economy, A. T. Bryan, ed., North-South Center, 1995. Group of Experts, Overcoming Obstacles and Maximizing Opportunities: Smaller Economies and Western Hemispheric Integration, http : //ncfap . org/ftaawtol. htm . 1998.

PAGE 153

142 Haley, S. "US and World Sugar and HFCS Production Costs". Sugar and Sweetener Situation and Outlook Report, USDA, Economic Research Service, SSS 223 May, 1998a. Haley, S. Modeling the US Sweetener Sector: An Application to the Analysis of Policy Reform. International Agricultural Trade Research Consortium (IATRC), Working Paper, # 98-5, 1998b. Hall, S., K. Cuthbertson, and M. Taylor. Applied Econometric Techniques. Hemel Hempstead, UK: Philip Allan, 1992. Hannah, A. and D. Spence. The International Sugar Trade. New York: John Wiley & Sons, 1997. Haraksingh, K. "The Future of Preferential Trade and Sugar." Caricom Perspective. 67(June 1997): 10-13. Harvey, A. "Application of the Kalman Filter in Econometrics," in T.F. Bewley, (ed.), Advances in Econometrics: Fifth World Congress. Cambridge: Cambridge University Press, 1(1987): 285-313. Harvey, A. Time Series Models (2 nd ed.), Cambridge, Massachusets: The MIT Press, 1993. Harvey, A. "Trends, Cycles and Autoregressions." Economic Journal 1 07(1 997): 192201. Harvey, A., S. Henry, S. Peters, and S. Wren-Lewis. "Stochastic Trends in Dynamic Regression Models: An Application to the Employment-Output Equation." The Economic Journal, 96(1986): 975-85. Helpman, E. "Politics and Trade Policy." NBER Working Paper No. 5309, Cambridge, MA, 1995. Hildreth, C. and J.P. Houck, "Some Estimators for a Linear Model with Random Coefficients," Journal of the American Statistical Association, 63(1968): 584595. International Sugar Organization (ISO). Quarterly Review. London, Various Issues James, K. "Prospects for the Caribbean Sugar Agro-Industry." Paper presented at the UWI Ag.-50 Workshop on Agriculture in the Caribbean Issues and Challenges, Trinidad and Tobago 16-23 August 1998. Johansen, S. "Statistical Analysis of Cointegrating Vectors." Journal of Economic Dyaamics and Control, 13(1988): .231-254.

PAGE 154

143 Johnston, J. and J. Dinardo. Econometric Methods, 4 th edn. New York: Mcgraw-HilL, 1997. Kalman, R.E. "A New Approach to Linear Filtering and Prediction Problems," Transactions ASME, Journal of Basic Engineering, 82(1960): 35-45. Kalman, R.E. and R. S. Bucy , " New Results in Linear Filtering and Prediction Theory," Transactions ASME, Journal of Basic Engineering, 83(1961): 95-103. Kennedy, P. A Guide to Econometrics. 4 th edn. Cambridge, Mass.: The MIT Press, 1998. Kennedy, P. L., R. Harrison, M. Piedra. "Analyzing Agribusiness Competitivenes: The Case of the United States Sugar Industry" International Food and Agribusiness Management Review, l(1998):245-57. Krugman, P. R "Increasing Returns, Monopolistic Competition and International Trade." Journal of International Economics, 9 (1979)469-79. Kwaitowski, O., P. Phillips, P. Schmidt, and J. Shin. "Testing the Null Hypothesis of Stationarity Against the Alternative of a Unit Root" Journal of Econometics, 54(1 992):. 159-78. Landell Mills Commodities (LMC) "Annual Review of the World HFCS Industry," in Sweetener Analysis, New York, NY: LMC International, December, 1998. Langham, M. and M. Mara. "More Realistic Single Equation Models Through Specification of Random Coefficients." Southern Journal of Agricultural Economics, 5(1973)161-66. Leu, G., A. Schmitz, and R. Knutson. "Gains and Losses of Sugar Program Policy Options." American Journal of Agricultural Economics, 69(1987):591-602. Lindert, P. H. and T. A. Pugel. International Economics. 19 th edn., Chicago: Erwin,1996. Lopez, R. A., and J. L. Sepulveda. "Changes in the US Demand of Sugar and Implications for Import Policies." Northeast Journal of Agricultural Resource and Economics. 14 (1985): 177 82. Lord, R. " Sugar: Background for 1995 Farm Legislation" US Department of Agriculture, Economic Research Service, Commodity Economics Division. Report #711 June, 1995. Lucas, R. "Econometric Policy Evaluation: A Critique," in The Phillips Curve and Labor Markets, K. Brunner and A Meltzer, eds., Carnegie Rochester Conference Series on Public Policy, Amsterdam: North-Holland, 1(1 976): 19-46. Mankiw, N. G. Principles of Economics. New York: The Dryden Press, 1 997.

PAGE 155

144 Marks, S. "A Reassessment of Empirical Evidence on the US Sugar Program," in The Economics and Politics of World Sugar Policies, S. V. Marks and K. E. Maskus , eds., Ann Arbor: The University of Michigan Press, p. 79 108 ,1993. McFadden, D. "Cost, Revenue nd Profit Funcions," in Production Economics: A Dual Approach to Theory and Applications, M. Fuss and D. McFadden, eds., Amsterdam: North-Holland Publishing Co., 1978 Njchols, D. F. and A.R. Pagan, " Varying Coefficient Regression." in Handbook of Statistics, Vol. 5 E. J. Hannan, P.R. Krishnaiah, and M. M. Rao, eds., Amsterdam :North-Holland, p. 413-49, 1983. Nicholson, W. Microeconomic Theory: Basic Principles and Extensions, 6 th edn., New York: The Dryden Press, 1995. Pindyck, R. and D. Rubinfeld. Econometric Models and Economic Forecasts. 3 edn., New York: McGraw-Hill Inc., 1991. Polopolus, L. and J. Alvarez. Marketing Sugar and Other Sweeteners. Amsterdam: Elsevier Science Publishers, 1991. Rausser, G., E. Lichtenberg, and R. Lattimore. "Development in Theory and Empirical Applications of Endogenous Governmental Behavior," in New Directions in Econometric Modeling and Forecasting in US Agriculture, Gordon C. Rausser, ed., Amsterdam: North Holland Publishing Co., 1982. Rendleman, M. and T. Hertel. Economy Wide Effects of the Sugar Program. Paper presented at American Agricultural Economics Association (AAEA), 1989 Rosenberg, B.'The Analysis of a Cross-Section of Time Series by Stochastically Convergent Parameter Regression," Annals of Economic asn Social Measurement, 2(1973):461-4. Schmitz, A. "The Free Trade Myth and the Reality of European Subsidies." Paper presented at the 12 th International Sweetener Symposium, US Sweetener Policy in the Farm Bill: Charting a Course for the Future, Washington DC(27 June), 1995. Schmitz, A., and L. Polopolus. "Alcohol Fuel Tax Credit Policy: Sugar, Corn and the Environment" in F. Casey, A. Schmitz, S. Swinton and D. Zilberman, eds., Environmental Regulation and Technology Adoption in Agriculture. Boston, MA: Kluwer Academic Publishers, (Forthcoming). Schmitz, A. and D. Christian. "The Economics and Politics of US Sugar Policy." in The Economics and Politics of World Sugar Policies, S. V. Marks and K. E. Maskus, eds., Ann Arbor: The University of Michigan Press, p. 49 -78, 1993. Shen, C, D. Hakes, and K. Brown. "TimeVarying Response of Monetary Policy to Macroeconomic Conditions" Southern Economic Journal 65(1999): 584-93

PAGE 156

145 Shim, J., and J. Siegel. Dictionary of Economics. New York John Willey and Sons, 1995. Sudaryanto, T. "The Potential Impact of Liberalized Trade Policies in the United States and in the European Economic Community on International Markets for Sugar." Ph.D. Dissertation, North Carolina State University, 1987. Tanyeri-Abur, A. "An Agricultural Sector Analysis of the United States Sugar Import Policy." Ph.D. Dissertation, Texas A&M University, 1990. Thomas, C. Y. Sugar : Threat or Challenge? Ottawa: International Development Research Center, 1985. Thompson, R. A Survey of Recent US Development in International Agricultural Trade Models. Bibliographies and Literature of Agriculture, No. 21, Economic Research Service, USDA, 1981. US Department of Agriculture (USDA). Economic Research Service, Commodity Economic Division, Sugar and Sweeteners Situation and Outlook Report. Washington, DC, Various Issues. US Department of Agriculture (USDA), Economic Research Service, Commodity Economic Division, Sugar and Sweeteners Situation and Outlook Yearbook. Washington, DC, Various Issues. US Department of Agriculture. Economic Research Service (USDA), Commodity Economic Division, "Free Trade in the Americas." Situation and Outlook Series, WRS-98-1, Washington, DC, November 1998. Ward R. and L. Myers. "Advertising Effectiveness and Coefficient Variation Over Time." Agricultural Economics Research, 31(1979): 1-11. Ward R. and D. Tilley. "Time-varying Parameters With Random Components: The Orange Juice Industry." Southern Journal of Agricultural Economics , 12(1980): 5-13. West Indian Commission. Time for Action: Report of the West Indian Commission (Second Edition). Kingston : University of West Indies Press, 1993. Wint, A. "South-South Integration." Paper presented a Seminar on New Global Trading Arrangement, organized by the UWI and Sponsored by the European Union. Kingston, Jamaica November 19, 1997. Yule, G. U. "Why Do W e Sometimes Get Nonsense Correlation Between Time Series?" Journal of the Royal Statistical Society, 89(1 926): 1 -64

PAGE 157

146 Zietz, J., and A, Valdes. "The Potential Benefits to LDCs of Trade Liberalization Sugar and Beef by Industrialized Countries." Weltwirtschaftliches Archiv 22 (1986): 439 -464. Zitt, M. " A Complex Invention Process: Industrial, Technological and Scientific Dynamics in HFCS History." Paper presented at the 1 2 th International Economic History Congress, Spain, 1998.

PAGE 158

BIOGRAPHICAL SKETCH Edward A. Evans, alias Gilly, was born on November 3, 1956, in Jamaica W.I. to Evroy and Elizabeth Evans. He graduated from St. Georges College, Kingston, Jamaica W.I., in 1974. He enrolled in the University of West Indies (UWI), St. Augustine, Trinidad and Tobago and graduated in 1979 with a Bachelor of Science (Hons.) in agriculture. In 1979, he was awarded a UWI scholarship to pursue a master in agricultural economics at the same university in the Department of Agricultural Economics and Farm Management (DAEFM). He graduated with his Master of Science (M.Sc.) degree in 1982 and taught for a short time at the UWI Extra Mural Studies Department and in the DAEFM. In 1983, he joined the staff of the Caribbean Community (CARICOM) Secretariat, Guyana, as an economist and worked there for a period of eight and one-half years conducting economic analyses in the area of international trade and trading arrangements. In 1991, he joined the staff of the Caribbean Agricultural Research and Development Institute (CARDI), Trinidad W.I., as a senior economist responsible for performing all economic analysis pertaining to the CARDI executed, and Canadian International Development Agency (CIDA)-funded Sheep and Goat Production and Marketing Project. In the fall of 1995, he enrolled in the Ph.D. program of the Food and Resource Economics Department at the University of Florida. He completed his course work with a GPA of 3.96 and was awarded the doctor of philosophy (Ph.D.) in 1999 with specialization in international trade and policy analysis. He has written extensively and has authored or coauthored more than 30 publications/presentations, many in the area of international trade. 147

PAGE 159

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosopl; Carlton i Distinguished Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. ax R. Langham Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. uU2 Ronald W. Ward Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation andjsjully-adequate, in scppe~and quality, as a dissertation for the degree of Doctor ofMuosophy./ jrt D. Emerson Professor of Food and Resource Economics I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. Andrew; Schmitz • Eminent Scholar in Food and Resource Economics

PAGE 160

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. favid A. Denslow Distinguished Professor of Economics This dissertation was submitted to the Graduate Faculty of the College of Agriculture and Life Sciences and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctonof Philosophy. December, 1999 Dean, College of Agriculture and Life Sciences Dean, Graduate School


38
As noted before, most of the international trade in HFCS occurs within North
America. Although the Canadian and the US-HFCS industries are difficult to separate,
from a custom point of view, Canada is a net exporter to US, which in turn is a net
exporter to Mexico. The growth in this trade is reportedly due to increases in the Mexican
demand (USDA, Situation and Outlook Report, 1998). However, owing to a dispute
concerning the legitimacy of the US-Mexican Side Agreement on trade in sugar under the
North American Free Trade Agreement (NAFTA), the Mexicans have recently (January
1998) placed a 100% tariff on imports of HFCS from the US. The issue is now being
addressed by both the WTO and the NAFTA dispute settlement bodies. The tariff has
retarded the growth in trade but the US has maintained current levels of exports to
Mexico. Total exports of HFCS from the US increased by 37 % in 1998 and now account
for as much as 6.2 % of domestic production compared with 3.7% and 4.7 % in 1996 and
1997, respectively.
Finally, even with the shipping problems of the liquid product, an increasing
amounts of HFCS are currently being shipped to markets such a Japan and Trinidad and
Tobago (USDA Situation and Outlook Report, 1997). Industry experts believe that
shipping problems are surmountable. Within CARICOM, Trinidad and Tobago is the
only country, which currently imports HFCS for use in its soft drink industry, however
there is the possibility that other countries might soon follow.


12
Methodology
The purpose of this study is to assess the potential economic impact that likely
changes to the US sugar program, and an expanded High Fructose Com Syrup (HFCS)
industry, might have on the CARICOM sugar industry within the framework of the
proposed Free Trade Area of the Americas (FTAA). This will be accomplished in two
stages. First, a conceptual model of the US sugar and HFCS markets will be constructed.
Although constructed within a static framework, the model will attempt to highlight some
of the dynamics between sugar and HFCS. Based on the model, the implications of
possible changes in the US sugar program, and the impact of an expanded HFCS industry
on non-US sugar producers will be discussed.
The second stage of the exercise will be to empirically model the demand for
HFCS in the US with a view of assessing the maturity of the industry, and drawing
implications for the various sugar concerns. The focus will be on the dynamic paths of
adjustments of various parameters of the estimated demand function. To facilitate such
an investigation a state-space modeling approach and the Kalman filter will be employed.
Among the advantages of utilizing the state-space form and the Kalman filter are that: 1)
it allows for modeling of time-varying parameters, if the situation so warrants, and can
provide an estimate at any point within the sample and hence a valuable way of
examining how a trend might have evolved; 2) it can deal fairly easily with nonstationary
time series which traditional econometric methods have difficulty in handling; and 3) it
allows modeling to be done in level form (rather than differenced) in situation where the
data is nonstationary thus facilitating interpretation of results. The flexibility of the state-
space model thus makes it an appropriate tool for estimation in situations, such as the


126
sugar would be marginal, since it would not affect the demand for HFCSi.e., the
increase in the demand would only be in relation to crystallized sugar. If on the other
hand, the price were lowered to a level that initially fell below the kink, then this would
set off a chain of events. Given declining cost in the production of HFCS and the current
view that the HFCS producers earn economic rent, then to maintain market shares the
price of HFCS could be lowered to such an extent to cause the effective demand for
liquid sugar to fall below the kink. Any increase in demand for sugar in the US market
would again have to be accommodated under the demand for tabletop (CS) sugar.
Implications are that structural changes have occurred within the US market and sugar
demand in the future will be limited further.
When the trade considerations are imposed on the model, a somewhat startling
result is obtained. The model suggests that within the context of a partial liberalizing of
the US sugar program and the prospects of HFCS penetrating overseas markets, the
world market price of sugar could fall rather than increase. This is due mainly as a result
of a portion of the rest of the world (ROW) domestic sugar demand being displaced with
HFCS and surpluses are forced on the world market. This revelation represents a radical
departure from the general findings of most studies. The discrepancy is due to a failure
of these studies to directly take into consideration the prospect of HFCS entering into
trade in their analysis. For example, HFCS could easily displace a substantial quantity of
the sugar used in Mexico's soft drink industry, with consequences of an increase in the
world surplus of sugar.
The welfare analysis associated with the first scenario suggests that whereas
consumers in general and the industrial users of sweeteners, in particular would benefit a


37
sugar imports from a level of 5.0 MMT in the seventies to current level of about 1.6
MMT (Hannah fe Spence, 1997).
The growth of the industry in the US is attributed to several factors including: 1) a
marketing environment with no restriction on supply; 2) advancements in technology;
and 3) the relative cheapness of the product in relation to sugar on a sweetness
equivalency basis. Since its inception in the US the product has sold at a discount to
sugar. Figure 2.3 shows a comparison of the real price of HFCS-42 and the real world
and US prices of refined sugar. The evidence indicates that the pricing of HFCS is now
much closer to that of the world market price of refined sugar.
Fig. 2.3. Real HFCS and refined sugar prices, 1982-98 (1992=100)
Source: USDA, ERS, Commodity Economic Division and Dept, of Commerce,
Bureau of Economic Analysis


128
have to be at a considerable discount to HFCS to cause a reversal in the current sweetener
pattern in the US.
The welfare implications of the full liberalization scenario were somewhat similar
to that of the partial liberalization, with consumers in general benefiting and the relatively
high cost sugar producers being the losers. Producers of HFCS again are expected to
benefit although to a lesser extent than in the case in which there is only a partial
liberalization of the program. And, as in the previous scenario the relatively high cost
CARICOM sugar producers would be the major losers.
Empirical Findings
The estimation of the US derived demand for HFCS was the only aspect of the
conceptual model investigated empirically and was the subject of Chapter 5. The ongoing
product development that has occurred in the market for HFCS, suggests that the
coefficients of the derived demand should be estimated within a framework that allows
for the estimation of time-varying parameters. The particular framework chosen was the
state-space model and the Kalman filters which permits one to observe the dynamic path
of adjustments taken by the varying parameter. In applying the technique it was necessary
to utilize a two-step approach. The first step essentially involved placing the model in the
state-space form and ensuring that all the explanatory variables were exogenous. The
second step involved carrying out the actual estimation of the equation using the Kalman
filter. The double log form was chosen to represent the form of the derived demand
equation so the estimated parameters are the respective elasticities. In keeping with the
focus of the dissertation, attention was only given to the dynamic path of adjustments
with respect to the intercept, and the own- and cross-price elasticities.


122
on hand within a rapidly changing international agricultural policy arena was
highlighted. Initiatives such as the WTO and FTAA were explored within the context of
trade liberalization and an improvement of the global/regional benefits as well as their
implied costs for some sectors. The CARICOM sugar industry was identified as one
likely to suffer adverse consequences of the rapid changes occurring in the international
agricultural policy arena.
It was noted in chapter 2 that the proposed FTAA comprises a wide diversity of
countriessome less developed such as those in the Caribbean, others developing such as
Brazil and Chile and others which are developed such as Canada and the US. A likely
consequence of this is that while the overall benefits to the region will improve as a
result of integration, some countries will benefit more than others. Indeed, given the
features of some of the smaller CARICOM countries, they could be among the losersat
least in the short-run.
A detailed review was made of the trends in the world, and Western Hemisphere
sugar and HFCS prices, production, consumption and trade, and the importance of the
sugar industry to the CARICOM sugar producing countries discussed. In the 1980s, there
have been major structural changes in the world sugar market as a consequence of
HFCS. Developing countries now dominated the import trade and their lower income,
and higher price elasticities suggest fewer fluctuations in world market price (Hannah and
Spence, 1997). A further implication is that the current relatively low world market prices
of sugar can be expected to continue into the future. This change in the status of the
market is attributed largely to the developed countries implementing policies, which
fostered the growth of their domestic sweetener industry. A consequence has been the


50
Nonstationary time series have been subjected to one of two detrending
procedures in order to make them stationary before subsequent analysis. The two
procedures are: 1) regressing the time series as a simple linear (or higher order) function
of time and then using the residuals as the detrended series; and 2) an approach which
involves using first (or higher) order differencing of the time series. The first method
works only if the series say Yt has a deterministic trend and is generated according to the
following equation:
Y, = a +pt + e, (3.1)
where etis white noise. If however, the time series is generated by say a random walk or
random walk with a drift as for example:
Yt = Yt_i + P + v, (3.2)
where vt is white noise, the first approach will not work. More formally, equations (3.1)
and (3.2) are referred to as trend-stationary process (TSP) and difference stationary
process (DSP) with a trend, respectively.
The second approach has had some measured success even though as Kennedy
(1998) points out some analysts have adopted it without any justification based on
economic theory. Within this approach, a variable is said to be integrated of order d,
written as 1(d) if it has to be differenced d times to be made stationary. A stationary
variable is integrated of order zero. Economic variables are seldom integrated of order
greater than two and if non-stationary they are usually 1(1) (Kennedy, 1998). Equation
(3.2) above represents a series, which is integrated of order one. Taking Yt.i from both
sides of the equation gives equation (3.3):
AYt = P +vt
(3.3)


65
Kalman filter can be used to address the concerns of critics of the REH, such as Friedman
(1979), with regard to the information-available assumption. This he states can be done
by extending Friedmans framework in which he advocates that given the true model yt
xtp + ut (u, is white noise) agents may sequentially update their estimate of the fixed true
parameter vector p as more information on (yt, Xj) becomes available (e.g. time-varying
parameter models). Hall et al (1992) suggested that this could be done by: 1) assuming
that agents have some prior information about P (at time t = 0); and 2) by allowing p to
vary stochastically. Thus, agents are not assumed to know instantaneously the true model
but they do use the information optimally (efficiently), hence making the information
available assumption of the REH much more plausible. They further point out that in
certain models, the Kalman filter can be viewed as mimicking a learning process by
agents as in the case of adaptive expectation where the adjustment parameter is updated
each period, based on new information. This he notes formalizes Flemmings (1976) idea
of a change in gear when forming expectations. He makes the point that while the
adaptive expectation has been shown to be optimal (in the sense of producing unbiased
forecasts) only when the data generation process is of the form of an integrated moving
average, IMA (1,1), or ARIMA (1,1,1), the Kalman filter is optimal under more general
conditions, and in fact produces minimum mean square estimators (MMSE) under the
normality assumption. Harvey (1987,1997) has pointed out several other advantages of
the state-space framework and the Kalman filter. These are summarized below14.
14 The approach allows unobserved components to be incorporated in a model and the Kalman filter
provides the means of estimating them by way of a likelihood function. Estimates of the unobserved
components can then be obtained by smoothing. It therefore allows the entire model to be set up in terms of
components, which have direct interpretation.lt provides a framework in which time-varying parameters
can be estimated. For example, a stochastic trend can be estimated in which the level and slope are allowed
to evolve over time. Moreover, the framework is flexible enough to facilitate estimation of those


102
The main beneficiaries would be the consumers in both the US and the rest of the world.
The global benefits could be expected to exceed the losses.


20
Table 2.3. Leading World Sugar Exporters and Importers, 1997-98
Exporters
Importers
Country
Quantity
(MMT)
% of World
Exports
Country
Quantity
(MMT)
% of World
Imports
Brazil
7.2
20.2
Russia
3.8
10.7
European Union
6.2
17.4
United States
2.0
5.6
Australia
4.6
12.9
European Union
1.8
5.1
Thailand
2.9
8.1
Japan
1.6
4.5
Cuba
2.3
6.5
Korea
1.5
4.2
World
35.6
100.0
World
35.6
100.0
Source: USD A, Sugar and Sweeteners Situation and Outlook Yearbook.
Owing to domestic government policies, significant changes have occurred in the
status of many of the major players in the market. For example, the European Union has
changed its status from being a net sugar importer through 1976 to becoming the worlds
largest exporter. In the United States, sugar imports have declined considerably from a
share of 20 percent of world imports in the 1970s to the current 5.6 percent. As noted by
Hannah and Spence (1997), the United Stateswhich until 1975 had been the largest
sugar importer by a wide margin-had slipped to second position during the 1975-80
period and to fifth position by 1994. Japanonce the worlds second largest importer of
sugarhas reduced its imports considerably, owing mainly to domestic support policy.
Table 2.3 however, does not reflect the significant change that has occurred in the
structure of the global sugar import market. The market was previously dominated by
developed countries but is now dominated by developing countries. Indeed, Hannah and
Spence (1997) reported that in 1994 the developing countries accounted for
approximately 68 percent of net imports, compared with only 25% in 1975. These
authors noted that this change in the structure of global import demand has implications
for the extent of the price volatility since, unlike the developed countries, these countries


15
conservation of natural environment for future generations. The process has five
elements: 1) liberalization of barriers to trade in goods and services; 2) elimination of
restraints on investments; 3) provisions of free labor movement for specialized workers;
4) harmonization of tax and monetary policies; and 5) establishment of supra regional
institutions to administer the arrangements and to engage in dispute resolution. Bryan
(1995) notes that the new integration process goes beyond the traditional concept of
countries simply extending reciprocal preferences to their trading partners. He cautions,
however, that while absolute reciprocity may be the ultimate goal of the hemispheric
trade liberalization, many of the smaller Caribbean economies simply cannot compete
with countries having more productive infrastructures and technologies. Stated
differently, it might not be possible for them to offer absolute reciprocity to industrial
countries at least in the short run, and there will be the need for some compromise on the
part of the more developed trade partners (Bryan, 1995).
To better appreciate some of the disparities among the countries, Table 2.1 shows
a relative ranking of the 34 countries that will comprise the FTAA with respect to the
selected indicators of population, land area and GDP. In general, the CARICOM
countries (underlined in Table) occupy the lower levels in all three cases. The subregion
accounts for only 0.67% of the Hemisphere's land area, 0.82% of its population and
approximately 0.2% of total GDP. The Table also reveals that the most populous and
highest income country is the United States and the least populous and lowest income
country is St. Kitts and Nevis. Barbados has the smallest land area while Canada has the
largest. What is not obvious from the Table however, is the fact that many


85
technology; and 3) specialized machinery. They further suggest that if scale economies
are substantial over a large range of output, then a likely result is that a few firms will
expand their operation in order to reap the scale economies, giving rise to oligopoly
market structure. As it is well known, firms in an oligopoly can influence prices, can earn
economic profit and can vary the level of competition. More importantly, the existence of
scale economies has been shown by authors such as Krugman (1979) to provide as solid a
basis for trade as is comparative advantage. This line of argument forms the basis of the
New International Trade Theories (NITTs).
Much of the above is true for the HFCS industry, which has benefited
substantially from path breaking gains in biotechnology and processes such as
fractionation technology. Indeed, there is the view that the HFCS industry represents the
successful marriage of two industries: the wet com processors and the enzymes industry.
In light of the above and the fact that the industry is still relatively young and continues
to benefit from technological improvements, the study assumed a long-run HFCS
industry average cost reflecting external economies as depicted in Figure 4.8 .
Noticeable is the shape of the cost curve (LRAC) implying possible gains from
increasing return to scale. The figure also shows two hypothetical demand curves Dus
and Dfta reflecting the possible increase in the demand for HFCS as a result of the
formation of a free trade area, emphasizing the possible gains to be made from the
lowering of the average cost. The assumption of the long run shape of the HFCS industry
cost function forms a crucial part of the subsequent analysis of the trade implications
within the framework of the proposed FTAA.


CHAPTER 3
LITERATURE REVIEW AND THEORETICAL BACKGROUND
Several studies have been done on modeling the US Sugar/Sweetener subsector
and drawing economic/welfare implications for various groups including domestic sugar
producers, processors, sweeteners' users, consumers and trading partners. Most of these
studies have either not taken into consideration the impact of HFCS in their analysis or
have only done so in a cursory manner. The first section of this chapter therefore
undertakes a critical review of those studies in which the HFCS industry was taken into
consideration. The review provides a basis for the conceptual model of the US sugar and
HFCS subsector developed in chapter 4. Because of the problem of time-varying
parameters in the HFCS industry and the need to provide more reliable parameters for the
HFCS industry, the second section of the chapter focuses on some critical time series
issues such as serial correlation, unit root and cointegration. The methodology of state-
space model and the Kalman filter, allows for estimation of time-varying parameters and
is employed later in estimating the derived demand for HFCS. Consequently, this
methodology is discussed in detail. The chapter ends with a discussion on the nature of a
derived demand and the general approaches used to estimates such functions.
Previous Empirical Studies
Sugar and HFCS Related Studies
One of the earliest studies to include HFCS in the analysis of the US sweetener's
market is a study conducted by Carman (1982). This study had as its objective the
39


49
Stationary
Stationary is an important aspect of a time series. Formally, an infinite sequence
of random variables is said to be strictly stationary if the joint probability function of any
finite subset is identical to any other subset (Darnell, 1994). Less formal, when a series is
generated by stationary process it tends to fluctuate around a constant level and there is
no tendency for its spread to increase or decrease over time. In other words, the mean of
say Yt, its variance, and its covariance with other Y values, say Yt.k, are invariant to time
and the covariance between say Y, and Yt+S is a function of s only, for all t. Hence an
immediate implication of a stationary time series is that it has a time independent mean
and is homoscedastic. Kennedy (1998) points out that although many scientific time
series data are stationery, most economic time series data are nonstationary since they
tend to exhibit some kind of trend over time (i.e., the mean changes over time). One
reason for this (Aoki, 1990) is that the circumstances facing optimizing economic agents
change with time. Another is that in several cases the dynamic structures generating the
data are mostly nonlinear.
Nonstationarity has implications for the standard regression model in
econometrics since the model usually makes assumptions regarding the stationarity of the
error term as well as that of the variables in the regression. It has been shown that running
regression on nonstationary data can give rise to misleading (or spurious) values of R ,
Durbin Watson, and t-statistics, which may lead to an erroneous understanding of the
relationships which exist among the regression variables. For instance, it has long been
recognized (Yule, 1926) that regressing one nonstationary (trended) variable on another
could result in spurious regression.


132
As pointed out in Chapter 2 the industry is of paramount importance to the
maintenance of social and economic stability. It provides valuable foreign exchange for
which the shadow price exceeds the official rate and the earnings used to purchase
valuable capital goods that the subregion must acquire from outside sources. The industry
has a history dating back to the seventeen-century when the subregion produced the bulk
of the world output (Hannah and Spence, 1997). A consequence of this tradition is that
the subregion has acquired knowledge and skill in the cultivation and processing of the
crop. Moreover, a substantial amount of investment has been amassed in the industry and
fixity of these assets make it difficult to shift resources without creating serious social
and economic dislocation (Haraksingh, 1997).
The importance of the sugar industry to the CARICOM countries and its
limitations in terms of competitiveness must therefore be taken into consideration in any
discussion about integrating the subregion into a wider hemispheric free trade area.
Negotiations cannot be based solely on overall efficiency and competitiveness. Rather,
such considerations will have to be based on objectives such as those considered in the
political economy literature. Such a body of literature (Helpman, 1995; Rausser et al,
1982) emphasizes distributional considerations as a means of explaining among other
things, various trade polices of countries. This body of literature for example, views trade
policy as a device for income transfers to preferred interest groups in a society or group
of societies. Although this theme and approach was not explored in the main body of the
dissertation, it is clear that this dissertation has a political economy aspect.
As pointed out in Chapter 1, a dilemma facing CARICOM is whether to integrate
with the Americas within the framework of the FTAA or with the EU. The dilemma


94
quantity of sugar demanded, from X2 to X3 ? and the import quota, it causes only a
movement along the inelastic portion of the sugar demand curve. This implies only a
slight increase in the demand for crystallized sugar (CS). Depending on where the kink
exists in relation to Pti. such a move could represent an appreciable fall in the support
price.
The second case considered is where the new support price falls below the
original kink in Figure 4.10ai.e. a movement from Pti to Pj3. Here it would appear that
the new support price of sugar makes it competitive with HFCS and so there is an
effective demand for some liquid sugar. Other factors remaining constant, the
consumption of sugar (both LS and CS) appears to increase substantially from X2 to Xs.
This would certainly be the case, all other factors remaining constant. However, one
needs to consider the changes, which would have been occurring almost simultaneously
within the HFCS industry.
As the hypothetical model of the long-run situation in the HFCS industry shows
(Figure 4.11), as a consequence of the external economies of scale there are potential
gains to be made as a result of the increased demand resulting from the opening up of the
HFCS industry to international trade. Furthermore, to the extent that the industry retains
some of its market power in the expanded market, producers will be able to obtain
economic rent as shown by the hatched area in Figure 4.11. As such gains are realized, it
would result in a lowering of HFCS, which would result in lowering of the kinked portion
of the US sugar demand curvei.e. from Dsi to Ds2 in Figure 4.10a.(see earlier discussion
on page 91). To the extent that the new kink Ds2 is below PT3, (as shown) the situation is
very much the same to that examined in the earlier casei.e. the change in the US sugar


YEAR
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
139
INTERCEPT
2.35060
2.35060
2.35060
2.35060
2.35060
4.46769
4.55585
4.75492
4.97237
5.18549
5.40252
5.58922
5.76038
5.94295
6.32377
6.69173
7.04554
7.38946
7.71357
7.93952
7.96047
7.97624
SMOOTHED ESTIMATES
HFCS
SUGAR
-1.15039
0.54531
-1.15039
0.54531
-1.15039
0.54531
-1.15039
0.54531
-1.15039
0.54531
-2.22674
1.10953
-2.14419
1.01180
-2.09098
0.90724
-2.04130
0.80051
-1.99177
0.69514
-1.94281
0.59014
-1.89038
0.49291
-1.83814
0.39991
-1.77051
0.28948
-1.74007
0.14978
-1.70745
0.01258
-1.67621
-0.11948
-1.66450
-0.23255
-1.66559
-0.32941
-1.68225
-0.38388
-1.66338
-0.40732
-1.63964
-0.43357
FLOUR
SOFT DRINKS
0.36625
4.37817
0.36625
4.37817
0.36625
4.37817
0.36625
4.37817
0.36625
4.37817
-2.91161
6.54083
-2.82031
6.28829
-2.67334
6.13299
-2.51831
5.99348
-2.36385
5.85107
-2.23221
5.74120
-2.15580
5.67193
-2.08330
5.60115
-2.02052
5.53505
-1.85404
5.50256
-1.70599
5.48071
-1.55618
5.45034
-1.38316
5.41328
-1.19581
5.37920
-1.03380
5.34817
-1.01011
5.34521
-0.98789
5.34171


58
(1993, 1997) that the state-space model is fundamental to dynamic modeling, and that
since most times series models can be put in this framework they provide a more robust
alternative to the modeling of time series analysis. The state-space framework is a
concept borrowed from engineering and is based on the original work of Kalman (1960)
and Kalman and Buey (1961). The concept centers around the possibility of tracking the
economy in a similar manner as engineers would track the state of a system, such as the
location of a satellite or a tanker, using noisy measurements. The Kalman filter
(discussed below) was used to derive optimal estimate of the state, given knowledge of
certain parameters. Engle and Watson (1987) note however, that there is a slight
difference between the approach used by engineers and that used by economists. In the
main, engineers usually have qualitative theories that describe the equation of motion of
physical systems and were primarily interested in the state of the system obtain from
noisy measurements. Economists on the other hand, were less fortunate in not having
such laws of motion of the economy at their disposal and consequently were much more
interested in discovering them from noisy data, rather than in merely estimating the state
of the economy. Consequently, economists had to overcome the initial problem of not
having the parameters of process in estimating the state. The focus for economists
therefore shifted slightly to finding a suitable methodology by which such parameters
could be estimated. The solution to the problem came in the mid-1970s when economists,
with the use of the Kalman filter were able to evaluate the likelihood function in
complex cases. Engle and Watson (1987) note that this development allowed the
parameters to be estimated using maximum likelihood methods as in standard
econometrics. It therefore opened up the way for the widespread use of the technique in


19
Table 2.2. Leading World Sugar Producers and Consumers, 1997-98
Producers
Consumers
Country
Quantity
(MMT)
% of World
Production
Country
Quantity
(MMT)
% of World
Consumption
European Union
19.3
15.4
India
16.7
13.1
Brazil
16.6
13.2
European Union
14.4
11.3
India
14.5
11.6
China
9.0
7.0
China
8.6
6.8
United States
8.9
7.0
United States
7.3
5.8
Brazil
8.8
6.9
World
125.4
100.0
World
126.9
100.0
Source: USDA, Sugar and Sweeteners Situation and Outlook Yearbook.
Only a relatively small proportion of the large world production of sugar is traded
each year. Schmitz (1995, p 54) notes that it is common to divide the market for sugar
into three distinct markets: 1) the domestic market within the sugar-producing countries.
This market, he contends, is the largest and accounts for about 75% of all world sugar
production; 2) the market for the various international agreements between certain
importers and certain exporters. These agreements include the import quotas under the
US program, the bilateral agreements between Cuba and the former Soviet Republics and
the agreements of various countries and groups of countries such as the ACP with the
EU. This market makes up about 10% of world production; and 3) the market for the
residual free market in world sugar, which accounts for the remaining 15% of world
production.
For the fiscal year 1997-98, approximately 35.6 MMT (28.3 percent of world
production) of sugar was traded globally (Table 2.3). The five most important sugar
exportersBrazil, European Union, Australia, Thailand, and Cubaaccounted for 65.1
percent of global exports. Imports, on the other hand, have been less concentrated, with
the share of the top five sugar-importing countries and regions in 1997-98Russia,
United States, European Union, Japan, and Koreaamounting to only 30.1 percent.


115
Fig. 5.1. Dynamic path of HFCS intercept over time, 1982-98
The exponential growth beginning toward the end of the eighties and continuing
to about 1996 is consistent with the fact that by the late 1980s all the major soft drink
producers had switched from using liquid sugar to HFCS. In addition, it also reflects
changes in productivity resulting from technical progress and increased capital stock
among the users of the product. It is well known that the producers of HFCS work in
concert with the users to modify the product so as to improve its acceptability. With the
general up turn in the state of the US economy beginning around 1992 there has been an
increase in the HFCS demanders (increased capital stock) resulting from an increase in
the demand for final HFCS-containing commodities. Figure 5.1 reveals however, that


113
necessarily act in unison. Since the double log formulation was used, the values of the
coefficients also represent the elasticities. The estimate of the own-price elasticity of
HFCS was -1.64 implying that the demand for the commodity was relatively elastic. In
the case of sugar the cross price elasticity was inelastic but not significant at any
reasonable level of confidence. While the price of flour was non significant at the 10%
level the price of soft drink was significant at this level. The last column of Table 5.2
shows the absolute values of the Beta (standardized) coefficients. These values give an
indication of the relative importance of the independent variables in the model and allow
for direct comparison (Pindyck and Rubinfeld, 1991). Thus, with respect to the price of
HFCS the value implies that a one standard deviation change in the price of HFCS leads
to a 2.27 standard deviation change in the quantity of HFCS demanded. The values also
reveal that among the explanatory variables chosen the price of soft drink was the most
important in explaining the relative variation in the quantity of HFCS demanded.
Attention is now turn to an examination of the implications of these estimates on
the time path adjustments for the selected parameter of the trend, own-price elasticity of
HFCS, and the cross-price elasticity with respect to sugar.
Model Implications
The state-space and the Kalman filter allow for the modeling of time-varying
parameters and for stochastic movements in the underlying trend variable. Addressing an
observation made by Engle and Watson (1987) that time-varying parameters when used
in most studies are typically interpreted as tests for the stability equation, the emphasis
here is on the economic importance of such variations. As such the concern is not so
much with whether the estimated parameters are stable and represent long-term


146
Zietz, J., and A, Valdes. The Potential Benefits to LDCs of Trade Liberalization Sugar
and Beef by Industrialized Countries. Weltwirtschaftliches Archiv 22 (1986): 439
-464.
Zitt, M. A Complex Invention Process: Industrial, Technological and Scientific
Dynamics in HFCS History. Paper presented at the 12th International Economic
History Congress, Spain, 1998.


13
demand for HFCS, in which the phenomenon being considered might be nonstationary
and still evolving.
Organization of the Study
This dissertation consists of six chapters. Following this introduction, chapter 2
provides background information on the trends and developments and the institutional
settings of the sugar and HFCS markets. It therefore places the study within the wider
context, highlighting the complex nature of the issue on hand and the varied interests.
Chapter 3 begins with a review of the literature, focusing on studies that have attempted
to include the HFCS industry in their modeling exercises. This is followed by a review of
the theoretical background to some of the issues, which are considered and used in the
subsequent chapters. In particular, a detailed overview of the state-space model and
Kalman filter is provided. In Chapter 4 a conceptual model is constructed of the US sugar
and HFCS subsector. The model is then used to examine a few scenarios within the
framework of the proposed FTAA. Implications of the results are drawn for the non-US
sugar producers, in particular the CARICOM producers. Chapter 5 takes the analysis a
step further and presents the two-stage estimation of the derived demand for HFCS in the
US. The first section of the chapter focuses on the specification of the model while the
latter section presents and discusses the implications of the results. The final chapter
summarizes the conclusions from the research and makes suggestions with regard to
some of the critical policies that political and regional planners might wish to take into
consideration in their ongoing integration negotiations as they relate to the future of the
CARICOM sugar industry.


97
some of their market power, then their gains would be even larger than that associated
with the traditional producer surplus, since they would be able to obtain economic rent.
In the case of the sugar exporting countries in general, the producers would lose as the
cheaper HFCS displaces some of their domestic market and with little hope of increased
access to the US sugar market. Moreover, revenues from sales on the world market would
decline as a result of the fall in price.
Consumers on the other hand would in general benefit from the lower all round
prices. The consumers in the US would gain considerably as both the prices of HFCS and
sugar fall. Contrary to the expectation from conventional trade theory which implies that
the consumers in the exporting countries are usually made worse off as a result of the
opening of the country to trade, in this case the opposite occurs as the users of HFCS in
the US will benefit from the reduction in prices associated with external economies of
scale.
Within the context of the FTAA, the major losers would be the relatively high
cost sugar industries of the small CARICOM countries. Firstly, they would be losing
their US import quota rent. Secondly, due to the permanent displacement of a portion of
the sugar market, it is likely that there would be a much smaller US market for the sugar
produced by these countries due to the fact that the cost of production within these
countries exceeds the current US domestic support price of approximately 18 cents per
pound of raw sugar. For example, production cost in Jamaica is reported to be US 33
cents per lb. (Caribbean Update, 1998). Also of consequence, is the fact that three of the
world's most efficient sugar producers; Brazil, Colombia and Guatemala would be
included as members of the FTAA and the region as a whole is a net supplier of sugar.


DYNAMICS OF THE US SUGAR AND HFCS MARKETS:
IMPLICATIONS FOR THE CARICOM SUGAR PRODUCERS WITHIN THE
FRAMEWORK OF THE PROPOSED FTAA
By
EDWARD ANTHONY EVANS
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1999

To Sabrina, Latoya, Akeem, Attli and ma dodo

ACKNOWLEDGMENTS
I would like to express my profound gratitude to Dr. Carlton Davis, chairman of
my supervisory committee, for all his help and encouragement throughout the course of
this study. Dr. Davis' patience, understanding, guidance, assistance and constructive
criticisms given to me during the preparation of this dissertation are greatly appreciated.
His desire for excellence, and steadfast refusal to lower the bar kept my shoulders to the
wheel and has contributed immensely to my education and self-development.
Drs. Max Langham and Ron Ward, members of my committee, deserve special
mention. I have benefited a great deal from Dr. Langham's criticisms and vast experience,
and even though we did not always agree, I am often reminded that the beauty of a choir
comes from the fact that not everyone sings the same melody. Dr. Ward has provided
useful suggestions and has welcomed me into his home on several occasions during the
night.
To the other members of my committee, Drs. R. Emerson, A. Schmitz and D.
Denslow I am eternally grateful. Without their criticisms and useful suggestions this
effort would not have been possible. Special thanks is also due to Drs C. Moss and R.
Kilmer who made worthwhile contributions in the initial stages of the dissertation.
I gratefully acknowledge the financial support I received from the Department of
Food and Resource Economics. In this regard I must again say thanks to Dr. Davis and
will always remember the words he spoke which jolted me into finally accepting his
offer"there are more dogs than bones. I would also like to acknowledge the stipend I
iii

received from the Caribbean Agricultural Research and Development Institute (CARDI)
during my initial years. Special mention must be made of Mr. Hugh Saul whose faith in
me and timely intervention made it possible for me to receive the latter financial support
and take up the offer at the University of Florida.
Perhaps my greatest depth of gratitude however, is owed to my wife, Sabrina,
children, Latoya and Akeem, my mother and my godfather, Attli, for their encouragement
and for the sacrifice they have made in favor of this effort. To them this dissertation is
gratefully dedicated.
Finally, I wish to express my sincere gratitude to all those who stood in my comer
cheering my every move. Among them are, my dear friend Dr. Kenrick Jordan, my most
ardent supporter, Dr. Pauline Lawrence, Dr. Curtis McIntosh, Mr. Karl James, Greta and
Danny Roberts.
IV

TABLE OF CONTENTS
page
ACKNOWLEDGMENTS iii
LIST OF TABLES viii
LIST OF FIGURES ix
ABSTRACT x
CHAPTERS
1 INTRODUCTION 1
Statement of Problem 3
Problematic Situation 3
Specific Problem 7
Hypotheses 10
Objectives 11
Methodology 12
Organization of the Study 13
2 BACKGROUND 14
Free Trade Area of the Americas (FTAA) 14
Review of World and Western Hemisphere Sugar Markets 17
World Production and Market TrendsSugar 17
Western Hemisphere Production and Market TrendsSugar 21
Importance of the Sugar Industry to CARICOM Sugar Producers 23
Socioeconomic Importance 23
The EU/Lome ACP Agreement 25
The US Sugar Program and Policy 27
Review of World and Western Hemisphere HFCS Markets 31
World Production and Market TrendsHFCS 33
Western Hemisphere Production and Market TrendsHFCS 34
3 LITERATURE REVIEW AND THEORETICAL BACKGROUND 39
Previous Empirical Studies 39
Sugar and HFCS Related Studies 39
v

Evaluation and Critique 46
Time Series Analysis 47
Serial Correlation 48
Stationary 49
Cointegration and Error Correction Models 51
Time-Varying Parameters 55
The State-Space Framework 57
The State-Space Model 57
Kalman Filter 61
Advantages of State-Space Approach and the Kalman Filter 64
Relevance of the State-Space Model and Kalman Filter to the Current Study 66
The Concept of a Derived Demand 67
4 CONCEPTUAL MODEL OF US SUGAR AND HFCS MARKETS 71
The US Sugar and HFCS Subsectors 72
The US Sugar Subsector 73
The US HFCS Subsector 78
The Interrelationship between the US Sugar and HFCS Markets 86
Trade Implications of Changes in the US Sugar Program within an FTAA 90
Free Trade Area 90
Scenario I- Partial Liberalization of the US Sugar Program 92
Trade implications 92
Welfare implications 96
Scenario II- Complete Liberalization of the US Sugar Program 98
Trade implications 98
Welfare implications 101
5 EMPIRICAL MODEL OF THE US DEMAND FOR HFCS 103
Empirical Model 103
Econometric and Statistical Issues 106
Description of Data 110
Estimation Results 111
Model Implications 113
Intercept Adjustment Over Time (gt) 114
Own-Price Elasticity (Pit) 116
Cross-Price Elasticity (P2O 118
6 SUMMARY, POLICY RECOMMENDATIONS AND SUGGESTIONS FOR
FURTHER RESEARCH 121
Summary and Conclusions 121
Theoretical Findings 125
Empirical Findings 128
Policy Implications and Recommendations 131
Limitations of the Study and Suggestions for Further Research 135
vi

APPENDIX
STATE AND SMOOTEHED ESTIMATES 138
REFERENCES 140
BIOGRAPHICAL SKETCH 147
Vll

LIST OF TABLES
Table page
Table 2.1. Ranking of Countries of the Western Hemisphere on Basis of Population,
Land Size and GDP 16
Table 2.2. Leading World Sugar Producers and Consumers, 1997-98 19
Table 2.3. Leading World Sugar Exporters and Importers, 1997-98 20
Table 2.4. Leading Hemispheric Sugar Producers and Consumers, 1997-98 22
Table 2.5. Leading Hemispheric Sugar Exporters and Importers, 1997-98 23
Table 2.6. Socioeconomic Importance of CARICOM Sugar Industries 24
Table 2.7. CARICOM/EU Market Real Earnings and Quota Rent (1992=100) 26
Table 2.8. CARICOM/US Market Real Earnings and Quota Rent (1992=100) 28
Table 2.9. World HFCS Production, 1989-97 ('000 metric tons dry basis) 34
Table 2.10. Changes in US Per Capita Sweetener Consumption and Relative Market
Shares, (1975 1997) Pounds, Dry Weight Basis 36
Table 5.1. Estimation of Price Equation 112
Table 5.2 Final Estimates Using Kalman Filter Over the Period 1977-98 112
vm

LIST OF FIGURES
Figure page
Fig. 2.1. World and US refined and raw real sugar prices, 1985-98 (1992=100) 18
Fig. 2.2. Changes in US per capita consumption of selected sweeteners, 1975-97 36
Fig. 2.3. Real HFCS and refined sugar prices, 1982-98 (1992=100) 37
Fig. 4.1. Flowchart of the US HFCS and Sugar Subsectors 72
Fig. 4.2. Hypothetical US demand for crystallized sugar (CS) and liquid sugar (LS),
respectively 74
Fig. 4.3. Hypothetical US aggregate demand for sugar 76
Fig. 4.4. Hypothetical US supply and demand for sugar 77
Fig. 4.5. Hypothetical US demand for HFCS 79
Fig. 4.6. Shift in hypothetical HFCS short-run marginal cost curve 82
Fig. 4.7. Hypothetical US HFCS short-run marginal cost and demand configuration 83
Fig. 4.8. Hypothetical HFCS industry average costs, reflecting external economies 86
Fig. 4.9. Hypothetical supply and demand for sugar, and hypothetical demand for HFCS,
respectively 88
Fig. 4.10. Trade implications under partial liberalization of US sugar program 93
Fig. 4.11. Hypothetical HFCS industry long-run demand and average cost 95
Fig. 4.12. Trade implications under completely liberalized US sugar program 100
Fig. 5.1. Dynamic path of HFCS intercept over time, 1982-98 115
Fig. 5.2. Dynamic path of adjustment in HFCS own-price elasticity, 1982-98 117
Fig. 5.3. Dynamic path of adjustment of cross-price elasticity of HFCS with respect to
sugar, 1982-98 119
IX

Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
DYNAMICS OF THE US SUGAR AND HFCS MARKETS:
IMPLICATIONS FOR THE CARICOM SUGAR PRODUCERS WITHIN THE
FRAMEWORK OF THE PROPOSED FTAA
By
Edward Anthony Evans
December 1999
Chairman: Carlton G. Davis
Major Department: Food and Resource Economics
To cope with growing regionalism, many developing countries must choose
between alternative and competing integration strategies. The Caribbean Community
(CARICOM) is faced with such a dilemma. The United States and the European Union
have proposed the formation of separate free trade areas (FTAs) with the subregion with
a tentative implementation date of 2005. The consensus emerging in CARICOM favors
participation in the proposed Free Trade Area of the Americas (FTAA)-a potential
grouping of 34 disparate countries in the Western Hemisphere. While the anticipated

benefits of such a group are unquestionable, there will be losses at the sector and industry
levels.
The CARICOM sugar industry is of considerable importance to the subregion but
could be adversely affected by the FTAA. First, it is possible that CARICOM could
violate the Lome Agreement through FTAA membership, denying continued access to
the lucrative EU sugar market. Second, CARICOM sugar producers could be forced to
x

compete with larger and more efficient sugar-producing FTAA member countries-such
as Brazil, Colombia and Guatemalafor a share of the regional sugar market. Third, and
the focus of this dissertation, a substantial portion of the regional sugar market might be
eroded as a result of the successful penetration of high fructose com syrup (HFCS)
produced in the United States.
A qualitative assessment of the potential economic impact of changes to the US
sugar program and an expanded HFCS industry on the CARICOM sugar industry was
conducted. First, a conceptual model of the US sugar and HFCS markets was developed,
and potential trade and economic impacts were discussed. Second, an empirical model of
the demand for HFCS in the United States was constructed. To capture the dynamics
within the industry a state-space representation utilizing the Kalman filter was employed.
The maturity of the HFCS industry was assessed, and implications for the various sugar
concerns were drawn.
The findings substantiated the view that CARICOM membership in an FTAA
could have potential negative economic impact on the CARICOM sugar industry and that
an FTAA could further depress, rather than increase, the world market sugar price.
Finally, it was concluded that the US HFCS industry is entering a mature phase, forcing
the need for com refiners to actively seek external markets for their products.
xi

CHAPTER 1
INTRODUCTION
The global economic environment is in a state of rapid transition. The new
dispensation in both the developed and developing countries is the opening up of national
and regional economies to market forces. In the developed countries, the principal forces
leading to trade liberalization stemmed from: 1) the growing budgetary pressure of
having to maintain protective regimes; 2) growing regionalism; and 3) multilateral
negotiations such as the General Agreement on Tariff and Trade (GATT/WTO).
Developing countries, on the other hand, have tended to liberalize their trading regimes
either unilaterally or as a result of following the dictates of multilateral financial and
developmental institutions (Wint, 1997). Whatever the reason, the result has been a
dramatic reduction in trade barriers.
For developing countries, the wide spread process of trade liberalization and
globalization is only one element of the new global dispensation to which they must
respond. An important related element is the erosion of non-reciprocal preferential
treatment under the Generalized System of Preferences (GSP) or other preferential
regimes such as Lome1, which in many cases has been a mainstay of several of the
economies. At the same time, LDCs excluded from regional blocsas in the case for the
1 Lome Convention is a contractual non-reciprocal trading arrangement between African, Caribbean and
Pacific (ACP) countries and The European Union (EU). The first Lome Convention (Lome 1) was
established in 1975 and the Convention is now in its fourth round (Lome 4). It expires in the year 2000 and
is currently being re-negotiated albeit, under new philosophical circumstances.
1

2
Caribbean and some of their Latin American neighbors vis--vis NAFTA-are likely to
face losses due to trade and investment diversions, higher non-tariff barriers (NTBs) and
possibly higher tariffs (Blanco, 1997).
As regionalism grows, many developing countries must choose among alternative
integration strategies. Accordingly, they are reassessing existing economic and trading
arrangements as well as considering the prospects of forming new relations and strategic
alliances so as to preserve, consolidate or establish stronger trading positions. The
Caribbean Community (CARICOM)2a group of 14 countries in the Western
Hemisphereis one such alliance of developing countries. With a population of
approximately 6.0 million and a total gross domestic product (GDP) at market prices of
only about US $14.0 billion, these countries represent a very small market by world
standards (Bryan, 1995). In addition to the Lome arrangement, these countries have
benefited a great deal from other marketing arrangements such as Caribbean and Canada
Agreement (CARIBCAN), Caribbean Basin Initiative (CBI), and GSP. Both the US and
the EU, however, have now separately proposed forming Free Trade Areas (FTAs) with
CARICOM with a tentative implementation date being the year 2005 or there about.
While details of such arrangements are still being refined, they will have far reaching
consequences for CARICOM's relationships with Europe, the other ACP countries, the
Americas and much of the developing world. Within CARICOM, these arrangements
2 Member countries of the CARICOM include: Antigua and Barbuda, The Bahamas, Barbados, Belize,
Dominica, Grenada, Guyana, Jamaica, Montserrat, St. Kitts/Nevis, St. Lucia, St. Vincent and the
Grenadines, Suriname, and Trinidad and Tobago. Haiti has formalized arrangements for full membership as
of August 1999 bringing the total membership to 15. However, in view of that countrys recent accession,
this study does not include it in the analysis.

3
will change how the subregion conducts its business with the rest of the world and how it
earns its foreign exchange.
Faced with a set of perplexing choices, the CARICOM countries have begun to
systematically weigh the benefits and costs of embarking on one of the following three
strategies: 1) becoming a part of a Free Trade Area of the Americas (FTAA) with the US
as the central figure; 2) seeking to form a Free Trade Area with the EU as part of the new
Lome arrangement; and 3) remaining outside of the formation of any free trade area with
the US or the EU. The emerging consensus favors (1). Irrespective of which strategy is
chosen, there will be serious internal economic conflicts because of gains and losses at
the sectoral and industrial levels. Moreover, the decisions made could have irreversible
economic consequences.
The CARICOM sugar industry, a major contributor to the economic and social
stability of the subregion, is undoubtedly one of the more important industries that will be
affected as the new economic arrangements emerge. Within the framework of the
proposed FTAA, for example, the industry is likely to be faced with the additional
economic challenges associated with significant structural changes that are occurring
within the US-one of its major sugar markets. In particular, this challenge could come as
a result of the rapid development of the US High Fructose Com Syrup (HFCS) Industry
and its expansion in a much freer trading environment.
Statement of Problem
Problematic Situation
The CARICOM countries are characterized as having certain features which,
when taken in combination, makes them extremely vulnerable. These features include

4
having: 1) small economies with a relatively high degree of openness as reflected in
trade-to-output ratios; 2) undiversified sectoral and general economic structures; 3)
export concentrations in one or two products; 4) small firms with limited opportunity to
take advantage of economies of scale; 5) a high degree of dependency on non-reciprocal
trading arrangements for historical reasons; and 6) a limited-resource based sugar
industry with technologies heavily oriented toward labor intensity and high cost
productions structures (West Indian Commission, 1993).
As indicated earlier, these countries have benefited appreciably from the
preferences they enjoyed in various marketing and trading arrangements such as the
EU/ACP Sugar Program and the US Sugar Program. A consequence has been a degree of
complacency and technical obsolescence on the part of regional producers that have led
in part to an appreciable reduction in the levels of production and marketing efficiencies3.
A recent study conducted by LMC International and reported by the Economic Research
Service (ERS) of the United States Department of Agriculture (USDA) indicated that
none of the CARICOM countries can be regarded as low cost sugar producers. Barbados
and Trinidad and Tobago were among the highest cost producers in the world. The
remaining four CARICOM producers (Belize, Guyana, Jamaica and St. Kitts/Nevis) were
classified as medium cost producers (Haley, 1998a). Production cost in Jamaica is
reported to be in the vicinity of 33 cents per pound, compared to the US and Brazil prices
of 22 and 14 cents, respectively (Caribbean Update, 1998; USDA, 1998). To complicate
3 This point has been conceded by representatives of the Sugar Association of the Caribbean (SAC)-- the
official producer association. They pointed out however, that while current initiatives are being taken to
modernize the industry and improve both field and factory levels efficiencies, efforts to attract much
needed capital investments are being hampered by the uncertainty which surrounds the future of the
subregion's industry. They have also indicated that levels of efficiencies that can be obtained will not match
some of those reported in parts of the US due to their economies of scale (James, 1999 Personal Interview).

5
matters iirther, a considerable amount of the land used in the subregion to produce sugar
cane is marginal with respect to sugar cane production but has few viable alternatives.
Also, from a wider socioeconomic point of view, labor and capital could become socially
disruptive in these small undiversified economies if other alternatives were pursued too
quickly (Haraksingh, 1997).
Notwithstanding these drawbacks, the industry remains of considerable import to
the subregion, provides much of the valuable foreign exchange earned, and is a major
contributor to the employment of labor. For example, the industry provides over one half
of the foreign exchange earnings in St. Kitts and Nevis and employs about 97 thousand
persons or about 5 % of the active working population within CARICOM (James, 1998).
Moreover, the socioeconomic importance of sugar production to these countries goes
much further than the above figures might suggest as many of these industries do play a
vital role in the maintenance and upgrading of rural infrastructure, health, recreational
and educational facilities.
With the general consensus that CARICOM should seek to become part of the
proposed FTAA, there has been a genuine concern with regards to the future of its sugar
industry (Group of Experts, 1998). Such a move toward integrating the economies brings
with it the challenge of competing in a liberalized trading environment without the
benefit of preferences to which the subregion has grown accustomed.
The following three interrelated issues with respect to the subregional sugar
industry are of importance. First, there is the possibility that should the subregion pursue
such a path it could be in violation of the terms and conditions of its Lome Agreement.
Specifically, article 174 section 2a of the Agreement forbid recipients to enter into any

6
economic arrangement with any developed countries which would result in such
countries receiving terms and conditions more favorable than those given to the EU
(Group of Experts, 1998). As indicated earlier, more than three-quarters of the
subregion's exports of sugar goes to the EU market where producers receive a high
enough price to cover the relatively high cost of production. Becoming part of the FTAA
could therefore mean that CARICOM might either have to sacrifice its preferential
market in the EU or at best offer the EU similar terms and conditions as those given to
the US and Canada, the developed members of the proposed FTAA. Resolving this issue
within the framework of an FTAA might not be as simple as it seems in view of the fact
that the US took a similar stance, within the framework of its CBI initiative, to that taken
by the EU in their Lome Agreement.
Second, an FTAA, which includes the subregion, would place the CARICOM
sugar industry into direct competition with the larger and more efficient producers in the
hemisphere. The study conducted by LMC, referred to above, notes that three of the
world most efficient sugar producers, Brazil, Columbia and Guatemala are in the
hemisphere. These low-cost producers could easily supply the sugar requirements of the
other members of the FTAA.
Third, is the fact that the US while being the largest sugar importing country
within the hemisphere is also a major producer of all the principal sweetenerscane
sugar, beet sugar, com sweeteners and several non-caloric sweeteners. Its sugar and
sweetener industry is reported to be the largest in the world, with annual consumption of
caloric sweeteners approaching 20 million short tons (Lord, 1995). Moreover, with the
technological advances in the com sweetener industry, the United States has become the

7
world's lowest-cost producer of sweeteners. The world average cost of production is
20.91 cents per pound. It is 13.73 cents per pound in the most efficient sugar producing
countries. In comparison the US produces HFCS in 1994/95 at an average cost of only
10.60 cents per pound dry weight equivalence (Haley, 1998a). HFCS competes directly
with sugar in most industrial uses. In a situation of no foreign tariffs, duties, taxes,
quotas, and other border barriers against com sweetener imports, the US could export its
technology and/or com sweetener product to some of the higher-cost markets within the
FTAA. The US would also benefit as the world's largest and cheapest producer of com-
the main input used in the production of HFCS. The potential of HFCS to displace sugar
has already been seen in the US and to a lesser degree in Canada. And, though it has not
yet adversely affected the other sugar producing and consuming countries, those in the
planned FTAA are likely to feel its impact.
Specific Problem
This study is concerned with the threat that an expanded HFCS industry could
pose to the CARICOM sugar industry within the framework of a liberalized trading
environment. The US sugar industry and its sugar program have been adequately dealt
with in the literature. The same is not true in the case of the HFCS industry. Not much
has been written on the likely impact that an expanded HFCS industry could have on
world and regional prices, production, consumption and trade patterns of sugar. Also, the
dynamics of the US sugar-HFCS interrelationships and the implications of such
relationships for trade in an extended market have not been fully explored. Thirdly, as
noted by Tanyeri-Abur (1990) there appears to be an absence of studies to form a basis
for studying the trade implications of an expanded HFCS industry. As a consequence,

8
studies of production and consumption of the commodity have been largely confined to
the domestic market. With the advancements in HFCS technology on both the supply and
demand sides and the reduction in trading barriers in the international arena, the prospects
of this commodity assuming a greater level of importance in trade have increased
considerably.
One aspect of the HFCS market is the dual nature of the product, whereby within
a given range of the sweetener demand curve it substitutes almost perfectly for sugar but
within another range it does not. Either of the sweeteners (HFCS or sugar) can be used in
soft drinks, but for some other uses that require a crystallized sweetener, HFCS is not a
viable substitute. This dual nature of the substitutability relationship has implications for
the extent to which the commodity will be traded internationally, the shape of the
sweetener demand curve and the econometric approach to be used to estimate HFCS
demand parameters.
Another related aspect is the issue of the extent to which the industry has reached
a level of maturity. Thomas (1985), pointed out that HFCS when considered by industrial
standard is regarded as a relatively new product. Among other things, he noted that this
implies that there will almost certainly be further technological improvements and that
the behavior of the industry can be expected to be different from that of a mature industry
such as sugar. If this is indeed the case, then there are implications with regard to the
econometric approach used to estimate the important economic parameters. A cursory
investigation into the history of the industry seems to support the point made by Thomas.
On both the supply and demand sides of the market advances in technology have resulted
in noticeable development of the product and product usage. For example, the price of

9
HFCS since its inception has been lower than that of sugar on a sweetness equivalency
basis. Nevertheless, the adoption of this product had been cautious. It took a decade for
the major soft drink industries to convert from liquid sugar to HFCS. On the demand side
the adoption of the product has been aided by the producers of HFCS working in concert
with the users, providing technical assistance and adapting the product to the special
requirements of the various clients. It is in this context that HFCS-55 was developed as a
more effective substitute for liquid sugar than HFCS-42 in the soft drink industry.
Likewise on the supply side there has been a constant effort aimed at improving the
technology involved in the production of the commodity, with a view to producing an
effective substitute sweetener at the lowest cost. The implication is that the industry is
evolving and as such the parameters of the industry might be changing. This time-varying
issue must be recognized in characterizing the industry in any analytical model.
Ward and Tilley (1980) pointed out that in markets, such as the one we are
concerned with where significant adjustments are occurring, the parameters are likely to
be time-varying. They noted that technological changes result in structural changes in
economic phenomenon, giving rise to varying parameters. They cautioned that a failure
to take into consideration such variations in the estimated parameters could cause serious
errors in projections and policies drawn from them.
This issue of the degree to which the HFCS industry in the US can be considered
mature is therefore important from an estimation point of view. It is also important in the
context of this study since the dynamic paths of adjustments of various economic
parameters have such implications for the CARICOM sugar industry such as: 1) the
extent of saturation of HFCS in the US market and by extension the size of the US

10
domestic sugar requirement; and 2) likely rent-seeking behavior on the part of the com
refiners.
From the above it is clear that any assessment of the likely impact of an FTAA on
the CARICOM sugar industry must include a thorough understanding of developments
occurring within the US sweeteners industry. To focus only on sugar per se would be to
overlook an important component of the dynamics occurring within the regions
sweeteners market. Understanding the dynamics between HFCS and sugar and generating
reliable parameters for the HFCS industry, are required for a meaningful assessment of
the manner in which the HFCS industry might respond in a freer trading environment.
Hypotheses
1. The parameters of the supply and demand configuration of the HFCS industry
have been varying over time because of the relative "newness" of the HFCS
industry and the influence of both scale and technological factors.
2. The market for HFCS is approaching an asymptotic level, giving rise to the need
to seek external markets for the product, and a basis for a new round of rent
seeking by com refiners.
3. If the FTAA becomes a reality, the US could become a major exporter of HFCS
in the near future with a sizable proportion of the product going to the countries in
the regional trading blocs.
4. If HFCS remains competitive and can be substituted for sucrose causing a
structural shift in the demand and supply of sugar, there will be an excess supply
of sugar in the FTAA resulting in depressed regional and world prices.

11
5. Given the above, membership of CARICOM countries in the FTAA will in the
short to medium term have potentially negative economic impacts on the sugar
industry of the subregion.
Objectives
The study has two general objectives. Firstly, to qualitatively assess the potential
economic impact that likely changes to the US sugar program, and an expanded High
Fructose Com Syrup (HFCS) industry, might have on the CARICOM sugar industry
within the framework of the proposed Free Trade Area of the Americas (FTAA).
Secondly, to provide quantitative estimates for the HFCS demand configuration that can
be used in subsequent modeling exercises. Specific objectives are to:
1. Assess selected trends in the US sweeteners industry.
2. Develop a conceptual model of the US sugar and HFCS markets.
3. Qualitatively assess the trade and welfare implications of likely changes in the US
sugar program and an expanded HFCS industry on the sugar producing countries
of the FTAA, with specific reference to the CARICOM sugar producers.
4. Develop and estimate the derived demand for HFCS in the US sweeteners market.
5. Investigate the demand side dynamics within the HFCS industry, and
6. Identify critical policies regional planners might wish to take into consideration in
their ongoing integration negotiations, as they relate to the future of the
CARICOM sugar industry.

12
Methodology
The purpose of this study is to assess the potential economic impact that likely
changes to the US sugar program, and an expanded High Fructose Com Syrup (HFCS)
industry, might have on the CARICOM sugar industry within the framework of the
proposed Free Trade Area of the Americas (FTAA). This will be accomplished in two
stages. First, a conceptual model of the US sugar and HFCS markets will be constructed.
Although constructed within a static framework, the model will attempt to highlight some
of the dynamics between sugar and HFCS. Based on the model, the implications of
possible changes in the US sugar program, and the impact of an expanded HFCS industry
on non-US sugar producers will be discussed.
The second stage of the exercise will be to empirically model the demand for
HFCS in the US with a view of assessing the maturity of the industry, and drawing
implications for the various sugar concerns. The focus will be on the dynamic paths of
adjustments of various parameters of the estimated demand function. To facilitate such
an investigation a state-space modeling approach and the Kalman filter will be employed.
Among the advantages of utilizing the state-space form and the Kalman filter are that: 1)
it allows for modeling of time-varying parameters, if the situation so warrants, and can
provide an estimate at any point within the sample and hence a valuable way of
examining how a trend might have evolved; 2) it can deal fairly easily with nonstationary
time series which traditional econometric methods have difficulty in handling; and 3) it
allows modeling to be done in level form (rather than differenced) in situation where the
data is nonstationary thus facilitating interpretation of results. The flexibility of the state-
space model thus makes it an appropriate tool for estimation in situations, such as the

13
demand for HFCS, in which the phenomenon being considered might be nonstationary
and still evolving.
Organization of the Study
This dissertation consists of six chapters. Following this introduction, chapter 2
provides background information on the trends and developments and the institutional
settings of the sugar and HFCS markets. It therefore places the study within the wider
context, highlighting the complex nature of the issue on hand and the varied interests.
Chapter 3 begins with a review of the literature, focusing on studies that have attempted
to include the HFCS industry in their modeling exercises. This is followed by a review of
the theoretical background to some of the issues, which are considered and used in the
subsequent chapters. In particular, a detailed overview of the state-space model and
Kalman filter is provided. In Chapter 4 a conceptual model is constructed of the US sugar
and HFCS subsector. The model is then used to examine a few scenarios within the
framework of the proposed FTAA. Implications of the results are drawn for the non-US
sugar producers, in particular the CARICOM producers. Chapter 5 takes the analysis a
step further and presents the two-stage estimation of the derived demand for HFCS in the
US. The first section of the chapter focuses on the specification of the model while the
latter section presents and discusses the implications of the results. The final chapter
summarizes the conclusions from the research and makes suggestions with regard to
some of the critical policies that political and regional planners might wish to take into
consideration in their ongoing integration negotiations as they relate to the future of the
CARICOM sugar industry.

CHAPTER 2
BACKGROUND
The purpose of this chapter is to provide background information on key players
and institutional arrangements involved with sugar and HFCS. The chapter contains a
brief description of the proposal to form a Free Trade Area of the Americas and
highlights some of the disparities among the potential members. This is followed by a
review of world and Western Hemispheric trends in the production and marketing of
sugar. Next is a discussion of the importance of the sugar industry to CARICOM. Key
marketing arrangements such as the EU/ACP Sugar Protocol and US Sugar Program are
examined. Implications of the GATT/WTO rules are also briefly examined as they relate
to these special marketing arrangements. The final section of the chapter focuses on the
HFCS and reviews world and Western Hemisphere production and marketing trends for
this commodity.
Free Trade Area of the Americas (TTAA)
A Free Trade Area of the Americas (FTAA) is one of several trade agreements
that the United States is encouraging and actively pursuing. The decision to establish an
FTAA and to conclude the negotiation for its implementation by the year 2005 was made
at the Summit of the Americas held in Miami, Florida in December 1994. Among other
things, the Plan of Action calls for the promotion of prosperity among the 34
participating countries through economic integration and free trade, the eradication of
poverty and discrimination, and the guaranteeing of sustainable development by the
14

15
conservation of natural environment for future generations. The process has five
elements: 1) liberalization of barriers to trade in goods and services; 2) elimination of
restraints on investments; 3) provisions of free labor movement for specialized workers;
4) harmonization of tax and monetary policies; and 5) establishment of supra regional
institutions to administer the arrangements and to engage in dispute resolution. Bryan
(1995) notes that the new integration process goes beyond the traditional concept of
countries simply extending reciprocal preferences to their trading partners. He cautions,
however, that while absolute reciprocity may be the ultimate goal of the hemispheric
trade liberalization, many of the smaller Caribbean economies simply cannot compete
with countries having more productive infrastructures and technologies. Stated
differently, it might not be possible for them to offer absolute reciprocity to industrial
countries at least in the short run, and there will be the need for some compromise on the
part of the more developed trade partners (Bryan, 1995).
To better appreciate some of the disparities among the countries, Table 2.1 shows
a relative ranking of the 34 countries that will comprise the FTAA with respect to the
selected indicators of population, land area and GDP. In general, the CARICOM
countries (underlined in Table) occupy the lower levels in all three cases. The subregion
accounts for only 0.67% of the Hemisphere's land area, 0.82% of its population and
approximately 0.2% of total GDP. The Table also reveals that the most populous and
highest income country is the United States and the least populous and lowest income
country is St. Kitts and Nevis. Barbados has the smallest land area while Canada has the
largest. What is not obvious from the Table however, is the fact that many

16
Table 2.1. Ranking of Countries of the Western Hemisphere on Basis of Population,
Land Size and GDP
Rank
Population
1995
(MiniS"'
ns)
Land Size
(000 sq.
Km.)
GDP
1995
(US$ M
1990=100)
34
St. Kitts & Nevis*
0.04
Barbados
0.3
St. Kitts & Nevis
150
33
Antigua & Barbuda
0.06
Grenada
0.3
Dominica
178
32
Dominica
0.07
St. Kitts & Nevis
0.3
St. Vin. and Gren.
218
31
Grenada
0.09
Antigua & Barbuda
0.4
Grenada
229
30
St. Vin. and Gren.
0.11
St. Vin. and Gren.
0.4
Suriname
334
29
St. Lucia
0.14
St. Lucia
0.6
Antigua & Barbuda
366
28
Belize*
0.21
Dominica
0.7
Belize
491
27
Barbados*
0.26
Trinidad & Tobago
5.0
St. Lucia
509
26
Bahamas
0.27
Jamaica
11.0
Guyana
602
25
Suriname
0.42
Bahamas
14.0
Haiti
1,642
24
Guyana*
0.82
El Salvador
21.0
Barbados
1,717
23
Trinidad & Tobago*
1.26
Belize
23.0
Nicaragua
2,590
22
Jamaica*
2.53
Haiti
28.0
Bahamas
3,053
21
Panama*
2.63
Dom. Republic
49.0
Honduras
3,378
20
Uruguay*
3.19
Costa Rica
51.0
Jamaica
4,171
19
Costa Rica*
3.07
Panama
76.0
Trinidad & Tobago
5,707
18
Nicaragua*
4.54
Guatemala
109.0
Bolivia
6,496
17
Paraguay*
4.83
Honduras
112.0
Panama
6,570
16
El Salvador*
5.64
Nicaragua
130.0
El Salvador
6,674
15
Honduras*
5.95
Suriname
163.0
Costa Rica
7,027
14
Haiti*
7.18
Uruguay
177.0
Paraguay
7,177
13
Dom. Republic*
7.91
Guyana
215.0
Dom. Republic
7,341
12
Bolivia*
8.06
Ecuador
284.0
Guatemala
9,706
11
Guatemala*
10.62
Paraguay
407.0
Uruguay
11,431
10
Ecuador*
11.46
Chile
757.0
Ecuador
15,132
9
Chile
14.20
Venezuela
912.0
Peru
47,618
8
Venezuela
21.64
Bolivia
1,099.0
Chile
48,326
7
Peru*
23.53
Colombia
1,139.0
Colombia
56,379
6
Canada*t
29.61
Peru
1,285.0
Venezuela
64,980
5
Argentina*t
34.59
Mexico
1,958.0
Argentina
196,949
4
Colombia*
35.10
Argentina
2,767.0
Mexico
250,936
3
Mexico*t
94.78
Brazil
8,512.0
Brazil
432,433
2
Brazil*
159.22
United States
9,809.0
Canada
608,658
1
United States*f
263.06
Canada
9,976.0
United States
6,173,900
* Major sugar producing and exporting countries
t Major HFCS producers
Source: Adapted from Bernal (1998)
of these CARICOM countries are relatively small, open and undiversified economies
with over one-quarter of their total exports concentrated on one or two products. In the
case of countries such as St. Kitts and Nevis, Dominica and St. Vincent and the

17
Grenadines and St. Lucia, one primary exported commodity, either raw sugar or bananas,
accounts for more than 50% of exports.
Review of World and Western Hemisphere Sugar Markets
Sugar4 is one of the few commodities that can be produced from crops grown in
both temperate and tropical climates. A consequence of this is that widespread
governmental interventions have long been a feature of the world sugar market, making it
one of the most volatile of all primary commodity markets (Borrell and Duncan, 1993).
To gain some independence from the volatile world market, producers in most countries
have lobbied their governments to operate intervention schemes aimed at controlling
domestic prices, supply, and demand. Numerous pricing tiers, and trade and stockholding
mechanisms (policy wedges) have been devised and have resulted in distorted
production, trade, consumption, and world price. The trends in real world and US market
prices of raw and refined sugars are shown in Figure 2.1 for the period 1985-98. Among
other things, Figure 2.1 reveals the relative volatility of prices in the world market
compared with those in the US market.
World Production and Market TrendsSugar
Global sugar output in 1998-99 is projected by the US Department of Agriculture
(USDA) to increase by about 1.14 million metric tons (MMT) (0.9 percent) to 126.5
MMT, whereas consumption is projected to increase by 0.4 percent to 127.5 MMT,
reflecting some degree of market tightening. The five leading world sugar producers and
4 In this study, the word sugar refers to sucrose obtained from sugarcane and sugar beets.

18
consumers in 1997-98 are shown in Table 2.2. The European Union, followed by Brazil,
continues to dominate production. The United States is listed as the fifth largest sugar
producer, reflecting growth in domestic sugar beet production. In terms of consumption,
the leading country is India, followed by the European Union and China. The United
States occupies the fourth position in spite of its massive consumption of HFCS. As seen
from the Table, both production and consumption are relatively concentrated, with
Fig. 2.1. World and US refined and raw real sugar prices, 1985-98 (1992=100)
Source: Compiled with USDA sugar data and GDP deflator
the top five producing countries (out of 120 sugar-producing countries) and top five
consuming countries (out of all countries) accounting for 53 percent and 45 percent of
global production and consumption, respectively. Finally, it should be observed that
developing countries occupy three of the top five positions in both cases, reinforcing the
importance of this commodity to these economies.

19
Table 2.2. Leading World Sugar Producers and Consumers, 1997-98
Producers
Consumers
Country
Quantity
(MMT)
% of World
Production
Country
Quantity
(MMT)
% of World
Consumption
European Union
19.3
15.4
India
16.7
13.1
Brazil
16.6
13.2
European Union
14.4
11.3
India
14.5
11.6
China
9.0
7.0
China
8.6
6.8
United States
8.9
7.0
United States
7.3
5.8
Brazil
8.8
6.9
World
125.4
100.0
World
126.9
100.0
Source: USDA, Sugar and Sweeteners Situation and Outlook Yearbook.
Only a relatively small proportion of the large world production of sugar is traded
each year. Schmitz (1995, p 54) notes that it is common to divide the market for sugar
into three distinct markets: 1) the domestic market within the sugar-producing countries.
This market, he contends, is the largest and accounts for about 75% of all world sugar
production; 2) the market for the various international agreements between certain
importers and certain exporters. These agreements include the import quotas under the
US program, the bilateral agreements between Cuba and the former Soviet Republics and
the agreements of various countries and groups of countries such as the ACP with the
EU. This market makes up about 10% of world production; and 3) the market for the
residual free market in world sugar, which accounts for the remaining 15% of world
production.
For the fiscal year 1997-98, approximately 35.6 MMT (28.3 percent of world
production) of sugar was traded globally (Table 2.3). The five most important sugar
exportersBrazil, European Union, Australia, Thailand, and Cubaaccounted for 65.1
percent of global exports. Imports, on the other hand, have been less concentrated, with
the share of the top five sugar-importing countries and regions in 1997-98Russia,
United States, European Union, Japan, and Koreaamounting to only 30.1 percent.

20
Table 2.3. Leading World Sugar Exporters and Importers, 1997-98
Exporters
Importers
Country
Quantity
(MMT)
% of World
Exports
Country
Quantity
(MMT)
% of World
Imports
Brazil
7.2
20.2
Russia
3.8
10.7
European Union
6.2
17.4
United States
2.0
5.6
Australia
4.6
12.9
European Union
1.8
5.1
Thailand
2.9
8.1
Japan
1.6
4.5
Cuba
2.3
6.5
Korea
1.5
4.2
World
35.6
100.0
World
35.6
100.0
Source: USD A, Sugar and Sweeteners Situation and Outlook Yearbook.
Owing to domestic government policies, significant changes have occurred in the
status of many of the major players in the market. For example, the European Union has
changed its status from being a net sugar importer through 1976 to becoming the worlds
largest exporter. In the United States, sugar imports have declined considerably from a
share of 20 percent of world imports in the 1970s to the current 5.6 percent. As noted by
Hannah and Spence (1997), the United Stateswhich until 1975 had been the largest
sugar importer by a wide margin-had slipped to second position during the 1975-80
period and to fifth position by 1994. Japanonce the worlds second largest importer of
sugarhas reduced its imports considerably, owing mainly to domestic support policy.
Table 2.3 however, does not reflect the significant change that has occurred in the
structure of the global sugar import market. The market was previously dominated by
developed countries but is now dominated by developing countries. Indeed, Hannah and
Spence (1997) reported that in 1994 the developing countries accounted for
approximately 68 percent of net imports, compared with only 25% in 1975. These
authors noted that this change in the structure of global import demand has implications
for the extent of the price volatility since, unlike the developed countries, these countries

21
have relatively low income and high price elasticities. Consequently, the change has
resulted in greater stability of world market prices.
Western Hemisphere Production and Market TrendsSugar
Of the thirty-four countries comprising the FTAA, twenty-five are major sugar
producers/exporters. Thus in the 97/98 production year the Hemisphere produced about
39.61 MMT or about 31.6% of world production, estimated at about 125.38 MMT.
Consumption of sugar within the Hemisphere was estimated at 31.38 MMT, which
represented 24.7% of world consumption. This implies that the region5 is a net surplus
producer of sugar of approximately 8.23 MMT. If production of sugar in Cuba6 were to
be added, output would increase by another 3.0 MMT bringing the regional surplus to
10.54 MMT. Table 2.4 shows the top six sugar producing countries plus CARICOM and
the top seven consuming countries in the region together with their respective market
shares. On the production side the largest producer is Brazil with a share of 39.6%,
followed by the US and Mexico with shares of 18.4% and 13.9%, respectively. Total
CARICOM production, comprising output from Barbados, Belize, Guyana, Jamaica, St.,
Kitts and Nevis, and Trinidad and Tobago, amounted to less than one million metric ton
in the 97/98 fiscal year7. This amount represents about 1.9% of regional output and less
than 1% of world output.
5 In the document the word region will be used to refer to the Western Hemisphere, and the subregion
will be used to refer to the CARICOM countries, unless otherwise define.
6 Cuba is excluded from the prospective countries slated to form the FTAA.
7 These countries are referred to as the CARICOM sugar quota holders and comprise the Sugar Association of
the Caribbean (SAC).

22
On the consumption side the US and Brazil were the main consumers each
accounting for about 28% of regional consumption. Next in line were Mexico, Argentina,
Columbia and Canada, respectively. The fourteen CARICOM countries consumed 0.32
MMT or about 1% of regional consumption.
Table 2.4. Leading Hemispheric Sugar Producers and Consumers, 1997-98
Producers
Consumers
Country
Quantity
%of
Country
Quantity
%of
(MMT)
Hemispheric
(MMT)
Hemispheric
Production
Consumption
Brazil
15.70
39.6
US
8.91
28.4
US
7.28
18.4
Brazil
8.80
28.0
Mexico
5.49
13.9
Mexico
4.24
13.5
Columbia
2.15
5.4
Argentina
1.45
4.6
Guatemala
1.79
4.5
Colombia
1.37
4.4
Argentina
1.75
4.4
Canada
1.27
4.0
CARICOM8
0.75
1.9
Peru
0.91
2.9
a] Represents the total output of six CARICOM countries
Source: International Sugar Organization
With respect to the hemispheric trade, in the 97/98 period total exports amounted
to 12.77 MMT representing about 35.9% of world exports, estimated at 35.59 MMT. In
contrast, total imports amounted to just 4.46 MMT or about 12.5% of world imports.
Consequently, the region is a net exporter of 8.3 MMT (excluding net exports from Cuba
of 2.3 MMT). Table 2.5 shows the leading sugar exporters and importers in the region.
The Table also shows exports as a percentage of domestic production and imports as a
percentage of consumption. The leading sugar exporters within the region are Brazil,
Guatemala, Mexico and Columbia, respectively. Of interest however, is the fact that the
CARICOM exports, while accounting for only 5.1% of total regional exports, represent
more than 85% of their domestic production--an indication of the importance of this
commodity in trade for these countries.

23
Table 2.5. Leading Hemispheric Sugar Exporters and Importers, 1997-98
Exporters
Importers
Country
Share (%)
of Hemispheric
Exports3
Exports as a %
of Domestic
Production
Country
Share (%)
of Hemispheric
Imports
Imports as a %
of Domestic
Consumption
Brazil
56.4
45.9
US
44.1
22.1
Guatemala
10.7
76.0
Canada
27.3
95.7
Mexico
10.0
23.3
Peru
5.8
28.2
Columbia
6.5
38.5
Venezuela
5.4
30.8
SACb
5.1
86.5
Chile
4.9
32.4
Dominican Rep.
2.3
51.2
Ecuador
2.9
32.5
El Salvador
1.6
45.4
Uruguay
2.1
86.4
a] Include re-exports
b] Total exports from six CARICOM countries
Source: International Sugar Organization
In terms of regional imports, the United States and Canada dominated the trade in
97/98. Together these two countries accounted for in excess of 70% of the region's sugar
imports. Peru and Venezuela were the other major importers, accounting for about
11.2% of the import market. The fourteen CARICOM countries import a total of about
0.1 million metric ton. This amount represents about 2.2% of the overall regional imports
and less than a third of the total CARICOM sugar requirements. The bulk of such imports
was from the world market and in some cases was done in order that the exporting
country would be able to satisfy its sugar export quota commitments.
Importance of the Sugar Industry to CARICOM Sugar Producers
Socioeconomic Importance
The sugar industry is of considerable importance to CARICOM countries. This is
particularly true for the six members that comprise SAC. Among other things, the
industry contributes substantially to valuable foreign exchange earned by these countries.
For example, in St. Kitts and Nevis the industry provides more than one-half of that

24
countrys foreign exchange earnings. In Jamaica, sales from exports of sugar were in
excess of US $100 million in 1996 (Caribbean Update, 1998). In the subregion, the
industry directly employs about 97,000 persons, or about 5 percent of the active working
population, and contributes appreciably to the Gross Domestic Product (GDP) of the
respective countries (Table 2.6). Moreover, the socioeconomic importance of sugar
production to these countries is much more than these figures might suggest since these
industries play a vital role in the maintenance and upgrading of rural infrastructure,
health, recreational, and educational facilities (James, 1998).
Table 2,6. Socioeconomic Importance of CARICOM Sugar Industries
Country
Revenue
as % of GDP
Revenue
As % of Agrie.
Production
Foreign Earnings
as % of Total
% of Active
Work Force
Hectares
Cultivated
Barbados
1.7
33.2
17.0
3.6
11,700
Belize
10.5
61.1
32.8
45.3
24,000
Guyana
26.0
41.0
27.7
9.6
42,000
Jamaica
2.6
29.7
8.0
4.1
45,000
St. Kitts/Nevis
49.0
85.0
64.1
12.5
4,000
Trinidad &
Tobago
1.4
67.0
1.6
3.8
19,000
Source: James 1998
Historically, sugar in the Caribbean has been marketed under preferential trading
arrangements. Currently, these arrangements are the Lome Agreement (via the Sugar
Protocol) and the US Preferential Quota (via the Sugar Import Tariff Quota). In 1997-98,
of total SAC countries sugar output of 0.75 MMT (raw sugar) produced, 0.66 MMT
(86.5 percent) were exported. Of this amount, 79.9 percent went to the European Union,
and 11.3 percent went to the United States. The remaining 8.8 percent of sugar exports
from CARICOM were sold on the subregional and world markets, with each accounting
for an equal share. Prices obtained in both the EU and US markets were significantly
higher than the world market price (James, 1998).

25
The EU/Lome ACP Agreement
The EU provides special support to ACP countries, which includes CARICOM,
through a series of Protocols annexed to the Lome Convention. These Protocols give free
access to EU for fixed quantities of bananas, sugar, beef, veal and rum. Producers usually
receive EU prices, which, by virtue of the Unions Common Agricultural Policy (CAP),
are usually two to three times the world market price. Of all the Protocols, the Sugar
Protocol is considered unique in that it is characterized by its indefinite period, and its
claimed autonomy. Signatory countries benefit from a formal contractual commitment
by the EU to buy specific quantities of cane sugar at guaranteed prices. The quota is fixed
at about 1.3 MMT (of which 0.43 MMT is allocated to CARICOM) and ACP countries
are permitted to import sugar for domestic purposes if their surplus in a year is
insufficient to fill the quota. Additional preferential access was granted to ACP sugar
producers on July 1995, for a six-year period, under the Special Preferences Sugar (SPS)
quota bringing the total allocation to the CARICOM subregion to 0.53 MMT (James,
1998). Table 2.7 gives an indication of the real earnings and the quota rent8 from sale of
sugar to the EU market. The values are calculated on the basis of only the main quota
(does not include shipment under the SPS) and assumes SAC countries fulfill their
quotas.
Trade liberalization puts at risk the benefits of the Protocol to the signatory parties
and by implication its very existence. One problem stems from the fact that the Lome
Convention was ruled in 1994 by a GATT panel to be inconsistent with GATT rules by
8 Quota rent is used here loosely to refer to the extra revenue (surplus) earned by the exporting country-in
excess of what the country would have earned selling at the world market priceIf om selling in a particular
market where the price is higher than world market price because of import restrictions.

26
virtue of its non-reciprocal nature (Blanco, 1997). In particular, the Arrangement was
ruled to be discriminatory vis--vis other LDCs and thus not covered by the enabling
clause permitting GSP schemes. This GATT ruling led to the EU seeking a waiver,
which was subsequently granted by the WTO. However, the non-reciprocity permitted by
Table 2.7. CARICOM/EU Market Real Earnings and Quota Rent (1992=100)
Year
SAC
QUOTA
(MMT)
UK Price
(Pounds
/MT)
Real UK
Price raw
(USS/MT)
Real World
Price raw
(US$/MT)
Total Real
Earnings
(US $M)
Quota Rent
(US $M)
90-91
0.4281
346.10
628.35
204.27
269.00
181.55
91-92
0.4281
351.12
587.79
199.98
251.64
166.03
92-93
0.4281
403.00
629.91
215.66
269.67
177.35
93-94
0.4281
400.28
587.39
252.06
251.47
143.56
94-95
0.4281
420.56
608.52
273.07
260.51
143.61
95-96
0.4281
445.58
649.58
240.39
278.09
175.18
96-97
0.4281
414.44
577.45
227.90
247.21
149.64
97-98
0.4281
364.64
512.96
177.08
219.60
143.79
Source: SAC data
the GATT/WTO waiver is contrary to Lome principles and is only valid until the end of
the present Convention in the year 2000. What this means in practical terms for the ACP
countries and CARICOM in particular, is that the Sugar and Banana protocols will have
to stand alone as either viable WTO-compatible commercial arrangements between the
EU and the nations to which they apply, or they will disappear completely. The WTO
1999 negotiations will undoubtedly include strong calls for free trade in sugar and the
demise of the Protocol.
A second factor which stands to reduce the potency of the Sugar Protocol,
irrespective of the shape or form of any future EU-ACP Agreement, is the reductions in
domestic support measures by the EU as it seeks to implement its WTO liberalization
commitments. Firstly, as the EU implements its agreed tariff reductions, access by third

27
(non-ACP) countries, presently granted Most Favored Nation (MFN)9 status, will be
enhanced thereby reducing the preference margin for exports from ACP. Secondly, as
support is reduced, particularly subsidization of exports to domestic producers in the EU,
the domestic prices paid to EU producers will begin to fall. This will in turn, trigger a
reduction in ACP prices since the Sugar Protocol guarantees that ACP quota holders will
be paid the same prices as EU producers.
Other factors which will or have begun to erode the value of the EU/ACP Sugar
Protocol include: 1) the pre-EU membership arrangements with the countries of Central
and Eastern Europe some of which are sugar producers; 2) a surge in EU centered
regional trade arrangements, in particular, preferential agreements in the pipeline with
some of the sugar producing countries in Latin America; 3) the fact that markets have
evolved significantly since the first Lome Convention in 1975, with the result that Europe
has become a leading sugar exporter; and 4) that cane sugar can be bought on the world
market at much lower prices than those guaranteed by the Protocol which now cost the
EU several hundred million Euros10 each year. A final issue that is unresolved is the
impact of the Euro, the EUs common currency, on the sugar prices that quota holders
receive.
The US Sugar Program and Policy
In spite of the less lucrative nature of the US market relative to the EU market, the
former nonetheless has provided CARICOM sugar producers with a much more viable
9 The Most Favored Nation (MFN) clause of a commercial treaty is a binding contract by the signatories to
confer upon each other all the most favored trade concessions that either may grant to any other nation
subsequent to the signing of the agreement.

28
alternative than selling sugar on the world market. It has also permitted the subregional
sugar industry to survive and, through inter-industry economic linkages, has fostered
growth in services and other sectors. Table 2.8 gives a rough indication of the real quota
rent earned by CARICOM as a result of the US program over the period 1990-98. The
data are based on the assumption that CARICOM fulfilled its quota commitment to the
US, which is approximately 4.8 percent of the total tariff rate quota (TRQ)11. Also, no
Table 2.8. CARI COM/US Market Real Earnings and Quota Rent (1992=100)
Year
TOTAL
TRQ
(MMT)
SAC
TRQ
(MMT)
Real US
Price raw
(US$/MT)
Real World
Price raw
(US$/MT)
Total Real
Earnings
(US $M)
Quota Rent
(US $M)
90-91
2.1027
0.1051
487.41
204.27
51.24
29.77
91-92
1.3863
0.0693
468.82
199.98
32.50
18.63
92-93
1.1347
0.0567
464.86
215.66
26.37
14.14
93-94
1.1347
0.0567
458.00
252.06
25.98
11.68
94-95
1.3241
0.0662
466.49
273.07
30.88
12.81
95-96
1.9702
0.0985
439.92
240.39
43.34
19.66
96-97
1.9065
0.0953
414.98
227.90
39.56
17.83
97-98
1.4545
0.0727
403.56
177.08
29.35
16.47
Source: Compiled from USDA data.
adjustments have been made to accommodate shipping expenses. As can be seen, the
amounts of quota rents in real terms are substantial and their loss would have serious
negative social repercussions. However, as indicated in Fig. 2.1, prices have trended
upward in the world market and downward in the US market during the period. An
obvious implication of this observation is that the real premium enjoyed by quota holders
and US producers has been declining. As will be seen later, this factorcoupled with a
10 The Euro is currently valued at approximately 1 Euro = US $1.06.
11 The Tariff Rate Quota (TRQ) system is a key component of the current US sugar program. This
component of the program is elaborated on below.

29
reduction in the quota levelsimplies a considerable loss of real earnings for the quota
holders.
A stated objective of the US Sugar Program is to ensure a reliable supply of sugar to
American consumers at competitive prices, while providing some stability for American
sugar interests (growers and processors). Key components of the program include: 1)
loan rates; (2) an implicit market stabilization price (MSP); and 3) Tariff-Rate Quota
(TRQ). These components include:
1. Loan rates-a loan rate of US $0.18 per pound of raw sugar is made available to
processors. The loan rate of 22.9 cents per pound for refined beet sugar was
historically set in relation to raw sugar with a prescribed formula. Loans were
previously non-recourse in that, at any time, the processors could refuse to repay the
loan and the government would accept the sugar in lieu of cash. However, with the
implementation of the Federal Agricultural Improvement and Reform Act (FAIR) of
1996 current loan rate programs are to operate on a recourse basis under certain
conditions. Whenever the tariff-rate quota (TRQ, discussed below) is 1.36 MMT or
less, the USDA can demand repayment of the loan with interest at maturity,
regardless of the price of sugar. The loans are considered to be non-recourse only
when the TRQ is above the stipulated 1.36 MMT. In addition, loan rates were fixed at
the 1995 levels of 18.0 and 22.9 cents per pound for raw and refined sugar,
respectively.
2. Market stabilization price (MSP)a major stipulation of the program in the past was
that it should run at no cost to the government. To ensure that such provision was
met, an MSP was explicitly established at a level to ensure that commercial market

30
channels were cleared rather than forfeited to the Commodity Credit Corporation
(CCC). The MSP was set at a level of $0.22 per pound of raw sugar. However, under
the FAIR program the no net cost provision was eliminated and the MSP is no longer
set explicitly although it is expected that supply would be managed in such a way as
to ensure the market price would be above the loan rate.
3. Tariff-rate quota (TRQ)-the TRQ for sugar and sugar-containing products is not
technically a part of the domestic sugar support legislation (Buzzanell, 1998).
Nonetheless, it is one of the principal tools used to support the loan program and to
ensure that the domestic price does not come within the forfeiture range. The US
Trade Representative (USTR) allocates the TRQ for raw sugar to 40 countries. The
total quota varies from year to year in keeping with the objective to stabilize domestic
raw-sugar prices. Although the quota can vary from year to year, it cannot fall below
1.14 MMT, which is the commitment given by the United States under the Uruguay
Round of the GATT (USD A, Situation and Outlook Series, 1998). Of the 40 countries
participating in the program, 23 are from the Western Hemisphere and include all the
SAC sugar exporters. Country allocations are based on historic trade with the United
States. Haiti, and St. Kitts and Nevis were included in the allocations as minimum
quota-holding countries, and the allocation to Mexico was increased to fulfill
obligation pursuant to North American Free Trade Agreement (NAFTA). About 66
percent of the raw cane sugar TRQ are allocated to the hemispheric countries. The
Dominican Republic, Brazil, CARICOM, and Argentina account for the bulk of the
overall regional quota with TRQ shares of 17 percent, 14 percent, 5 percent, and 4.6
percent, respectively. All countries in the Hemisphere, with the exception of Brazil,

31
receive preferential treatment (zero-tariff) in that they are able to sell their in-quota
sugar to the United States at the established US domestic price and retain the export
quota rent. In the case of Brazil, a small tariff of 0.625 cent/pound is charged on the
in-quota exports. Imports outside of the quota have been subject to a high tariff of
17.62 cents per pound since January 1, 1995. In accordance with GATT/WTO
commitments, this tariff is expected to decline to 15.36 cents per pound by the year
2000.
Since the quota is reserved for selected countries and thus can be considered
discriminatory towards the LDC countries, there is the view that the TRQ will be
seriously questioned in the 1999 WTO review round (Contrearas, 1998). In addition,
there is increasing pressure within the US to eliminate the current sugar program in
keeping with the Federal Agricultural Improvement and Reform (FAIR) Act of 1996,
also known as the Freedom to Farm Act. FAIR seeks to eliminate or greatly reduce
government intervention into the agriculture sector as a means of making the sector more
competitive. While the issue of the elimination of the US sugar program is somewhat
debatable, there is the unconfirmed view that the once strong lobby, comprising the
growers of cane and beet sugar and the manufacturers of caloric and non-caloric
sweeteners, is losing its political clout, in part because manufacturers of HCFS no longer
need the protection accorded in order to make their operations viable.
Review of World and Western Hemisphere HFCS Markets
High Fructose Com Syrup is a liquid caloric sweetener made from ordinary
cornstarch. It can be substituted for sugar (sucrose) in most liquid uses. As a consequence
of its relative cheapness in comparison to other forms of caloric sweeteners, it has been

32
used in a wide range of processed food products such as beverages, baked goods, dairy
products and jams and jellies. Although converting cornstarch into sweet substances was
discovered as early as 1811, it was not until the late nineteen-century that sweeteners
such as glucose com syrup and dextrose were produced in com wet milling. However,
these products were only about 70% as sweet as sugar and hence were not competitive
with sugar on a sweetness equivalency basis. The search to find a sweetener comparable
to sugar did continue and although promising results were obtained in the early 1960s
when scientists discovered an effective method of obtaining fructose from glucose a
substance which is 110 to 170 percent sweeter than sucrose the break through did not
occur until 1967 and was based on research carried out in Japan and the US. However,
commercialization of the product did not occur until 197212.
HFCS comes in two strengths, HFCS-42 and HFCS-55. The numbers indicate the
strength of the fructose percentage. Thus, in the case of HFCS-42, the product contains
42% fructose, and about 50% and 8% of dextrose and other saccharides, respectively.
This product is approximately 90% as sweet as sugar. HFCS-55 contains 55% of fructose
and about 40% and 5% of dextrose and other saccharides, respectively. This product has
sweetness of about 110%. Commercial production of HFCS-42 began in 1972 while that
of HFCS-55 began in 1977. In 1985, through further processing of HFCS-55 a crystalline
form of the product was prepared for commercial use. However, certain technical and
economical blotches still hamper its manufacture and limit its widespread use as a direct
substitute for crystallized (tabletop) sugar. The main problems have to do with its still
relatively high cost of production compared with sugar and the fact that its sweetness
12 For details on the invention process of the product and facts about its diffusion see Zitt (1998).

33
appears to vary depending upon the particular use (Polopolus and Alvarez, 1991;
Thomas, 1985).
Although both syrups share certain favorable features such as stability, high
osmotic pressure or crystallization control, each offer special qualities to food
manufactures and consumers. Thus HFCS-42 is popular in canned fruits, condiments and
other processed foods which need mild sweetness that won't mask natural flavors. The
sweeter HFCS-55 is used mainly in the soft drinks industry, in ice cream and frozen
desserts. The main drawbacks to HFCS are that it is available mainly in liquid form, and
therefore restricted to only certain industrial uses. In addition, due to its high water
content it costs more to transport than an equivalent quantity of sugar, and it is difficult to
handle, since it must be maintained at 80 -100 F temperature when stored or
transported (American Sugar Alliance, 1998).
World Production and Market TrendsHFCS
Table 2.9 shows world HFCS production for the period 1989-1997. It shows that
in 1997 world output of HFCS amounted to 10.41 MMT which is about 8.5% of the
combined world output of sugar and HFCS. The US dominates production with output of
7.71 MMT or 74% of total output in 1997. Japans production of 0.8 MMT accounted for
another 7.6% of total output. Thus, together these two countries accounted for almost
82% of world output. It should be pointed out however, that while production of HFCS in
the US is unrestricted, production levels in both Japan and the EU (third largest producer)
are restricted.

34
Owing to the problems alluded to earlier with regards to storing the commodity,
very little is held in stock. Of particular interest is the fact that, between 1990 and 1997,
world consumption of HFCS increased at an annual rate of 4.0% while consumption of
sugar grew at a lackluster rate of 1.5% per annum. With the demand growth for sugar
failing to keep pace, HFCS has captured an increasing share of the combined sugar
market for these sweeteners, increasing from 7.2% in 1990 to an estimated 8.5% in 1997
(LMC International, 1998).
Table 2.9. World HFCS Production, 1989-97 ('000 metric tons dry basis)
Country
1989
1990
1991
1992
1993
1994
1995
1996
1997
US
5,345
5,677
5,852
6,038
6,459
6,813
7,143
7,408
7,711
Japan
744
783
794
747
727
806
789
792
800
EU
276
280
284
286
288
290
303
303
303
Korea
244
270
276
263
246
260
277
279
280
Canada
239
245
252
250
255
255
265
260
295
Argentina
146
161
184
184
195
207
190
198
210
Taiwan
51
67
110
125
150
171
182
195
210
Mexico
0
0
0
0
0
0
0
110
175
Others
156
210
197
229
258
308
354
388
423
World
7,201
7,693
7,950
8,122
8,579
9,109
9,493
9,933
10,407
Adapted from LMC International, (1998)
Currently, only a small amount of HFCS output enters into trade. As indicated
earlier, this is due to the technical difficulties and high costs associated with transporting
the liquid sweeteners. Consequently, trade is confined mainly to cross-border
transactions. In this regard, the only region where considerable quantities are traded
internationally is in North America.
Western Hemisphere Production and Market Trends-HFCS
With respect to the Western Hemisphere, total output of HFCS in 1997 amounted
to 8.43 MMT, representing 81.0% of world production. The main producing countries

35
within the region are the US (91.5%), Canada (3.5%), Argentina (2.5%) and Mexico
(2.1%). Production in Mexico started in 1996 by US interests and already output level is
close to that of Argentina and Taiwan. At the moment, the bulk of the com used in the
production of this commodity in Mexico is imported from the US. The 1997 regional
consumption of HFCS was estimated at 8.32 MMT. Again the US was the main
consuming country accounting for 91.1% of total demand. This was followed by Mexico
(3.7%), Argentina (2.5%) and Canada (2.5%). Only a small quantity of HFCS is currently
consumed in CARICOM and none of the member states engage in the production of the
commodity.
The considerable success in HFCS as a substitute for sugar is best illustrated in
the case of the US. Table 2.10 shows the changes in the per capita consumption and the
relative market shares of caloric sweeteners consumed in the US over the period 1975 to
1997. And, Figure 2.2 shows a comparison of the trends in per capita consumption of
HFCS, total com sweeteners and sugar for the same time period. In particular, the Figure
reveals that between 1975 and 1997, the per capita consumption of sugar fell from 89.2
lbs. to 67.1 lbs. after reaching a low of 60.8 lbs. in 1986, while the per capita
consumption of HFCS increased from 5 to 61.4 lbs. over the same period. This meant that
whereas in 1975 sugar accounted for about 76 % of the caloric market shares and HFCS
only 4%, by 1997 the market share of sugar had fallen to 43% while in the case of HFCS
it had risen to 40%. The rapid and considerable increase in the production and
consumption of HFCS in the US and the concomitant displacement of a portion of the US
sugar demand, has come largely at the expense of the sugar quota holders and by
implication the US sugar refiners. This is easily seen when one considers the fall in US

36
Table 2.10. Changes in US Per Capita Sweetener Consumption and Relative Market
Shares, (1975 1997) Pounds, Dry Weight Basis
Year
Refined
Sugar
Relative
Share (%)
HFCS
Relative
Share (%)
Total Corn
Sweet'rs
Relative
Share (%)
Honey & Edible
Syrups
Relative
Share (%)
Total Caloric
Sweet'rs
1975
89.2
76
5.0
4
27.5
23
1.4
1
118.1
1976
93.4
75
7.2
6
29.7
24
1.3
1
124.4
1977
94.2
74
9.5
7
31.2
25
1.4
1
126.8
1978
91.4
72
12.1
10
33.7
27
1.5
1
126.6
1979
89.3
70
14.9
12
36.4
29
1.4
1
127.1
1980
83.6
67
19.1
15
40.2
32
1.2
1
125.0
1981
79.4
63
23.2
19
44.5
36
1.2
1
125.1
1982
73.7
60
26.7
22
48.2
39
1.3
1
123.2
1983
71.1
57
30.7
25
52.2
42
1.3
1
124.6
1984
67.6
53
36.3
29
57.8
46
1.4
1
126.8
1985
63.2
49
44.6
35
63.9
50
1.5
1
128.6
1986
60.8
48
45.1
36
64.6
51
1.6
1
127.0
1987
63.1
48
47.1
36
66.8
51
1.7
1
131.6
1988
62.6
47
48.3
36
68.3
52
1.5
1
132.4
1989
62.8
47
47.5
36
68.0
51
1.6
1
132.4
1990
64.8
47
49.2
36
70.4
51
1.6
1
136.8
1991
64.4
47
50.0
36
72.0
52
1.6
1
138.0
1992
64.4
46
51.6
37
74.3
53
1.6
1
140.3
1993
64.6
45
54.4
38
77.8
54
1.6
1
144.0
1994
65.8
45
56.4
38
80.2
54
1.5
1
147.5
1995
66.2
44
58.4
39
82.6
55
1.5
1
150.3
1996
66.9
44
59.8
39
84.2
55
1.5
1
152.6
1997
67.1
43
61.4
40
86.3
56
1.5
1
154.9
Source:USDA, ERS, Commodity Economic Division, Sugar and Sweetener: Situation and Outlook Report,
various issues
Fig. 2.2. Changes in US per capita consumption of selected sweeteners, 1975-97

37
sugar imports from a level of 5.0 MMT in the seventies to current level of about 1.6
MMT (Hannah fe Spence, 1997).
The growth of the industry in the US is attributed to several factors including: 1) a
marketing environment with no restriction on supply; 2) advancements in technology;
and 3) the relative cheapness of the product in relation to sugar on a sweetness
equivalency basis. Since its inception in the US the product has sold at a discount to
sugar. Figure 2.3 shows a comparison of the real price of HFCS-42 and the real world
and US prices of refined sugar. The evidence indicates that the pricing of HFCS is now
much closer to that of the world market price of refined sugar.
Fig. 2.3. Real HFCS and refined sugar prices, 1982-98 (1992=100)
Source: USDA, ERS, Commodity Economic Division and Dept, of Commerce,
Bureau of Economic Analysis

38
As noted before, most of the international trade in HFCS occurs within North
America. Although the Canadian and the US-HFCS industries are difficult to separate,
from a custom point of view, Canada is a net exporter to US, which in turn is a net
exporter to Mexico. The growth in this trade is reportedly due to increases in the Mexican
demand (USDA, Situation and Outlook Report, 1998). However, owing to a dispute
concerning the legitimacy of the US-Mexican Side Agreement on trade in sugar under the
North American Free Trade Agreement (NAFTA), the Mexicans have recently (January
1998) placed a 100% tariff on imports of HFCS from the US. The issue is now being
addressed by both the WTO and the NAFTA dispute settlement bodies. The tariff has
retarded the growth in trade but the US has maintained current levels of exports to
Mexico. Total exports of HFCS from the US increased by 37 % in 1998 and now account
for as much as 6.2 % of domestic production compared with 3.7% and 4.7 % in 1996 and
1997, respectively.
Finally, even with the shipping problems of the liquid product, an increasing
amounts of HFCS are currently being shipped to markets such a Japan and Trinidad and
Tobago (USDA Situation and Outlook Report, 1997). Industry experts believe that
shipping problems are surmountable. Within CARICOM, Trinidad and Tobago is the
only country, which currently imports HFCS for use in its soft drink industry, however
there is the possibility that other countries might soon follow.

CHAPTER 3
LITERATURE REVIEW AND THEORETICAL BACKGROUND
Several studies have been done on modeling the US Sugar/Sweetener subsector
and drawing economic/welfare implications for various groups including domestic sugar
producers, processors, sweeteners' users, consumers and trading partners. Most of these
studies have either not taken into consideration the impact of HFCS in their analysis or
have only done so in a cursory manner. The first section of this chapter therefore
undertakes a critical review of those studies in which the HFCS industry was taken into
consideration. The review provides a basis for the conceptual model of the US sugar and
HFCS subsector developed in chapter 4. Because of the problem of time-varying
parameters in the HFCS industry and the need to provide more reliable parameters for the
HFCS industry, the second section of the chapter focuses on some critical time series
issues such as serial correlation, unit root and cointegration. The methodology of state-
space model and the Kalman filter, allows for estimation of time-varying parameters and
is employed later in estimating the derived demand for HFCS. Consequently, this
methodology is discussed in detail. The chapter ends with a discussion on the nature of a
derived demand and the general approaches used to estimates such functions.
Previous Empirical Studies
Sugar and HFCS Related Studies
One of the earliest studies to include HFCS in the analysis of the US sweetener's
market is a study conducted by Carman (1982). This study had as its objective the
39

40
projection of the rate of adoption of HFCS in the US and an assessment of its economic
implications for major sweetener industry participants. The study used a logistical trend
model with estimates of a ceiling market share of HFCS and total demand for caloric
sweeteners to carry out the projections. Among other things, the study concludes that the
total quantity of sugar demanded in the United States would decrease for several years as
HFCS demand increases. Also, that the impact of the HFCS on domestic sugar producers,
under the restrictive sugar policy, would be minimal since the bulk of the costs would be
borne by those countries exporting sugar to the US. Carman assumed that the saturation
of the HFCS sweetener market would occur between 25 -30% of the total caloric
sweetener demand. This assumption has proven to be too conservative since HFCS
currently accounts for as much as 40% of the caloric sweeteners' market (Table 2.10).
Lopez and Sepulveda (1985) developed and estimated a demand model for sugar
and on the basis of the models results derived implications for sugar import policies. In
that model they addressed what they considered to be some shortcomings of earlier
studies. They: 1) estimated separate demand functions for industrial and nonindustrial
sectors; 2) compared the periods before and after the introduction of high fructose com
syrup in industrial uses; and 3) utilized a partial adjustment framework in an attempt to
measure the speed of demand adjustment. Their empirical results indicated that: 1)
changes in consumer preferences and the availability of the cheaper HFCS in the food
processing segment were exerting downward pressure on sugar demand; and 2) the
demand for sugar had become less responsive to sugar price change after the introduction
of HFCS. They estimated that the short-run own price elasticity went from -0.15 to -
0.04 and the long-run (full adjustment) own price elasticity went for -0.31 to -0.06 after

41
the introduction of HFCS. They attributed the decreased in own price elasticity of sugar
to the substitution taking place in the more price elastic segment of the sweetener market
such as the soft drinks and canning industries. They concluded that as the US demand for
sugar decreases and the food industry adjusts faster to sweetener choices, the US
government would have to impose more restrictive import barriers to maintain prices to
domestic sugar and HFCS producers.
Leu, Schmitz, and Knutson (1987) used a general equilibrium model to analyze
the policy options for the US Sugar program and to empirically estimate the
substitutability between sugar and HFCS by including the price of HFCS as a
determinant of sugar demand. They used a composite price of com sweeteners glucose,
dextrose and HFCSin order to generate a series dating back to 1955. Demand
elasticities were estimated for sugar and HFCS while supply elasticities were obtained
from previous studies. Their results indicated that the net social cost of the quota program
with an elasticity of excess supply of 2.37 was $203 million when substitution with
HFCS was allowed and $253 million with no HFCS substitution. Under the assumption
of a perfectly elastic excess supply curve the net social cost with HFCS substitution was
about $1 billion. They concluded that a deficiency payment program would be more
costly than an import quota. However, Marks (1993) stated that the study suffered since
price data on the HFCS were only available after 1975. He also questioned the composite
commodity assumption because glucose and dextrose are much less substitutable for
sugar than is HFCS.
Sudaryanto (1987) developed a trade model to analyze the effects of trade
liberalization by the US and EEC on international markets for sugar. In relation to the

42
US, he pointed out that analysts in previous sugar studies ignored the derived nature of
the demand for sugar by treating all sugars as if they were consumed directly. Utilizing
Fouriers flexible functional forms he separately estimated, a derived demand and direct
demand for sugar. He then computed an aggregate elasticity of demand for sugar using
the weighted average of elasticities obtained from the two models. While he did take into
consideration the impact of the possible substitution between HFCS and sugar in
estimating the various demand functions, he followed previous studies and utilized a
composite price index of HFCS, glucose and dextrose. The results of his analysis
indicated that in the United States, sugar for final consumption is far less elastic than
sugar used as intermediate input. Sudaryanto's results suggest that consumption for final
uses had reached the near saturation point and a change in sugar price had little effect on
consumption. In contrast, the demand for sugar in the food processing industry was more
elastic because of available substitutes. The result of his trade model showed that
unilateral trade liberalization in the US market would force an increase in US imports by
114% and would be accompanied by a decrease in both production and supply price by
8.3% and 11.2%, respectively. Complete removal of protection in the EEC was estimated
to cause a reduction in supply price and production in that region of 13.2% and 5.6%,
respectively.
Rendleman and Hertel (1989) examined the impact of policy changes on sugar
and HFCS within a general equilibrium framework. The results of their analysis showed
that any substantial increase in the production of com sweeteners would cause an
increase in the production of byproducts, which would have the effect of lowering the
byproducts prices and thus raising the cost of production of com sweeteners. They

43
estimated that free trade would result in the US prices for sugarcane, sugar beet and corn
sweeteners declining by 34%, 31% and 10%, respectively. Com sweetener byproducts
would decline by 2% but prices would increase by 12 %. In terms of the changes to
output, they estimated that US sugarcane production would decline by 77%, sugar beet
production by 47%, and com sweetener production by 2%. This study was an
improvement over many of the previous studies in that an attempt was made to seriously
analyze the interrelationship among the various segments of the sweetener market.
Tanyeri-Abur (1990) analyzed the implications of current and potential changes in
the US sugar policy with a price endogenous agricultural sector model. The sugar and
HFCS markets and the industries, which used these two sweeteners, were modeled.
Several policy runs were made under varying assumptions pertaining to the level of
imports, target prices and substitution relationship between sugar and HFCS. The results
were compared with the based year (1996) in which imports of raw sugar was restricted
to 1.6 million short tons. Her work indicated that: 1) the removal of US sugar quotas
would benefit US consumers by $657 million and increase net foreign surplus by $350
million; and 2) that US producers surplus would decline by $288 million leaving a net
welfare gain of about $718 million. However, when the agricultural sector with all
commodities was considered, there was a reduction in net social welfare for the sector of
$1.2 billion dollars despite an increase in total consumer surplus of $1.4 billion. She
explained that this result was due mainly to the increase in government payments that
would result as the price of cotton and rice fall as land was shifted from sugarcane to
these crops.

44
She conceded that conclusions in relation to the HFCS subsector could be a
reflection of rigid assumptions that were made in relation to the supply and cost of
production of HFCS. For example, an unlikely result is that HFCS does not effectively
compete with sugar and is replaced by sugar when free substitution is allowed. In further
analysis in the said study, however, she showed that the above result was highly sensitive
to the assumptions made with respect to the HFCS supply elasticities. She concluded that
further analysis is needed into the behavior of HFCS, but that, by and large, the signs of
the welfare change and the distributional implications would remain the same.
Marks (1993) constructed a partial-equilibrium model in which he assessed the
long-run static welfare effects of alternative US sugar policies. He evaluated two policies.
The first focused on estimating the overall economic impact of the US sugar program for
crop years (1984/85 to 1988/89)years in which the domestic-foreign price differential
varied widely. The analysis considered the long-run effect of the program on sugar prices
in the US and the rest of the world (ROW), the economic welfare of US and foreign
consumers and producers, and U. S. government revenues. His second policy analysis
examined the implication of a 10% real reduction in the US sugar loan rate, by which the
US provides support to its sugar producers. His approach to modeling the US sweetener
market differs from most of the previous studies in two ways. Firstly, he estimated the
total demand for refined sugar plus HFCS (US sweetener demand) as function of a
composite index of their prices. Secondly, the prices of refined sugar and HFCS in turn
were modeled as functions of the US raw sugar prices, which according to Marks is
targeted by the policy makers. In addition, he estimated equations for aggregate sugar
demand and supply for the rest of the world (ROW). From these equations he was able to

45
infer a price elasticity of ROW export supply. The price elasticity is important because it
gives an indication of the extent to which changes in the US imports would affect the
world market price.
Marks (1993) investigation reveals that in all years considered, the unilateral
liberalization of the US program would result in the world market price increasing, by an
amount which depends on the price differential between the US and world market price.
For example, the US programs depressive effect on the world market price in the year
1984/85, when the price differential was high, was approximately 16% whereas in the
year 1988/89, when world market price was relatively high, it was estimated at less than
9%. Among his other conclusions was the suggestion that the US sugar program
benefited the rest of the world in the aggregate over the period by an average of $0.331
billion. The program depressed the world raw sugar price, and cost foreign sugar
producers an annual average of $1,867 billion. However, these costs were outweighed by
$2,109 billion in gains to foreign consumers. The paradox he explained, was due to the
effects of the rent to foreign quota holders who profit from the price differential between
the US and the rest of the world. In his analysis of a marginal reform to the US sugar
program the result indicated that a 10% reduction in the real US loan rate would lead to
higher welfare in the aggregate in both the United States and the rest of the world.
Marks (1993) study although novel in some respects, suffered from two
weaknesses. First, the approach in estimating the US sweetener demand used the
historical annual market shares of refined sugar and HFCS as weights in computing the
composite sweetener price index. This method assumed a kind of fixed proportion good.
Second, in estimating the effects of future policy changes the erroneous assumption was

46
made that the technical substitution of HFCS for liquid sugar was complete and that the
existing market share would continue into the future.
Evaluation and Critique
A common feature of the studies reviewed here is that all utilized time series data
in their analyses to estimate parameters in supply and demand equations. With the
exception of the study by Lopez and Sepulvelda (1987), none considered the issue of
time-varying parameters. Secondly, although some attempt was made to take into
consideration the substitution effect between HFCS and sugar in the US sweeteners
market, none of the studies directly included HFCS in their trade analysis. Thirdly, all
studies inferred that if the US were to liberalize its sugar program unilaterally, it would
have a positive (upward) impact on the existing world market price. Fourthly, all the
trade-related studies were either spatial or nonspatial equilibrium models. According to
Thompson (1981) although spatial equilibrium models represent one of the most popular
approaches to agricultural trade modeling, especially for comparative statics analysis of
the effect of a change in policy, their adequacy for this purpose is questionable. He noted
that such models were not consistent with how many international agricultural markets
normally operate and hence do not do well in accounting for actual trade flows. Chief
among the reasons given for the real world inconsistency of the models was their faulty
assumption of a homogeneous traded product, ignoring the fact that many importers
differentiate among exporters on historic or political grounds. Consequently, changing
from one source to another is usually much more difficult than assumed by such models.
The second class of models used was the nonspatial equilibrium types in which
simulation rather than optimization procedures are used to give the net trade position of

47
each country. Such models are usually less efficient as a means of examining the effects
of changes in policy variables, such as tariffs and transport costs. The main advantage of
these models is that they are usually much easier to solve (Thompson, 1981). Both spatial
and non-spatial models are based on a competitive framework. However, as shown by
authors such as Krugman (1979), models with imperfect competition features might be
more realistic when modeling international trade. Although it is recognized by this author
that of necessity any model is an abstraction of the real world, an attempt will be made to
address a few of the shortcomings alluded to.
Time Series Analysis
Aoki (1990) states that the main aims of time series analysis are one or a
combination of the following: 1) to describe the data behavior succinctly; 2) to explain
the behavior of the time series in terms of exogenous variables; and 3) to forecast and in
some cases to control. Harvey (1987) makes a distinction between a pure time series
model and a structural time series model. The distinguishing feature of the two is that no
attempt is made in the former to formulate behavioral relationships among variables, in
contrast to the latter. Rather, the movements in the variable are 'explained' solely in terms
of its own past, or by its position in relation to time.
In utilizing time series data to carry out analysis several issues must be confronted
including a determination of whether the data are: 1) serially correlated; 2) stationary; 3)
possess a deterministic or stochastic trend; and 4) the stability of the parameters (time-
varying parameters). In cases where two or more time series are to be regressed on each
other there is the added concern of cointegration and guarding against spurious
regression. The following discussion provides brief insight into each of these issues.

48
Serial Correlation
Serial Correlation (autocorrelation) is the extent to which a time series variable
(usually the error term of the time series model), lagged one or more time periods, is
correlated with itself (Shim and Siegel, 1995). When such a situation exist it has
implication for the classical linear regression (OLS) model. The OLS estimator of the
parameters remains unbiased but no longer efficient and in fact is asymptotically
inefficient. Moreover, the standard formula used by the OLS to compute the estimate of
the variance of the errors is no longer unbiased and could either over- or underestimate
the true variance. The implication is that all tests of reliability, using the t- and F-
distribution, which rely on the estimated variance-covariance matrix of the estimated
parameters, become invalid.
Several tests such as the Durbin Watson (DW) statistic are available to detect the
presence of serial correlation and depending on the causes, various remedial actions can
be taken. In pure times series analysis, taking account of the pattern of serial correlation
usually allow for better forecast of future observations. Darnell (1994) notes that
detection of a pattern of first-order positive autocorrelation by means of the residuals can
result from several factors other than autocorrelated true errors (unobserved). These
factors include mispecification of functional and structural breaks in the relationship
(time-varying parameters). Consequently, he suggests that a respecification of the model
might be more appropriate and only when the evidence strongly suggest autocorrelated
true errors should the model be re-estimated by feasible generalized least squares or
maximum likelihood.

49
Stationary
Stationary is an important aspect of a time series. Formally, an infinite sequence
of random variables is said to be strictly stationary if the joint probability function of any
finite subset is identical to any other subset (Darnell, 1994). Less formal, when a series is
generated by stationary process it tends to fluctuate around a constant level and there is
no tendency for its spread to increase or decrease over time. In other words, the mean of
say Yt, its variance, and its covariance with other Y values, say Yt.k, are invariant to time
and the covariance between say Y, and Yt+S is a function of s only, for all t. Hence an
immediate implication of a stationary time series is that it has a time independent mean
and is homoscedastic. Kennedy (1998) points out that although many scientific time
series data are stationery, most economic time series data are nonstationary since they
tend to exhibit some kind of trend over time (i.e., the mean changes over time). One
reason for this (Aoki, 1990) is that the circumstances facing optimizing economic agents
change with time. Another is that in several cases the dynamic structures generating the
data are mostly nonlinear.
Nonstationarity has implications for the standard regression model in
econometrics since the model usually makes assumptions regarding the stationarity of the
error term as well as that of the variables in the regression. It has been shown that running
regression on nonstationary data can give rise to misleading (or spurious) values of R ,
Durbin Watson, and t-statistics, which may lead to an erroneous understanding of the
relationships which exist among the regression variables. For instance, it has long been
recognized (Yule, 1926) that regressing one nonstationary (trended) variable on another
could result in spurious regression.

50
Nonstationary time series have been subjected to one of two detrending
procedures in order to make them stationary before subsequent analysis. The two
procedures are: 1) regressing the time series as a simple linear (or higher order) function
of time and then using the residuals as the detrended series; and 2) an approach which
involves using first (or higher) order differencing of the time series. The first method
works only if the series say Yt has a deterministic trend and is generated according to the
following equation:
Y, = a +pt + e, (3.1)
where etis white noise. If however, the time series is generated by say a random walk or
random walk with a drift as for example:
Yt = Yt_i + P + v, (3.2)
where vt is white noise, the first approach will not work. More formally, equations (3.1)
and (3.2) are referred to as trend-stationary process (TSP) and difference stationary
process (DSP) with a trend, respectively.
The second approach has had some measured success even though as Kennedy
(1998) points out some analysts have adopted it without any justification based on
economic theory. Within this approach, a variable is said to be integrated of order d,
written as 1(d) if it has to be differenced d times to be made stationary. A stationary
variable is integrated of order zero. Economic variables are seldom integrated of order
greater than two and if non-stationary they are usually 1(1) (Kennedy, 1998). Equation
(3.2) above represents a series, which is integrated of order one. Taking Yt.i from both
sides of the equation gives equation (3.3):
AYt = P +vt
(3.3)

51
which is a stationary process. If P equals zero then the series is said to be integrated
(random walk) without a drift and if p 0 then it is said to be integrated of order one
with a drift.
Because of the serious econometric implications of modeling variables which are
nonstationary, as pointed out earlier, tests for stationarity and the degree of integration
have all but become routine to econometric analysis when working with time series data.
The most popular tests continue to be the unit root tests which are designed to test the
order of integration of the variable. Kennedy( 1998) points out that although a wide
variety of unit root tests are available, none of them is very powerful. The most popular
unit root tests are the Dickey Fuller or the Phillips-Perron in which the null hypothesis is
that the variable is non-stationary. But as Kwaitowski et al (1992) have shown
conflicting results could come about from the use of these tests and the one they proposed
with the null hypothesis being that the variable is stationary.
Cointegration and Error Correction Models
A major problem stemming from the use of variables made stationary by
differencing is that valuable information pertaining to the long-run equilibrium properties
of the data would be lost. The solution to this problem appears to be in a concept known
as cointegration. Two times series Y, and X, are said to be cointegrated if both series are
individually integrated of the same order say 1(1) and there exists a linear combination of
the two variables which is not integrated. In such circumstances it will be possible to
carry out a regression on levels of the variables and the results are meaningful (not

52
spurious) and there is no loss of any valuable long-term information, which would result
if their differences were used instead.
Moreover, in such situations, it is possible to tie in the short-term behavior of a
variable obtained from differencing, with a long-term equilibrium relationship. This is
done with a special formulation which involves a mix of both level and difference terms.
Such models are referred to as error correction models (ECMs). The ECMs can therefore
be viewed as comprising the short-run transitory effects and the long-run relationship and
describes how the long-run solution is achieved by a combination of negative feedback
and error correction (Darnell, 1994). In this regard, it is seen as a way to model the short-
run dynamics without losing sight of the longer-term steady state as may be posited by
economic theory. In other word, the ECM specification provides a means by which the
short-run observed behavior of variables can be associated with their long-run
equilibrium growth paths. Kennedy(1998) notes that this is one of the reasons why
economists have shown such interest in the concept of cointegration, since it provides a
formal framework for testing for and estimating long-run (equilibrium) relationships
among economic variables. He suggests that if a co integration relationship can be found,
advantage should be taken of the ECM framework.
It should be pointed out however, that the use of a single representation, implicitly
assumes that all the explanatory variables are exogenous. If this is not the case, then it is
suggested in the literature that one needs to use a vector auto regression (VAR) approach.
Furthermore, as long as the equation contains more than two variables, there is the
possibility that there could exist more than one cointegrating relationship. In such cases,
the traditional estimation procedures become inappropriate and again there is need to

53
switch to the VAR approach in which each variable is modeled in terms of lagged values
of all the other variables (Kennedy ,1998). Within such a general framework, testing can
be done to determine the number of cointegration relationships and the exogeneity of the
variables. The most common method is that of Johansen (1988). Manipulation of the
single equation VAR can produce a vector error correction model (VECM) in which the
vector terms are expressed as lagged differenced vector terms plus a lagged level term
which represents the error correction mechanism. As in the case of the ECM discussed
earlier, the intent is to exploit the steady state conditions resulting from cointegration
relationship and the short-term effects from the differenced variable in a single
framework.
A survey of the literature reveals, however, that there are unresolved issues with
the concept of co integration and the use of ECMs and VECMs. For one thing, although
ECMs and VECMs are used widely in the literature, several analysts (Harvey, 1997; Hall
et al, 1992; Aoki, 1990) have expressed serious reservations about the mixing of level
and difference variables in a given equation. Secondly, Monte Carlo studies have shown
that the estimates of the cointegration regression have considerable small sample bias in
spite of excellent large sample properties ( superconsistency). In this connection,
Cheung and Lai (1993) point to several finite-sample shortcomings in the Johansen
method, and have attempted to address some of these shortcomings by adjusting upwards
the critical values of the Johansen test. However, an implication of this is that the
likelihood of finding cointegration with finite sample correction is much more difficult.
Thirdly, in situations in which there are more than two variables in the regression
equation there is the likelihood that there could be more than one cointegration

54
relationship. In such situations, the standard OLS procedure does not produce estimates
that are consistent (Kennedy, 1998). Also, in these circumstances when there are more
than one cointegrating relationship there is usually a certain degree of ambiguity
surrounding the interpretation of the estimated cointegrating vectors (Johnston and
Dinardo, 1997; Kennedy, 1998). Kennedy (1998) further points out that the interpretation
of multiple cointegrating vectors can be frustrating. He refers to a practice among some
researchers, in dealing with this issue, of ignoring those cointegrating vectors that seems
not to make good economic sense as tantamount to imposing slightly false restrictions to
improve mean square error.
Fourthly, according to Johnston and Dinardo (1997) VARs have serious
limitations as a tool for the analysis of economic systems and are susceptible to the
problem of vanishing degrees of freedom, since the number of unknown coefficients can
rapidly approach the available sample size. An implication of this is that researchers
wishing to use this technique, but having limited data (observations), are forced to be
quite parsimonious in their model specification. However, this could be disastrous since
the failure to include an important variable could result in absence of any detection of
cointegrating relationship among the variables (Kennedy, 1998). On the other hand, as
pointed out by Johnston and Dinardo (1997), as more variables are added to the VARs,
problems arise in testing the number of cointegrating relationships. This comes about
because the test statistics are nonstandard distribution and require simulation. At present
the available tables can only account for eleven variables.

55
Time-Varying Parameters
Another potential problem with time series regression models is that the estimated
parameters may change. A review of the literature indicates that econometricians have
been aware of the problems of structural change since the late sixties as evidence by the
work on random coefficients (Hildreth and Houck, 1968; Langham and Mara, 1973).
However, it was not until the work of Rosenberg (1973) and Cooley and Prescott (1973)
that much more attention was focused on this particular phenomenon. Cooley and
Prescott (p. 463, 1973), in making a case for such analysis wrote:
In recent years economic theory has increasingly abandoned the relative security of
static equilibrium and perfect certainty...[as] it has become increasingly clear that to
assume behavioral and technological relationships are stable over time is in many cases
not only heroic, but completely untenable on the basis of economic theory.
Since then there has been a proliferation of articles in the literature dealing with these
issues (see for example survey by Nichols and Pagan, 1983). Ward and Myers (1979) for
example, used a distributed lagged advertising model with coefficients that had random
and systematic adjustments to demonstrate the dynamic effects of advertising on
consumer demand. Ward and Tilley (1980), using time-varying parameters with random
component, illustrated quite clearly the problems that can occur when parameters that
have changed are, in fact, ignored in the specification of the econometric model. Two
other interesting examples include the work of Engle and Watson (1987) in forecasting
electricity sales with time-varying parameters, and the work of Harvey et al (1986) on
stochastic trends in dynamic regression models, focusing on the employmentoutput
equation. In all such models one or more of the parameters is allowed to evolve over
time, and estimation techniques (ranging from simple models of varying parameters to

56
more complex ones involving state-space framework and Kalman filter (discussed
below)) are used for both the fixed operators and the varying parameters.
Time-varying parameters is regarded as a problem because standard regression
analysis of economic phenomenon assumes that the relationship being studied is stable
within the sample of estimation. For example, in the simple regression model of Y, = Xp
+ ut it is presumed that the p vector, which represents the effect of changes in the
explanatory variables upon the dependent variable, is constant. However, there are good
reasons why this might not be so and in fact should be allowed to vary. Ward and Myers
(1979) grouped the possible sources of parameter variation under three broad categories:
1) structural changes in the economic phenomenon being studied; 2)model
mispecification; and 3) aggregation. They noted that technology and institutional changes
always cause structural changes in economic phenomena. The Lucas (1976) critique of
economic analysis provides a good example. Lucas critique is to the effect that if the
policy regime is changed and economic agents takes account of the general policy
environment into their decision-making, then agents will adjust their behavior and the
coefficients that held in the previous regime will change to new values.
Sources of misspecification are varied but include omission of independent
variables, wrong functional forms and the use of proxy variables. For example, if the true
relation is:
Yt = ai + ct2Xt + 0C3X2t + ut (3.4)
and the analyst considers the linear relationship give by
Yt Pi + p2Xt + vt
(3.5)

57
then it is clear that the effect of B2 which is equal to (X2 + 2ct3Xt is not a constant. In the
case of aggregation as source of parameter variation, Ward and Myers (1979) cites as an
example the fact that over time, the relative importance of microeconomic agencies will
change and that such changes might not get reflected in the aggregate weight. In addition,
they noted that aggregated(macro) variables are discrete (Le., indexed as discrete time
points) whereas the underlying microvariables may be continuous, providing another
reason why parameter variation are likely to occur in macromodels.
In the literature, parameter variations are classified under the two broad headings
of stochastic and nonstochastic parameter variation. Each of these classifications is
further broken down into sub-categories. Hence, nonstochastic parameter variations,
which are caused by structural change in the economic phenomenon being studied, can be
divided into: 1) discrete variation or switching regression; and 2) systematic variation.
Likewise, stochastic models are divided into: 1) those which are stationary and are
referred to as random coefficient models; and 2) those which are nonstationary and are
termed sequential (Markovian) models13.
The State-Space Framework
The State-Space Model
Kennedy (1998) states that the state-space model can be regarded as a
generalization of the linear regression model and provides a unifying framework for all
dynamic linear models used in econometrics. A similar view is expressed by Harvey
13 For additional information 1 the various aspects of these types of models see Ward and Myers, 1979
and Nichols and Pagan, 1983.

58
(1993, 1997) that the state-space model is fundamental to dynamic modeling, and that
since most times series models can be put in this framework they provide a more robust
alternative to the modeling of time series analysis. The state-space framework is a
concept borrowed from engineering and is based on the original work of Kalman (1960)
and Kalman and Buey (1961). The concept centers around the possibility of tracking the
economy in a similar manner as engineers would track the state of a system, such as the
location of a satellite or a tanker, using noisy measurements. The Kalman filter
(discussed below) was used to derive optimal estimate of the state, given knowledge of
certain parameters. Engle and Watson (1987) note however, that there is a slight
difference between the approach used by engineers and that used by economists. In the
main, engineers usually have qualitative theories that describe the equation of motion of
physical systems and were primarily interested in the state of the system obtain from
noisy measurements. Economists on the other hand, were less fortunate in not having
such laws of motion of the economy at their disposal and consequently were much more
interested in discovering them from noisy data, rather than in merely estimating the state
of the economy. Consequently, economists had to overcome the initial problem of not
having the parameters of process in estimating the state. The focus for economists
therefore shifted slightly to finding a suitable methodology by which such parameters
could be estimated. The solution to the problem came in the mid-1970s when economists,
with the use of the Kalman filter were able to evaluate the likelihood function in
complex cases. Engle and Watson (1987) note that this development allowed the
parameters to be estimated using maximum likelihood methods as in standard
econometrics. It therefore opened up the way for the widespread use of the technique in

59
applied economics and the model became a natural generalization of latent variable
models to a full dynamic framework A brief taxonomy of empirical applications of such
modeling in economics can be found in Engel and Watson (1980).
To illustrate the state-space model, first consider the following standard single
equation regression:
m
y, = a++ Yj 8ix, (3-6)
i=0
where yt is a scalar observable data series, x,, is the single observable explanatory
variable with the structure of the lagged coefficients being 60...6m, t is the trend variable
with level and slope coefficients a and P, respectively, and et is the stochastic disturbance
term distributed such that st ~ NID(0,oe2). Harvey (1987) asserts that while equation
(3.6) is typical of many regression equations and that while many economic time series
show a trend, it is equally apparent that unless the time period is fairly short the trends
cannot be adequately captured by a deterministic (straight line) representation. He
observed that a great deal of applied economic work fall prey to the above error, since
they start off by detrending the data by regressing on time, thus making all that follows
invalid. He further points out that there is no reason in principle, why the parameters of
the explanatory variables of time series should not also be allowed to vary over time. He
suggests that a better approach would be to formulate the model in a state-space
framework. Such a representation could account for any unobservable components as
well as time-varying parameters.
Following along the lines suggested by Harvey (1987), reparametizing equation
(3.6) in terms of a simplified version of the state-space model, appropriate for time-
varying parameters yields the following:

60
m
y. =Mx +IXxt-¡+s
(3.7a)
i=0
M, = Mx-l+PxA + 7t
(3.7b)
A=A-,+£
(3.7c)
Sn =SA+vx
(3.7d)
where it is assumed that s, ~ NID(0,oE2), qt~ NID(0,Ot,2 ), £,~ NlD(0,c^2), and v,~
NID(0,ov2). The component pt is the trend, whereas p, is the slope of the trend and 6¡t are
the parameters of the explanatory variables, which are assumed to vary over time. The
introduction of the stochastic disturbance terms r|t and allows the level and slope of the
trend to change slowly over time. Likewise the disturbance term v,, allows 6it to be
generated by a random walk. More formally, the above state-space representation with
the simplifying assumption of no lags is:
Y, = (1 0 l)o, + e, (3.8)
where
1 1 0
Mx-i
7,
a, =
Px
=
0 1 0
PxA
+
kJ
0 0 1
k J
(3.9)
In terms of state-space description, such models usually consist of two parts: 1)
the measurement equation which describes how the data actually observed is generated
from the state variables (equation (3.7a) or equation (3.8)); and 2) the transition
(dynamic) equation which describes the evolution of a set of state variables (equations
(3.7b to 3.7d) or equation (3.9)). The vector at comprising the elements |it, pt, and 6, is
regarded as the state vector to be estimated and the measurements that are used to

61
estimate this state are the y,.. The statistical algorithm used to derive these state
parameters is the Kalman filter. Yt is the measurement equation. The parameters on the
right hand side of the transition equations (3.7a to 3.7c) or more precisely the system
matrix in equation 3.9, are described as the hyperparameters. They are regarded as being
fixed and known although they may change over the period in question. Engle and
Watson (1987) points out that on the basis of empirical experience and theoretical
arguments, in most cases these transition equations should to have unit root.
Concerning equations 3.7b to 3.7c, it should be noted that in the special case
where o^ o^2, and ov2 equal zero, the hyperparameters remain constant and equation
(3.7) collapses to the standard regression equation (3.6) with the deterministic trend and
fixed regression parameters. Harvey (1997) also showed that if oe2 = o^2 = ov2 = 0 and if
o^2 > 0, then the first diiference formulation is obtained as:
Ayt p + XS,Ax,_j + rj, (3.10)
Kalman Filter
The Kalman filter is simply a statistical algorithm, which allows certain
computation to be carried out for a model cast in the state-space form. Following the
approach of Harvey (1987) the typical model for which the Kalman filter is applied takes
the form of:
Y, = xA + s, (3.11)
5t = Gt5t-i + where as previously noted, equations (3.11) and (3.1 la) comprise the state-space system
in which the former is defined as the measurement equation and the latter as the dynamic

62
equation. Also, it is assumed that et ~ NID (0, oe2), and cot ~ NID (0, oj ). Although x,,
Gt, ,oe2 and oj may ultimately depend on a set of unknown parameters they are for the
purpose of the Kalman filter regarded as being fixed and known. Gt, oe and are
defined as the hyperparameters and are normally supplied by the researcher. In cases
where such information is not provided each defaults to the identity matrix. The
parameter of interest in the model is the state variable 5t, which evolves over time.
The Kalman filter uses a recursive estimation procedure to update the state as new
observations become available. In order to begin the procedure one must specify values
for ,oe2, Oa,2, xt, G, and the starting value for 5o. However, since such values are not
usually available the usual procedure is to construct starting values from the observations
themselves. Thus, in the case of a classical linear regression model with k explanatory
variables the starting values can be obtained by applying ordinary least squares (OLS) to
the first k observations with G = I and o(02 = 0.
The operation of the Kalman filter is easier understood when interpreted from a
Bayesian point of view. In this regard the goal of estimating 5t can be though of as being
carried out in two stages (Shen et al, 1999). In the first stage the 5t is estimated prior to
observing yt through equation 3.1 la. This can be represented as:
8t=GtSul (3.11b)
where, 8X is regarded as the best guess of the true value based on all information
available up to time t-1. In the second stage when the actual value of yt becomes
available, 8t is updated as follows:
t = Gtt., +A,e,
(3.11c)

63
where, Atet is considered to be the correction term. This term comprise two components.
The first component, the prediction error e,, is obtained as the difference between the
actual yt and the estimated or forecast value yt, which is based on the prior (un
updated) estimate of 5t i.e. e, = yt yt and yt = xt£, + s,. The second term, A,, can be
considered as the regression coefficient of the regression of 5t on et (Shen et al, 1999).
The entire process can be viewed as one in which the Kalman filter takes the
starting estimate of 5t and corrects it as new information becomes available, through a
one step at a time process, so that the estimator at any time t is a better predictor of the
actual observation. As mentioned earlier, from a practical standpoint, the starting values
are usually obtained from the computation of an OLS regression. Consequently, if the
equation contains k variables then k observations will be required to construct the starting
values. Thus, the first k estimates are usually disregarded. Having obtained these starting
values the procedure continues in a recursive manner until the final set of information at
time T becomes available, producing T- k one-step-ahead prediction errors.
At the stage when all information has been accounted for, a better estimator can
be obtained. The techniques for computing such an estimator are known as smoothing.
There are three basic smoothing algorithms; fixed point, fixed lag, and fixed interval
(Anderson and Moore, 1979). The one most commonly used is the fixed interval
smoother, which starts with the final Kalman filter and works backwards. It thus ensures
that an optimal estimator is available at all point in time based on all information up to
and including the final observation, YT. This information as pointed out by Harvey (1987)
can be quite valuable for examining the way in which a component such as trend or
elasticity evolved in the past. Finally, it should be noted that when the model is linear and

64
time invariant the Kalman filter will usually converge to a steady state, in that the
covariance matrix becomes time invariant and the estimates of the parameters are the
same as those obtained from fixed coefficient regression (Harvey, 1987).
Advantages of State-Space Approach and the Kalman Filter
Several authors (Harvey 1987,1997; Engle and Watson 1987; and Aoki 1990;
Hall et al, 1992) have alluded to the advantages of the state-space model over the
traditional parametrization of time series models such as ARMA, ARIMA, VARs and
VECMs. They noted that from a strictly theoretical point of view both types of models
are equivalent, because generic models in one representation can be transformed into the
generic ones in the other. However, when judged on other grounds such as numerical
stability, sensitivity with respect to small specification errors, statistical properties of
parameter estimators, or simply ease of dealing with nonstationary series, the state-space
models are superior. Moreover, they point out that in order to avail oneself of the
theoretical results and computational algorithms that have been developed over the years
in system literature, time series must be put in this newer state-space representation.
Hall et al (1992) in motivating the discussion for the use of the state-space
approach and the Kalman filter notes that although it is widely used in certain branches of
engineering and by applied statisticians, it is only just emerging as a possible useful tool
of the applied economists. He notes that the Kalman filter can be interpreted in terms of
agents forming expectations. He points out that while the paradigm for modeling
expectation is the rational expectation hypothesis (REH) where agents act as if they know
the true model of the economy up to a set of white noise errors, the consideration of the
information availability assumption has led critics to label it unrealistic. He notes that the

65
Kalman filter can be used to address the concerns of critics of the REH, such as Friedman
(1979), with regard to the information-available assumption. This he states can be done
by extending Friedmans framework in which he advocates that given the true model yt
xtp + ut (u, is white noise) agents may sequentially update their estimate of the fixed true
parameter vector p as more information on (yt, Xj) becomes available (e.g. time-varying
parameter models). Hall et al (1992) suggested that this could be done by: 1) assuming
that agents have some prior information about P (at time t = 0); and 2) by allowing p to
vary stochastically. Thus, agents are not assumed to know instantaneously the true model
but they do use the information optimally (efficiently), hence making the information
available assumption of the REH much more plausible. They further point out that in
certain models, the Kalman filter can be viewed as mimicking a learning process by
agents as in the case of adaptive expectation where the adjustment parameter is updated
each period, based on new information. This he notes formalizes Flemmings (1976) idea
of a change in gear when forming expectations. He makes the point that while the
adaptive expectation has been shown to be optimal (in the sense of producing unbiased
forecasts) only when the data generation process is of the form of an integrated moving
average, IMA (1,1), or ARIMA (1,1,1), the Kalman filter is optimal under more general
conditions, and in fact produces minimum mean square estimators (MMSE) under the
normality assumption. Harvey (1987,1997) has pointed out several other advantages of
the state-space framework and the Kalman filter. These are summarized below14.
14 The approach allows unobserved components to be incorporated in a model and the Kalman filter
provides the means of estimating them by way of a likelihood function. Estimates of the unobserved
components can then be obtained by smoothing. It therefore allows the entire model to be set up in terms of
components, which have direct interpretation.lt provides a framework in which time-varying parameters
can be estimated. For example, a stochastic trend can be estimated in which the level and slope are allowed
to evolve over time. Moreover, the framework is flexible enough to facilitate estimation of those

66
Relevance of the State-Space Model and Kalman Filter to the Current Study
The state-space framework was used in this study to estimate the derived demand
for HFCS. First, as pointed out elsewhere, the industry is relatively new which limits the
number of available observations (approximately 22). Also as discussed earlier, the VAR
based cointegration approach is biased with finite samples. In addition, one runs the risk
of encountering the vanishing degree of freedom problem referred to by Johnston and
Dinardo (1997). Second, there is strong evidence (Chapter 2 ) to suggest that the
parameters of the demand equation have been varying overtime For instance, the
growth rate of the demand for the product has varied considerably over the investigated
period. This suggests some kind of a stochastic trend rather than a deterministic trend.
Also, as pointed out elsewhere, the adoption of the product was gradual, implying that
parameters which do not vary over time, since the hyperparameters, which allow the parameter to vary, in
these cases would be set to zero, resulting in the estimation of fixed coefficients. These models in
forecasting put more weight on the most recent observations; the fasto- the level and slope change, the
more past observations are discounted.The state-space approach allows one to work in level data, rather
than differencing, even in cases where the data are nonstationary. Working in level form makes
interpretation of results much easier. He notes that traditional time series analysis, in situations where the
data is nonstationary, stresses the role of differences. Within the state-space framework it is not usually
necessary to difference in order to specify a suitable model. A similar view is expressed by Aoki (1990)
that while economic time series are usually nonstationary because circumstances facing optimizing
economic agents change with time and do not remain the same, the Kalman filter can deal more effectively
with nonstationary time series than can the traditional time series methods. Within the state-space
framework it is not necessary to test for unit root, as concerns over the degree of integration of a series are
not crucial. The state-space frame work is flexible enough that if the slope parameter is deterministic, the
hyperparameters which allow it to change over time will be estimated as zero or close to zero, hence little is
lost if indeed the series is not integrated The situations where VECMS can be usefully employed are quite
limited and one should have reservations about using them to provide a general vehicle for modeling
economic time series. These reservations stem from the fact that the VAR-based cointegration methods are
based on auto regressive models and like the unit root tests can have very poor statistical properties. In
addition, it is very difficult to fit autoregressive models to data with slowly changing seasonality or trend.
The strongest objection, centers not on the feet that when there are two or more cointegrating relationships
they can only be identified by drawing on economic knowledge, but rather that the VECM does not provide
a sensible vehicle for modeling the short-run...since it confounds long-run and short-run effects. As an
alternative to the VECMs Harvey proposes the use of multivariate structural time series model within the
state-space framework noting that such modeling can be formulated to incorporate long-run components of
cointegration directly by means of common trends. An added advantage of the state-space framework and
the Kalman filter is that the researcher need not understand the workings of the filter to apply the technique
and as such is free to concentrate on the selection of a suitable model and its interpretation.

67
time was needed for firms to learn about the product, determine its compatibility with the
manufacturing process, to formulate new recipes as well as make adjustment to existing
ones and to assess new product attributes. Also, with regard to some of the other
parameters such as the own price and cross price elasticity, for similar reasons mentioned
above, there is a strong suspicion that these might have undergone changes over the time
period, in view of the considerable changes which have taken place in the development of
the product making it far more attractive to the end users and consumers.
Third, is the fact that the objective of the study is not so much concerned with
whether there exists a long term relationship among the chosen explanatory variables, as
it is to identify the dynamic path of adjustments which might have occurred in the
industry and to draw inferences. As a consequence, the issues of cointegration and unit
root, though important, are of lesser concern within the present context. As Engel and
Watson (1987) observed that in most of the studies where time-varying parameters have
been used the predominant consideration seems to be on the stability of the regression
equation, only in a few cases have the variations been interpreted as economically
important. The current study intends to add to those few studies.
Finally, the use of the state-space framework within this study is an attempt to
make use of some of the theoretical and practical advances that have been made in the
area of systems literature and to provide yet another case where the techniques have been
used in economics.
The Concept of a Derived Demand
The term derived demand is normally used to denote the demand for inputs that
are used to produce the final product. According to Shim and Siegel (1995) the term

68
refers to the demand for a factor of production or product that is derived from the demand
for other goods. Neoclassical theory suggests that a profit maximizing firm will employ a
combination of inputs up to the point at which the additional cost of employing one more
unit of that input is equal to the additional revenue generated from the presence of that
that input. Given such a framework, a firms input demand functions (derived demands)
can be obtained in one of several ways. First, by solving the firm's profit maximization
objective function with respect to the given input. Alternatively, a system of derived
demands can be obtained by application of duality theory. This theory suggests that given
either the firms cost function or its profit function and assuming that such functions
satisfy a set of regularity conditions, the system of derived demands can be obtained by
applying Shephards lemma and Hotellings lemma to the given cost and profit function,
respectively (McFadden, 1978). For example, given a simple firm which produces one
output using three inputsthe third of which is fixed in the short runthe profit value
(optimized) function can be represented as follows:
n* = 7t* ( P, v, w, Z) = Max [Pq vxi WX2 : f(xi, X2, Z) > q ]
where n represents the maximum profit that the firm may obtain by allowing a subset of
inputs and output to vary while another subset of input is held fixed; P is the price of the
output; v and w are the prices of inputs xj and X2, respectively; xi and X2 are the levels of
the variable inputs used in the production process, while Z is the fixed input; q is the
quantity of output and f represents the production function. Since the profit function as
specify above is the result of a maximization process, then by applying Hotellings
lemma to the function the derived demands as well as the output supply can be obtained
as follows:

69
7*'/v = -x](P,v,w,Z)<0 =>
derived demand for input one
Sk/p = v, w, z) > o =>
supply function
The main difference between the derived demand obtained by way of the profit function
and that obtained from the cost function is that in the former output is allowed to vary
whereas in the latter it is held constant.
Once the derived demand is obtained, the own price elasticity of demand for the
input as well as the cross price elasticity of demand for an input can be computed. In the
case of the former, the elasticity will always be negative once the input is not considered
to be inferior. This is because in production theory, unlike the consumption theory, the
substitution and output effects always work in unison. In the case of the cross price
elasticity the sign is ambiguous. The ambiguity is due largely to the fact that in this case
the substitution and the output effects may not always move in the same direction. The
magnitude of both the own-price and cross-price elasticities is less predictable and
depends on factors such as the elasticity of substitution, the importance of the input in
terms of its share of total cost of production of the final good, and the price elasticity of
demand for the good being produced. According to Nicholson (1995) in general, the
own- or cross-price elasticity of demand for any input will be greater (in absolute value):
1) the larger the elasticity of substitution of that input for other inputs; 2) the larger is the
share of total cost represented by expenditures on that input; and 3) the larger is the price
elasticity of demand for the good being produced.
Thus, it is possible that given the profit function of the industries utilizing HFCS
as an input, and the assumption that such industries are operating within a perfect
competitive framework with a profit maximization objective, the derived demand curve

70
for HFCS can be obtained. However, in light of the difficulties surrounding the
estimation of a reliable profit function, such as obtaining cost sensitive information and
quantities used, a more straight forward approach can be taken in which the derived
demand is estimated directly on the basis of the explanatory variables suggested by the
above theory. These explanatory variables include the price(s) of the final
commodity(ies) and the respective prices of the various inputs used in the production
process. This latter approach will be used in this current study.

CHAPTER4
CONCEPTUAL MODEL OF US SUGAR AND HFCS MARKETS
As the process of development takes place, sweeteners in general are increasingly
being consumed as raw materials and inputs of industrial production and to a lesser extent
by households as essential foodstuffs15. Thus, in developed countries and regions such as
those of the EU, US, Canada and Japan, between 60%-70% of the sweeteners consumed
are in such goods as cakes, soft drinks and ice creams. However, in the developing
countries the situation is the reverse where only about 30%- 40% of the consumption of
sweeteners is in the form of sugar containing products (Atlas, 1996). This has obvious
implications for the patterns of demand, the direction of technical progress and the
underlying scientific investigation. The purpose of this chapter therefore is two fold. First
we construct a conceptual model of the US HFCS and sugar markets focusing on the
derived demand nature of the demand for sweeteners. While the model is essentially
static, some attempts are made to examine the dynamics of the relationship between
HFCS and sugar. Second, the conceptual model is used to assess some of the likely trade
and welfare implications resulting from changes to the US sugar program and an
expanded HFCS industry within the framework of the proposed FTAA. In carrying out
this assessment two scenarios are examined. The first focuses on the likely impact
resulting from a partial liberalizing of the US domestic sugar program, while the second
considers the more unlikely situation of the complete abandonment of the program.
15 Industrial (manufacturing ) uses include those for food (bakeries and cereals, confectioneries, soft drinks
and canned and frozen dairy products) and non-food items such as ethanol.
71

72
The US Sugar and HFCS Subsectors
Figure 4.1, presents a flowchart of the US sugar and HFCS sweeteners
subsectors. As stated earlier, HFCS is available mainly in a liquid form and is an almost
perfect substitute for liquid sugar, hence it is used in its entirety in the industrial sector.
Fig. 4.1. Flowchart of the US FIFCS and Sugar Subsectors
As evident from the Figure, useful byproducts are produced in the production of both
sweeteners. However, the revenue generated from the sales of the byproducts as a
proportion of the cost of the main input sugarcane, sugarbeet and com, respectivelyare
more substantial in the case of the production of HFCS.

73
The US Sugar Subsector
Focusing first on sugar, it should be noted that several studies have attempted to
model the US sugar market (Haley 1998b; Atlas, 1996; Bemirschka, Koo and Lou, 1996;
Marks 1993; Schmitz and Christian, 1993 ; Barros, 1992; Sudaryanto, 1987; Leu et al,
1987; Zietz and Valdez, 1986; Gemmill, 1976). Sudaryanto pointed out that many of the
studies fail to distinguish between direct and indirect consumption. They omit variables
explaining the derived demand characteristics of the consumption pattern. This study
recognizes the point made by Sudaryanto and others (Haley, 1998b; Barros, 1992) on the
need to disaggregate the market for sugars into derived and direct demand components.
However, a slightly different approach was taken in separating the market demand
components.
For convenience, a distinction is made between crystallized sugar (CS) and liquid
16
sugar (LS) In this study, crystallized (tabletop) sugar include sugars used both by
households and those segments of the industrial sugar users, in which it is currently
technically impossible to substitute sugar with liquid HFCS. The importance of this
distinction will become evident later in the analysis. However, for the time being there is
some evidence suggesting that at this juncture, it is in respect of the demand for liquid
sugar that a structural shift in demand has taken place. Only the liquid HFCS competes
with sugar. As pointed out earlier, liquid sugar and liquid HFCS are regarded as almost
perfect substitutes (Marks, 1993; Barros, 1992). Further supporting evidence for this
16 Refined sugar is available in three main product categories: white granulated sugar, liquid sugar and
specialty sugars. Granulated sugar is the most common form of sugar used in the households. Liquid sugar
is a mixture of water and sugar that is preferred by some manufacturers and specialty sugars include icing
sugar, brown or yellow sugars. Chemically, there is no difference between liquid sugar and crystallized
(granulated and specialty).

74
assertion exists when consideration is given to the nature of the penetration of HFCS in
the sugar consuming industries. For example, Haley (1998b) points out that in the
beverage industry, where a relatively large proportion of liquid sugar was used, HFCS
has almost totally replaced the higher-priced sugar. Indeed, all major sweet drink
manufacturers in the US use FIFCS as their main caloric sweetener.
Heuristically, one can therefore envisage the US sugar market as exhibiting two
distinct demand configurations in the early stages of the HFCS developmentone
inelastic and the other being relatively elastic. The inelastic curve represents the demand
for crystallized sugar (CS) as shown in Figure 4.2a. The curve is assumed to be inelastic
due to the absence of any close substitute. It should be noted however, that the exclusion
of the non-caloric sweeteners from this model implicitly assumes no formidable
Fig. 4.2. Hypothetical US demand for crystallized sugar (CS) and liquid sugar (LS),
respectively

75
challenge by that industry to this segment of the market. The second demand curve
represents the demand for liquid sugar (LS). As argued earlier, it is liquid sugar for which
HFCS is almost a perfect substitute (Figure 4.2b). The presence of this substitute (HFCS)
causes the demand curve for LS to become relatively elastic (rotating to the right as
shown in Fig. 4.2b) reflecting the fact that manufacturers are much more apt than
households to switch to a lower cost substitute because of the need to remain
. 17
competitive .
This characterization of the demand for sugar although novel, is nonetheless
anchored by evidence from the literature. For instance, Sudaryanto (1987) in partitioning
the US sweeteners market into a direct and derived demand for sugar found that the
elasticity of the derived demand was far greater than that for direct consumption. His
explanation was that since the consumption of final uses (direct) had approached
saturation, the responsiveness to a change in the price of sugar would be minimal
(inelastic). On the other hand, the demand for sugar in the food and beverage industry
was more elastic since many more substitution possibilities were available. The
Sudaryanto explanation of the differences in sugar demand configuration is somewhat
similar to that advanced by Thomas (1985). Specifically, Thomas noted that the US
market favored the development of the HFCS industry, since there existed high price
elasticity of demand for sugar among industrial sweetener users. These users, he
suggests, are far more ready than households to substitute one sweetener for another,
17 Economic theory holds that the price elasticity of demand for any good depends on many economic,
social, and psychological factors. Included among the economic factors are the availability of close
substitutes and necessity of a good. In general, necessities tend to have inelastic demand whereas goods
with close substitutes tend to have a more elastic demand (Mankiw, 1997)

76
given favorable price changes, and the technical characteristics of the individual
sweeteners.
Horizontally summing the two demand curves, CS and LS, in Figures 4.2a and
4.2b, gives us Figure 4.3, which represents the total market demand for refined sugar in
the US. Of significance in our model is the kinked nature of the resulting demand curve
ake. Point k on the curve, which represents the kink, can be interpreted as follows: At
prices above pK there is no demand for liquid sugari.e. it is completely substituted for
by HFCS. Below the kink however, there is both a demand for crystallized sugar (CS)
and liquid Sugar (LS), the latter in competition with HFCS.
Price
Fig. 4.3. Hypothetical US aggregate demand for sugar
In Figure 4.4, the US domestic sugar supply curve (Ss) is introduced and the
established US domestic support (loan) price, PT, set at approximately US 23 cents per
pound refined sugar. As drawn, the Figure indicates a situation in which there still

77
remains a market demand for liquid sugar. Thus, Xi and X3 indicate the quantities of
sugar produced and consumed, respectively. As such X3 Xi represents the US tariff-rate
quota (TRQ), which cannot fall below 1.14 MMT. X2 represents the quantity of CS
demanded and X3- X2 is the quantity of LS demanded. It should be noted that the
quantity of LS demanded does not represent the total demand for liquid sweetener that
would include the amount being satisfied by HFCS. Finally, Pw gives an indication of the
Fig. 4.4. Hypothetical US supply and demand for sugar
world market price in relation to the support price. Thus, the US sugar market might be
represented by the following equations:
Demand for CS:
Dcs = f(Ps,PF,Z)
(4.1)
Demand for LS:
Dls = f(Ps,PH(.),Qn(.),PF,Z)
(4.2)

78
Demand for Sugar:
Ds = Dcs + Dls
(4.3)
Supply of Sugar:
Ss = Sc{)+Sb(.)
(4.4)
Import Demand:
Id = Ds Ss > 1.256tonnes
(4.5)
Equilibrium Condition: Ds = Ss + Id
where:
(4.6)
Ps
domestic price of sugar in the US
Pf
index of sugar and HFCS containing products
Ph
price of HFCS
Qh
quantity of HFCS supplied
Sc & Sb
supply of cane and beet sugar, respectively
Z
a vector of other factors
The US HFCS Subsector
Turning attention to the HFCS subsector, the demand curve for HFCS is
conceptualized, as shown in Figure 4.5. Such a demand configuration is based on the
following grounds: 1) the observation that although HFCS is almost a perfect substitute
for liquid sugar (LS) it has always been sold at a discount to LS on a sweetness
equivalency basis (Fig. 2.3); and 2) the dual nature for the commodity, as discussed in
chapter one, which suggests that HFCS can only be substituted for sugar within a certain
range of the sugar demand function. It should be pointed out that above the controlled
price for sugar, Pt, the particular shape assumed by the demand curve is very ambiguous.
This is due to the fact that the price of HFCS has never exceeded that of sugar or even
come close to it. If one is willing to assume the conclusion reached by Tanyeri-Abur

79
(1990) that if the price of HFCS should ever exceed that of sugar the industries would
revert to using sugar completely, then this would suggest that the demand curve kinks at
or just above the controlled price and become highly or even infinitely elastic. However,
if one were to accept the arguments put forward by Marks (1993, p. 80 ) that "a reversal
of the time path of the real US sugar prices would hardly likely bring about the same
substitution pattern in reverse" then one might conceive of the portion of the demand
curve above the controlled sugar price, tapers off as shown in Figure 4.5.
Price
Several factors could account for a shift in demand for HFCS. As discussed in Chapter 3
in the section on derived demand, the demand for an input depends considerably on the
demand for the final good, hence any change in the demand for the latter will cause the
derived demand for the input to also shift. Thus, a shift in the demand for HFCS could
come about owing to any or a combination of the following: 1) changes in the prices of

80
the other inputs used in producing the final good, especially the price of sugar; 2) an
increase in the number of products using HFCS; 3) seasonality ; 4) changes in the price of
the final commodity; and 5) an improvement in technology which improves the
substitutability of HFCS for sugar. Consequently, the derived demand for HFCS can be
represented by the following general specification:
Demand for HFCS: Dh = /{Ph, Ps, Pf, T, Z) (4.7)
where:
Dh
=
demand for HFCS
Ph
=
price of HFCS
Ps
=
domestic price of sugar in the US
Pf
=
price index of sugar and HFCS containing products
T
=
technology (trend variable)
Z
a vector of other factors
With respect to the supply, it is important to note that the consensus of industry
experts is that the HFCS industry operates in an oligopolistic framework (Polopolus and
Alvarez, 1991, Thomas, 1985). A consequence of this is that even if the exact profit
function of such an industry were available it would not be possible to obtain the input
supply function of the industry by way of the duality theory, since the underlying
assumption of the theory is the existence of a perfectly competitive framework.
Moreover, in light of the prevailing circumstances under which the FfFCS is produced,
notably a kind of captive market resulting from the restrictive policies of the US sugar
program and the implicit floor price of sugar, it is argued that there is no supply function
for the industry but rather a reaction function. This reaction function can be viewed in the
contextalthough not identicalof the Stackelberg model in which one producer of the

81
substitutes has market intelligence and knows the reaction function of his competitor but
the other is naive. Consequently, the producer with the market intelligence take
advantage of his competitor by supplying an amount at a price which will maximize his
profit at the expense of his competitor. Here, the HFCS producers could be regarded as
the sophisticated suppliers, while those supplying sugar are regarded as naive. Taking this
analogy a step further, as in the case of the Stackelberg model in which it is possible for
one competitor to force another out of the market by sufficiently differentiating the
product, in a similar manner such an opportunity exist for HFCS to force liquid sugar out
of the market.
More specifically, it is assumed that the profit maximizing objective fimction of
the HFCS suppliers can be represented as follows:
= (Ph-Ch)*Dh(.) (4.8)
where n represents the maximum industry profit; Ph is the price of HFCS; Ch is the cost
of producing HFCS and DH is the demand for HFCS as defined in equation 4.7.
Assuming other factors remain constant, the reaction fimction of the HFCS suppliers is
obtained by differentiating the profit function with respect to the price of HFCS, setting it
equal to zero and solving for the price of HFCS. The reaction function will therefore
reflect the best price to charge for HFCS given the price of sugar and the other exogenous
factors that affect profit of the HFCS industry.
Within the framework of the above conceptualization, it is assumed that: 1) the
demanders of HFCS are price takers; and 2) that the marginal cost of producing the
HFCS remains constant over the entire range of output. The first assumption seems
plausible since in comparison to the number of firms producing HFCS there are large

82
number of firms that use HFCS as input into their production process. With respect to
the second assumption, this again seems plausible given that, the industry is highly
capital intensive and that there currently exist a considerable amount of excess capacity
within the HFCS industry In addition, the amount of com used in the industry is less
than 10% of the overall US com production, and consequently a moderate increase in
demand for HFCS is not expected to have any major impact on the market price of com.
Finally, the industry is highly capital intensive and as such one might expect that the cost
of an additional unit energy will remain fairly constant. On this basis, and in the absence
of concrete cost of production data, the short-run marginal cost curve of the industry (the
sum of the marginal cost over the variable cost) is conceptualized as shown in Figure 4.6.
The curve is drawn showing a section which is nearly perfectly elastic (constant marginal
cost) over some quantity range and becoming vertical as output capacity is reached.
Price
Fig. 4.6. Shift in hypothetical HFCS short-run marginal cost curve
In Figure 4.6, SMChi represents the industry short-run marginal cost function in
period one and SMCh2 is the short-run marginal cost curve in period two. As drawn the
shift in the marginal cost curve represents an overall reduction in the cost of production.

83
The shut down price is represented by P. The main factors which appear to influence the
marginal cost of producing HFCS are: 1) the net price of com (price of com less the price
of the byproducts); 2) the price of enzymes used in the conversion process; 3) the cost of
energy and capital; 4) labor cost; and 5) improvement in technology.
The hypothetical US HFCS market, as depicted in Figure 4.7, is obtained by
combining Figures 4.5 and 4.6. This representation is consistent with the view of industry
experts that the producers operate within a basically oligopolistic market structure,
Price
Fig. 4.7. Hypothetical US HFCS short-run marginal cost and demand configuration
implying among other things that demand and supply of the product do not determine the
pricing of the commodity. Moreover, because of problem with storage, production
closely matches that of the quantity demanded at the established price (See Chapter 2). In
Figure 4.7, it is assumed that the general pricing policy followed is that discussed earlier,
whereby the producers of HFCS having determined the best price for the commodity,
based on their reaction function and knowledge of the sugar subsector, supply a quantity

84
of the product which closely matches that demanded. This pricing strategy implies,
among other things, that the quantity supplied and demanded does not occur at the
intersection of the industry short-run marginal cost curve and the demand curve, but
rather at the point where the demand curve intersect with the established price. The
significance of this as far as estimation is concerned is that price and quantity
observations for the HFCS industry traces out the demand for HFCS. To the extent that
the prices received by the producers of HFCS is above the average cost of production,
they will be receiving economic rent and are in a position to adjust price of HFCS within
a certain range, in order to stave off the challenge from sugar in order to maintain or
increase market shares.
Based on the above line of argument, it is therefore theoretically possible to
estimate the derived demand function for HFCS using shifters and the observed price and
quantity data. This is done in the next chapter.
With respect to the long-term prospect for the supply of HFCS, there is
consensus that the industry is expected to benefit much more in the future than the sugar
industry owing to the fact that there exist both internal as well as substantial external
economies of scale (Marks, 1993; Thomas, 1985). Economic theory holds that in cases
where there are internal economies, one can expect a reduction in the average cost of the
particular firm as it increases its output. It also suggests that in the case of external
economies of scale the average cost of the typical firm declines as the output of the
industry expands (Lindert and Pugel ,1996). These authors noted that external scale
economies can come about in a variety of ways, including: 1) specialized services for the
industry; 2) the spill over effects of new knowledge about the product and production

85
technology; and 3) specialized machinery. They further suggest that if scale economies
are substantial over a large range of output, then a likely result is that a few firms will
expand their operation in order to reap the scale economies, giving rise to oligopoly
market structure. As it is well known, firms in an oligopoly can influence prices, can earn
economic profit and can vary the level of competition. More importantly, the existence of
scale economies has been shown by authors such as Krugman (1979) to provide as solid a
basis for trade as is comparative advantage. This line of argument forms the basis of the
New International Trade Theories (NITTs).
Much of the above is true for the HFCS industry, which has benefited
substantially from path breaking gains in biotechnology and processes such as
fractionation technology. Indeed, there is the view that the HFCS industry represents the
successful marriage of two industries: the wet com processors and the enzymes industry.
In light of the above and the fact that the industry is still relatively young and continues
to benefit from technological improvements, the study assumed a long-run HFCS
industry average cost reflecting external economies as depicted in Figure 4.8 .
Noticeable is the shape of the cost curve (LRAC) implying possible gains from
increasing return to scale. The figure also shows two hypothetical demand curves Dus
and Dfta reflecting the possible increase in the demand for HFCS as a result of the
formation of a free trade area, emphasizing the possible gains to be made from the
lowering of the average cost. The assumption of the long run shape of the HFCS industry
cost function forms a crucial part of the subsequent analysis of the trade implications
within the framework of the proposed FTAA.

86
Price
Fig. 4.8. Hypothetical HFCS industry average costs, reflecting external economies
The Interrelationship between the US Sugar and HFCS Markets
As indicated earlier there is a unique relationship between the US sugar and
HFCS markets. Sugar can substitute in all cases for HFCS but HFCS can only replace
sugar in a selected number of usesalbeit in such cases it has proven to be quite an
effective substitute (Barros, 1992). The implication is that it is only one segment of the
demand curve for sugar that is affected by changes in the demand and supply conditions
within the HFCS subsector. As indicated earlier in relation to Figure 4.3, the kink in the
sugar demand curve is of importance since it indicates the price above which there would
be no demand for liquid sugar. Several factors affect the positioning of the kink. This
point is not stationary but shifts in relation to changes to these factors. The primary
factors determining the positioning of the kink are: 1) the availability of other substitutes;
2) the price of HFCS which itself is influenced by several factors including the price of

87
com, the prices of the by-products, the price of domestic sugar; and 3) the supply
behavior of HFCS. These three factors are discussed below in highlighting some of the
dynamics of the sweetener market.
First, assuming other factors remain constant, it can be expected that an increase
in the number of available substitutes and the closer the substitutes on a per unit
sweetness equivalency basis the lower will be the positioning of the kink. This follows on
the assumption that there does not exist any switching cost for the manufacturers (user of
sweeteners) in utilizing one input over another. This latter assumption however is not
entirely accurate since such costs do exist in one form or the other. However, the general
principle applies in cases where such costs are negligible compared with the anticipated
increase in profits resulting from the change.
Second, and of greater importance is the relative price movement. For instance a
decrease in the price of HFCS is expected to cause the kink in the sugar demand curve to
shift downward, from ketoke , implying a lower level trigger price that sugar would
have to be sold in order for it to become an effective HFCS substitute (Figures 4.9a and
4.9b). The converse situation would also hold, that is, an increase in the price of HFCS
relative to the price of sugar shifts the kink, ke, upwards, implying an increase in the
trigger price at which sugar becomes an effective substitute for HFCS. These movements
postulated are consistent with economic theory which holds that a reduction in the price
of one commodity that is a close substitute for another commodity will cause a downward
shift in the demand for the substitute commodity. Thus, the change in the relative price of
HFCS only caused that segment of the sugar demand, (below the point of the kink in
Figure 4.9a, which is ke) to shift, since it is this segment which represents the demand for

88
liquid sugar, the HFCS substitute. It should also be noted that, other factors remaining
constant, the movement of the segment of the curve ke occurs along the dotted portion of
the demand curve akb as shown in Figure 4.9a for the case of a shift from ake to ake .
The nature of this movement has important trade implications. For one thing, it implies
that lowering the US support price for sugar (Pt) would induce only a slight increase in
the quantity of crystallized sugar demanded if the final support price is still above the
kink. This postulated movement is contrary to popular view of a substantial increase in
sugar demand associated with sugar price reduction in the US (Atlas, 1996; Schmitz and
Christian, 1993; Sudaryanto, 1987). This is unlikely, unless the price of sugar is lowered
Price Price
Fig. 4.9. Hypothetical supply and demand for sugar, and hypothetical demand for HFCS,
respectively

89
substantially so that it falls well below the point of any kink in the demand curve. And, as
noted earlier, the location of the point of the kink is a function of the dynamics of the
rapid economic and technological changes occurring within the HFCS industry.
The third factor mentioned above which influences the positioning of the kink is
the supply behavior of HFCS. This factor is closely related to the previous two factors
discussed. Marks (1993) in alluding to the pattern of the substitution between HFCS and
sugar asserts that it is a function of both the demand for sweeteners and the supply
behavior of HFCS. Elaborating on the latter factor, he postulated that a reversal of the
time path of the real US sugar prices would hardly likely bring about the same
substitution pattern in reverse. His claim for the likely failure of the reverse pattern of
substitution to occur is based on the view that the HFCS industry is in place with
hundreds of millions of dollars of sunk costs to protect and the advantage of leaming-by-
doing and other scale economies (p. 80). The implication of his assertion is that due to
the advancement made in the technology of producing HFCS, the benefit of product
development and the need to protect investment, the shifts in the sugar demand could be
structural. Moreover, on the demand side there also exist switching costs associated with
changing from one form of sweetener to the other. For example, in switching from liquid
sugar to HFCS certain costs would have been incurred both in modifying recipes and in
putting in place certain physical infrastructure such as storage facilities. While these
factors do not guarantee complacency on the part of the HFCS supplier, they imply a
lowering of the kink and that the price of sugar would have to fall substantially to bring
about a reversal of the current trend of sweetener usage.

90
Related to the issue of the supply behavior of HFCS and the kink is the structure
of the HFCS industry. Specifically, the degree of market power which exists in the
industry can influence the positioning of the kink. In general, other factors remaining
constant, it can be assumed that the less the intensity of competition among the HFCS
producers (the higher the degree of market power among the producers) the greater will
be their influence on the positioning of the kink. This is evident since it implies a greater
probability that the producers of HFCS will be able to influence the price of HFCS
covertly, which in turn will influence the positioning of the kink.
Trade Implications of Changes in the US Sugar Program within an FTAA
Given the conceptual model developed for the US caloric sweeteners industry the
question now addressed is "What are the implications of changes in the US sugar
program within the context of an FTAA, for the rest of the sugar producing members, in
particular those of the CARICOM?" This question attains some primacy based on the
objectives of this study. In this section, an attempt to address this matter using the
conceptual model developed. The analysis starts with a brief review of the concept of a
free trade area (FTA) as background to the subsequent analysis.
Free Trade Area
In the literature, a free trade area in the strictest sense is defined as an association
of countries that have agreed to remove all trade barriers among themselves, but retain
their individual national barriers against trade with third countries. In such an area it
would still be necessary to have custom officials at the borders to ensure that goods from
third countries do not enter through lower barrier members and trade within the higher

91
barrier members. A Customs Union on the other hand, is where, in addition to the
removal of all barriers to trade among the members, there is the adoption of common set
of external barriers thus avoiding the need for custom inspection at the internal borders
(Lindert and Pugel, 1996). However, in practice it is not uncommon to find a hybrid of
the two integration forms such as is the case of NAFTA, where most but not all of the
barriers to trade are removed and there is an agreement on phasing and harmonization of
barriers to third countries. While details of the specific form that the proposed FTAA will
take are not yet available, one can nonetheless advance a number of scenarios and
examine the consequences for the sugar and HFCS industries.
In advancing these scenarios cognizance is take of the changes to the US sugar
program as a result of the FAIR 1996 Act. In particular, note is taken of the attempt by
the US authority to make the program much more market oriented by the introduction of
the recourse loan policy, which essentially weakens the domestic guaranteed price. At the
same time, the elimination of the no-net-cost provision of the program makes the drawing
of any definite conclusion with regard to future state of the program ambivalent. In this
regard consideration is given to the view of Schmitz and Polopolus (forthcoming) that the
changes to the program as reflective of the FAIR 1996 are minor. However, as the com
refiners seek to expand their markets overseas, in light of: 1) the importance of sugar to
the prospective members of the FTAA; 2) developments within the WTO/GATT
negotiations: 3) internal domestic pressure from other commodity groups for which the
changes under FAIR were more severe; and 4) the weakened (although still powerful) US
sugar lobby it is postulated that the current US sugar program will need to undergo
substantial changes within the FTAA framework. In this connection, the following

92
analysis which is predicated on the foregoing conceptual model of the US sweetener
market, considers two scenarios: 1) a partial liberalization; and 2) a complete
liberalization of the US sugar program.
Scenario I- Partial Liberalization of the US Sugar Program
Trade implications
In view of the sensitive nature of sugar and the still relatively strong sugar lobby
within the US, one possible scenario is a partial liberalization of the US, one possible
scenario sugar program. It is conceivable that within an FTAA negotiation framework the
US could offer initially, to reduce its current level of support to domestic sugar producers
with an agreement that other member countries of the FTAA open up their markets to
trade in HFCS For example, the US could propose lowering the domestic support price
to say 15 cents per pounda price most experts feel would be the true world free trade
price. The effect of this decision is reflected in Figure 4.10a in terms of our conceptual
model as a fall in the support price from Pti. Here, it is assumed that the initial situation
which existed in the US market was such that at the current support price there were no
demand for liquid sugar i.e. substitution had been completed. This is reflected by the fact
that the kink is placed below the support price Pti. With the lowering of the support price,
two cases are possible. The first case is where the new support price Pt2 is at or above the
kink in the US sugar demand curve (Dsi). For convenience it will be assumed initially
18 A similar agreement was negotiated within the framework of the NAFTA Agreement between the US
and Mexico whereby provisions were made for the phased opening up of the US sugar market to excess
supply from Mexico over a 15-year period, while agreeing that Mexicos import tariff on US exported

93
that the price ratio of HFCS to sugar remains constant. This assumption is made to
prevent the shifting of the kinked portion of the sugar demand curve as discussed earlier.
In this case it can be seen that such an outcome would not affect the US demand for
liquid sugar and hence the demand for HFCS. In other words, any changes in the US
sugar program which result in the new price still being at or above the original kink will
not affect the US demand for liquid sugar, as long as the price for HFCS remains below
the domestic price of sugar. This is so because the price decline, although it increases the
Price Price Price
Fig. 4.10a. US sugar market Fig. 4.10b. Trade diagram Fig. 4.10c. ROW
Fig. 4.10. Trade implications under partial liberalization of US sugar program
HFCS would be no more than 15% and would be reduced to zero over a 10-year period (American Sugar
Alliance, 1998).

94
quantity of sugar demanded, from X2 to X3 ? and the import quota, it causes only a
movement along the inelastic portion of the sugar demand curve. This implies only a
slight increase in the demand for crystallized sugar (CS). Depending on where the kink
exists in relation to Pti. such a move could represent an appreciable fall in the support
price.
The second case considered is where the new support price falls below the
original kink in Figure 4.10ai.e. a movement from Pti to Pj3. Here it would appear that
the new support price of sugar makes it competitive with HFCS and so there is an
effective demand for some liquid sugar. Other factors remaining constant, the
consumption of sugar (both LS and CS) appears to increase substantially from X2 to Xs.
This would certainly be the case, all other factors remaining constant. However, one
needs to consider the changes, which would have been occurring almost simultaneously
within the HFCS industry.
As the hypothetical model of the long-run situation in the HFCS industry shows
(Figure 4.11), as a consequence of the external economies of scale there are potential
gains to be made as a result of the increased demand resulting from the opening up of the
HFCS industry to international trade. Furthermore, to the extent that the industry retains
some of its market power in the expanded market, producers will be able to obtain
economic rent as shown by the hatched area in Figure 4.11. As such gains are realized, it
would result in a lowering of HFCS, which would result in lowering of the kinked portion
of the US sugar demand curvei.e. from Dsi to Ds2 in Figure 4.10a.(see earlier discussion
on page 91). To the extent that the new kink Ds2 is below PT3, (as shown) the situation is
very much the same to that examined in the earlier casei.e. the change in the US sugar

95
policy having no effect on the US demand for liquid sugar. Consumption of CS increased
marginally from X2 to X4. If instead the new price is below the kink in Ds2, then it could
imply a sharing of the market between HFCS and liquid sugar.
To complete the analysis we invoke the large country trade assumption and
construct the usual trade-diagram and rest of the world (ROW) exporters, Figures 4.10b
and 4.10c, respectively. Assuming that the increased demand for HFCS, resulting from
the opening of the industry to trade within the FTAA, works mainly through the demand
for sugar in the rest of the world (Figure 4.10c), this development can be modeled as a
combination of two effects. Firstly, the presence of a substitute for liquid sugar in the
market causes the demand for sugar in the ROW to become more elastic. This is modeled
as a slight rotation of the demand curve. Secondly, as the price of HFCS declines and
Price
Fig. 4.11. Hypothetical HFCS industry long-run demand and average cost
reflecting external economies non-US users make the switch, the demand for HFCS gains
at the expense of sugar. This results in a leftward shift in the rest of the world demand for
sugar, from RWDsi to RWDs2. In other words, the increased consumption of HFCS in the

96
rest of the world will be at the expense of sugar. This leftward shift in the ROW demand
curve is explained by assuming that as the income of the members of the FTAA
improves, as a result of integration, there is for example a decrease in home baking and
an increase in eating outside of the home. Since less food is prepared at home, and since
it is more in the away-ffom-home food that there is a demand for HFCS, an indirect
substitution of HFCS for sugar occurs. Assuming iurther that the ROW sugar supply
remains the samegiven the limited incentive in the US market resulting from the change
in policythe net effect would be a rightward shift of the excess sugar supply function in
Figure 4.10b, from ES, to ES2, implying a fall in the world market price for sugar, from
Pwl to Pw2. The implication is that the partial liberalizing of the US sugar program within
the context of an FTAA, could in fact depress the world market price of sugar. This
conclusion differs from previous studies of the likely effects of a partial or full
liberalization of the US sugar policy which suggested that such action would increase the
world sugar price (Atlas, 1996; Marks, 1993; Schmitz and Christian, 1993).
Welfare implications
The welfare implications of the effect of a change in the US sugar policy within
scenario I, can be deduced from Figures 4.10 and 4.11. Notably, the producers of sugar in
the US would lose, while the producers of HFCS would gain substantially. The producers
of sugar in the US would lose as a result of lost revenue or income owing to eut back in
output and a reduction in the domestic price of sugar. The producers of HFCS however,
would gain from the lower cost of production and the increased market access within the
framework of the FTAA. Furthermore, assuming that such industries manage to maintain

97
some of their market power, then their gains would be even larger than that associated
with the traditional producer surplus, since they would be able to obtain economic rent.
In the case of the sugar exporting countries in general, the producers would lose as the
cheaper HFCS displaces some of their domestic market and with little hope of increased
access to the US sugar market. Moreover, revenues from sales on the world market would
decline as a result of the fall in price.
Consumers on the other hand would in general benefit from the lower all round
prices. The consumers in the US would gain considerably as both the prices of HFCS and
sugar fall. Contrary to the expectation from conventional trade theory which implies that
the consumers in the exporting countries are usually made worse off as a result of the
opening of the country to trade, in this case the opposite occurs as the users of HFCS in
the US will benefit from the reduction in prices associated with external economies of
scale.
Within the context of the FTAA, the major losers would be the relatively high
cost sugar industries of the small CARICOM countries. Firstly, they would be losing
their US import quota rent. Secondly, due to the permanent displacement of a portion of
the sugar market, it is likely that there would be a much smaller US market for the sugar
produced by these countries due to the fact that the cost of production within these
countries exceeds the current US domestic support price of approximately 18 cents per
pound of raw sugar. For example, production cost in Jamaica is reported to be US 33
cents per lb. (Caribbean Update, 1998). Also of consequence, is the fact that three of the
world's most efficient sugar producers; Brazil, Colombia and Guatemala would be
included as members of the FTAA and the region as a whole is a net supplier of sugar.

98
Thus, these low cost sugar-producing countries could easily supply any additional import
demand that might become available. At present, Guatemala supplies sugarin addition
to its TRQ allocationat the world market price to the US. This sugar is refined and re
exported as sugar or in sugar productsan indication of the competitiveness of
Guatemalas sugar industry. Thirdly, as was mentioned in chapter one in the section
outlining the problem, there is the likelihood that the CARICOM countries in becoming
members of the FTAA may have to give up their lucrative preferential EU/ACP sugar
quota. This situation comes about in the wake of a possible violation of the terms and
19
conditions of their Lome Convention If indeed this were the situation, the consequence
for the CARICOM sugar industry would be catastrophic. Finally, the governments of the
countries would lose revenue that they would have obtained from tariff on sugar imports.
Scenario II- Complete Liberalization of the US Sugar Program
Trade implications
A second but more unlikely scenario to be considered is the complete
liberalization of the US sugar program. Figure 4.12 represents the usual large country
trade analysis and is similar to Figure 4.10. A major difference here however, is an
elimination of import restrictions in the US sugar market. While the exact outcome is
somewhat ambiguous, the analysis suggests that the world market price of sugar would
remain more or less at the current leveli.e. no substantial increase and a strong
19 Lome Convention is a contractual non-reciprocal trading arrangement between African, Caribbean and
Pacific (ACP) countries and The European Union (EU). Prices received by ACP members for export of
sugar to the EU are on average three times that of the world market price. However, Article 174 Sec. 2a of

99
possibility of a slight decrease. The actual analysis leading up to this result is as follows:
With the opening of the US sugar market under an FTAA the world market price of sugar
rises from the pre-liberalization price of Pwl to Pw2 ? the free trade market price (Fig.
4.12a). At this juncture, sugar now competes directly with HFCS with the potential to
regain some of the sweeteners market share. This is shown in Figure 4.12a as the
potential for the US consumption of sugar to increase from X2 to X3, which would imply
that both crystallized and liquid sugar are consumed. However, the increased demand for
HFCS in the non-US countries (ROW) and the realization of industry external economies
of scale cause the price of HFCS to fall (Fig. 4.11). This, as discussed earlier, results in a
downward shift in the US sugar demand curve from Dsi to Ds2, with the accompanying
shift in the excess sugar demand curve, EDi to ED2, in Figure 4.12b. At the same time,
the increase in demand for HFCS displaces some of the sugar demand in the non-US
countries (ROW) causing their demand function to slightly rotate counterclockwise while
shifting to the left, from RWDsi to RWDs2, as previously discussed.
The increased opportunity to export sugar to the US also causes some of the more
efficient sugar producers in the FTAA to increase their supply of sugar, resulting in the
ROW sugar supply curve shifting to the right, from RWSl to RWS2. The combination of
the leftward shift in the demand curve and the rightward shift in the supply curve in the
ROW result in a substantial rightward shift in the excess sugar supply curve, from ESi to
ES2 in Figure 4.12b. A new free trade price is established resulting from the intersection
of ED2 and ES2. As drawn in Figure 4.12b this new price Pw3 is the same as Pwi, the pre
liberalized price. While this might not be the exact outcome, the point to be made is that
the Agreement states that in order to be eligible for the benefits developing countries are obligated to
provide no less favorable treatment to the European Union than they provide to any developed country.

100
owing to the presence of HFCS in the market, and the fact there is a strong possibility
that it can compete with sugar at world market prices, there is no guarantee that the
elimination of the US sugar program will necessarily result in an appreciable rise in the
world market price of sugar The exact outcome will be determined, among other things,
by the extent to which the importing countries open up their market and the degree of the
penetration achieved and sustained by HFCS. Additionally, it will depend on the long-run
ROW export supply price elasticity. Marks (1993) noted that in general the lower the
ratio of net exports to production, the higher should be the elasticity of export supply.
Price Price Price
Fig. 4.12a. US Sugar Market Fig. 4.12b. Trade Diagram Fig. 4.12c. ROW Sugar Market
Fig. 4.12. Trade implications under completely liberalized US sugar program

101
Using his estimate of 13.12 as a rough guide, the implication is that the opening up of the
US market should not seriously affect the world market price since a 100 % increase in
the quantity imported by the US would only result in an approximately 7.6 % increase in
the world market price, ceteris paribus.
Figure 4.12a also indicates a final outcome in which, it is possible that both liquid
sugar and HFCS are consumed with sugar regaining some of the market share that was
lost to HFCS. One can envisage a situation similar to that which exists in Canada, where
HFCS is still consumed in an environment where there is no special protection given to
HFCS and sugar prices reflect those of world market prices. Thomas (1985) points out
that in spite of such a situation, HFCS consumption captured approximately one-third of
the sweeteners market in Canada.
Welfare implications
The welfare effects of scenario II are similar to those that occurred in the partial
liberalization case under scenario I, although the magnitude of the benefits and losses will
differ considerably. Noticeably, the sugar producers in the US are expected to experience
considerable loss, while the gains to the HFCS producers would be less than in the
previous case. This latter situation will result since HFCS can be expected to lose some of
the market share as it competes with liquid sugar. The situation with regards to producer
in the rest of the world, including the FTAA is not clear, but the indications are that they
would again suffer losses or at best be made no worse off than previously. There would
however, be a noticeable redistribution of the benefits and losses among the producers,
with the relatively high cost CARICOM sugar producers being unfavorably disposed.

102
The main beneficiaries would be the consumers in both the US and the rest of the world.
The global benefits could be expected to exceed the losses.

CHAPTER 5
EMPIRICAL MODEL OF THE US DEMAND FOR HFCS
The preceding chapter outlines the conceptual model of the US sugar and HFCS
subsectors. This chapter focuses on the estimation of the derived demand for HFCS. The
first part deals with the specification of the model and gives a description of the variables
and data used. In the latter sections the results of the demand estimation and the model
implications are discussed.
Empirical Model
In keeping with earlier discussion of a derived demand function (Chapter 3) it is
postulated that the demand for HFCS is a function of the price of HFCS, the prices of
substitutes, and the prices of other inputs used in the production of the final HFCS-
containing goods. The total amounts of HFCS-42 and HFCS-55 utilized (millions of short
tons dry weight equivalent) were used to represent the quantity of HFCS demanded. This
amount reflects the total US domestic disappearancetotal supply less exports and stock
of HFCS. The HFCS-42 price (cents per pound dry weigh) was used as a proxy for the
price of HFCS. Using this series was necessary in light of the incompleteness in HFCS-
55 price series. This should not pose a serious problem, since both syrup prices tend to be
highly correlated.
The US wholesale beet sugar price series (cents per pound) was chosen to
represent the price of substitutes. Wholesale refined sugar prices rather than the
wholesale raw sugar prices were used, since the former is more readily comparable to the
103

104
price of HFCS on a sugar equivalency basis. While other com sweeteners such as
dextrose and glucose can be substituted for HFCS, in general owing to their much higher
price per unit of sweeteners there is limited substitution amongst these products. For
reasons discussed earlier, the non caloric (artificial) sweeteners were excluded from the
analysis. Consequently, only the price of sugar was included in the equation to capture
the effects of substitutes.
With respect to the final HFCS-containing products, only the price of soft drinks
was used. Although there are several such final products spanning several industries, the
models were limited to only the price of soft drinks, since this product utilizes the bulk of
the HFCS produced. In addition, sufficient information was not available to compute a
weighted index across the other products. Given the high degree of pair-wise correlation
among these indices, inclusion of more than one of these indices in the equation could
give rise to multicollinearity problems. While the soft drink index is not completely
representative of the range of final products, it was used as a proxy for those
unrepresented final products.
The final variable included in the derived demand equation is the price of flour.
Flour was used since a fair amount of HFCS is utilized in the baking industry. The
inclusion of this index represents a proxy for the other inputs used in the production of
final HFCS-sugar containing products. Finally, all prices in the equation are expressed in
real terms using the GDP deflator. The final model is expressed with a double log
specification representing the derived demand equation:
LQh, = Po + P.LRPh, + p2LRPSt + p3LRPDt + p4LRPFt + e, (5.1)

105
where LQm is the log of the annual quantity of HFCS used in period t; LRPhi is the log of
the real price of HFCS; LRPst is the log of the real price of refined sugar; LRPd, is the log
of the real price of soft drinks; LRPFt is the log of the real price of flour; and et is the
error term.
Of major concern is the dynamic path of adjustments of the coefficients over time.
As discussed in Chapter 3, equation (5.1) can be expressed in the state-space framework,
which then allows for estimation, prediction and signal extraction to be carried out using
Kalman filters and a smoothing algorithm. In the state-space framework and allowing for
a stochastic trend and time-varying parameters equation (5.1) can be represented as
follows:
LQhi = Pt + PuLRPh, + P2tLRPst + P3tLRPot + P^LRPfi + et (5.2a)
Pt = Pt-i + Yt-i + fit (5.2b)
Yt = Yt-i + St (5.2c)
Pit = P>(t-i) + con for i = 1,2,3,4 (5.2d)
where, pt is the stochastic trend, and et, fit, St and co¡( are all disturbance terms assumed
to be distributed normally and independently of each other, with zero mean and constant
variance. Equation (5.2a) is the measurement equation and equations (5.2b 5.2d) are the
transition equations. In matrix notation the above set of equations could also be
represented as:
Y, = (1 0 Z,')a, + 8,
(5.3a)

106
"1 1 0 0 0 0'
Mt-I
7,
r.
010000
Ka
6
fin
001000
Pm)
a, =

Pm)
+
Pn
000100
*>2.
P*
000010
Pm)
3,
Pm.
000001
_Pm t-l).
_(M.
[A] [B] [C] [D]
(5.3b)
where Zt represents in this case a 4x1 vector of explanatory variables; the at is the state
vector to be estimated; jit denote the stochastic trend as defined in (5.2b) above, and p,s
are the time-varying coefficients included as part of the state vector; and Y corresponds
to the vector of LQt. Equations (5.3a & 5.3b) are the measurement and transition
equations, respectively. The covariance matrix of the disturbance in (5.3b) is oE2Q where
Q = diag {a], a], o]a ,cr22), d2A(0 } and a] = on2/oe2 for example, is the relative
variance (noise to signal ratio) with respect to which the likelihood function estimating
the state vector will be maximized (Harvey, 1987).
Econometric and Statistical Issues
To estimate equation (5.1) above, several econometrics and statistical issues were
encountered. First, while econometric theory usually assists us in identifying the
variables to be included in a particular equation, it rarely ever specifies the functional
form which relates the dependent and independent variables. In the absence of such
indication the researcher must adopt the appropriate functional form based on experience
and what is known of the particular phenomenon being modeled. Consequences of
choosing the wrong functional form are well documented. The usual approach is to use a

107
linear form or a suitable approximation of a linear form and then test such assumptions
using test procedures such as the Ramsey RESET or the Box-Cox. The latter approach
was used to test the linear specification versus the log-linear specification. The results of
the procedure showed that the likelihood function was maximized with the value of X =
.01. This indicated that the log-linear function was superior to the straight linear form and
provided the criterion for choosing this particular functional form to represent the derived
demand
A second issue relates to the exogenity of the independent variables. This issue
becomes important because it assist in determining the appropriate estimating technique
to be used and hence the validity of the results. For example, in applying OLS to estimate
a single equation, the assumption is made that all the independent variables are
exogenous to the system (Darnell, 1994). If such is not the case then it poses an
estimation problem since, in this case, the disturbance term in the equation would become
a determinant of both the price and quantity. In other words, a right-hand side variable is
correlated with the disturbance. In standard econometrics this can be handled in one of
several ways, including the use of instrumental variables, two stage least squares (2SLS)
or obtaining the reduced form of the price variable. Similarly to the case of the OLS
regression, one of the requirements for utilizing the state-space model with a single
equation is that the independent variables are exogenous to the system being estimated
(Harvey, 1987; Engle and Watson, 1987; Darnell, 1994).
Owing to concerns surrounding the price of HFCS, a test of exogeneity (Granger
causality test) was performed on the quantity of HFCS demanded and the price of HFCS.
The results confirm the suspicion that the HFCS price was not exogenous. The null

108
hypothesis that the quantity of HFCS demanded did not influence the price of HFCS
could not be rejected. This could be interpreted that although the users of HFCS are price
takers, there is a feedback mechanism, whereby the quantity demanded influences the
price set by the suppliers. Viewed in this way, it meant that even though the price of
HFCS is not operating within a perfectly competitive framework and does not adjust to
ensure that the market is cleared within each period, it is nonetheless responsive to the
level of excess demand, which is given by the difference between the actual quantity
demanded and the actual quantity supplied.
In order to address these issues and to facilitate use of Kalman filtering, it was
necessary to obtain a reduced form of the price variable. This meant specifying an
approximation of the way in which prices of HFCS were formulated in terms of a set of
exogenous variables that are consistent with our earlier notion of the HFCS suppliers'
response fimction(discussed in the previous chapter). This was done by first regressing
the price of HFCS on all of the exogenous variables in the system. In a sense the
approach adopted can be viewed as carrying out the first stage of a two stage least
squares (2SLS) in which the endogenous variable is regressed on all of the exogenous
variable in the system. The uncorrelated estimate of the price of HFCS could then be used
instead of the correlated price variable in the second stage. In view of the profit
maximization objective assumed for the supplier and the notion of a reaction function,
three additional explanatory variables were introduced in carrying out the regression.
This gave rise to the following price specification as our working hypothesis with regard
to price formation:
LRPhi = <|>o + 4>iLRPct + <|>2LRPe, + 3LRIt + (juLRPst +
5LRPDt + (jiLRPFt + vt
(5.4)

109
where, LRPHt is as defined earlier; LRPct is the log of the real price of com; LRPei is the
log of the real price of energy; LRIt is the log of the real long term interest rates; vt is the
usual error term; and the remaining variables are as defined in the demand equation (5.1).
The real price of com (LRPct) was included since it constitutes the main input
used in the production of HFCS. The price series for the yellow dent com was chosen
since this represents the variety most commonly used in wet-milling process.
Consideration was given to using the net cost of com starch, (i.e., the price of the starch
after the returns from the major byproducts have been netted out) rather than the gross
price of the com. However, econometrics experiments with both variables showed that
the price of com was the better predictor of the price of HFCS. The cost of energy
(LRPEt) was included because of the highly capital intensive nature of the industry.
Energy cost is the second highest variable cost used in the production of the product.
Finally, long-term real interest rate (LRIt) was included to reflect the cost of borrowed
capital. This was computed as the difference between the long term nominal interest rate
and the inflation rate. Equation (5.4) was estimated by OLS and the predicted value of the
dependent variable LRP1It* was obtained and used as a proxy for the original HFCS price
variable. The respecified derived demand equation which utilized the Kalman filter is:
LQm ~ Pt + PitLRPnt + p2tLRPst + P3tLRPsDt + P4tLRPFt + et (5.5)
A third issue that arose had to do with supplying the values for the
hyperparameters and the initial values of the state vector within the framework of the
state-space representation (See Chapter 3, section on Kalman filter). In situations in
which the initial values of the state vector are not provided by the researcher, the first k
(the number of elements in the state vector) observations are regressed using OLS and the

110
parameters obtained are used as the initial values. In the case of the hyperparameters the
identity matrix is usually assumed if the initial information is not provided. In the current
exercise the information for the initial transitional covariance matrix was supplied by
setting it equal to the covariance matrix obtained from estimating equation (5.5) by
means of OLS. Once the initial values have been supplied, the Kalman filter generates the
likelihood function which is evaluated in terms of the prediction errors. The function is
maximized numerically with respect to the relative variances of the state vector.
Subsequently, a smoothing algorithm is applied to obtain the optimal (minimum mean
square) estimates of the state vector. In this case the fixed-interval smoother was used.
Description of Data
Annual data covering the period 1977 to 1998 were used in the estimation of
equations (5.4) and (5.5). The choice of the period was limited by the fact that the HFCS
industry began operations in the early 1970s with commercial production of HFCS-55
commencing in 1977. Data on the prices of sugar, HFCS, com and com starch, and the
quantities of HFCS were obtained from various issues of the United States Department of
Agriculture (USDA), Economic Research Services (ERS), Commodity Economic
Division, Sugar and Sweeteners Situation and Outlook publications.
The consumer price index for carbonated drinks, the price index for flour, the
index for energy (electric power and natural gas utilities) and the GDP price deflator were
all obtained from the Bureau of Labor Statistics (BLS). Price index for flour and the
index for energy were taken from the BLS producer price index revision-current series.
As indicated earlier, the real long term interest rate was calculated as the
difference between the 30-year (Moodys) corporate bond and the rate of inflation

Ill
(obtained from the GDP index). Information on the interest rates was obtained from the
Economic Report of the President (1997).
Estimation Results
The remainder of this chapter discusses the empirical results achieved using the
two-step estimation process, with equations 5.4 and equation 5.5. First, for simplicity
equation 5.4 was estimated as a double log using OLS. However, the diagnostic results
showed that it had a goodness of fit value, as reflected by the adjusted R2, of 0.63 with
two of explanatory variables being significant at the 5% level. Moreover, the Durbin-
Watson value was relatively low (1.09) indicating a possibility of the existence of serial
correlation among the error terms. The equation was re-estimated, assuming the existence
of first-order serial correlation, and the maximum likelihood technique was employed.
The new equation resulted in an adjusted R2 of 0.79 and the Durbin-Watson value
increasing to 1.72. The prices of com, sugar and the final good were all shown to have
significant impacts on the pricing of HFCS in a manner consistent with a priori
expectation. The results of both equations are shown in Table 5.1 as equations (5.4a) and
(5.4b), respectively.
With the information obtained from equation (5.4b), in particular the estimates of
price of HFCS(LRPhi*), the Kalman filter was used to estimate equation (5.5). The values
of the most recent set of estimates of the parameters together with their p-values are
presented in Table 5.220. This Table shows that the trend (pt), price of HFCS and the
20 Since the HFCS industry is one of imperfect competition the observation at any point in time t lies on the
demand curve corresponding to the price at which the MC = MR. The use of time series data to estimate the
derived demand curve implicitly assumes that all such points lie on a given demand curve. However, in

112
Table 5.1. Estimation of Price Equation
Description
Equation 5.4a-OLS
Estimates
Equati
Lik
cm 5.4b~Maximum
ihood Estimates
Parameters
Variables
Coef.
Std. Error
p- value
Coef.
Std. Error
p- value
4>o
Intercept
0.7395
1.4951
0.628
0.2797
0.9267
0.763
.
Price of Com
(LRPq)
0.2873
0.4150
0.499
0.6347
0.3310
0.055
§2
Price of
Electricity(LRPEt)
-2.0969
0.9381
0.041
-1.0939
0.8579
0.202
3
Interest Rate
(LRI,)
-0.4680
0.5829
0.435
-0.2425
0.4541
0.593
Price of Sugar
(LRPs.)
0.5301
0.4065
0.212
0.5772
0.2522
0.022
5
Price of Soft
Drink (LRP[>)
-1.2976
1.0847
0.250
-1.1964
0.8114
0.140
6
Price of Flour
(LRPfO
6.1545
2.3772
0.021
3.5046
1.8791
0.062
Adjusted R2
0.62
0.79
Durbin-Watson Statistic
1.09
1.73
rho
0.70
prices of the soft drinks (final goods) were significant at the 10% level. The signs on the
trend and HFCS price variable were consistent with expectations. In the case of the other
three variables no a priori assigning of the signs was possible given the nature of the
derived demand curve and the fact that the output and substitution effects do not
able 5.2 Final Estimates Using Kalman Filter Over the Period 1977-98
Parameter
Variable
Coefficient
Standard Error
p-value
Absolute Beta
Coefficient
Mi
Intercept
7.97624
4.47635
0.075
Pi.
Price of HFCS
(LRPh,*)
-1.63964
0.88958
0.065
2.26979
Pa
Price of Sugar
(LRPSt)
-0.43357
1.30173
0.739
0.87828
Pa
Price of Flour
(LRPf.)
-0.98789
1.82528
0.588
2.80602
P4
Price of Soft
Drink (LRPd.)
5.34171
2.77963
0.055
23.10574
view of the dynamic nature of the HFCS industry this is highly unlikely. Consequently, caution should be
exercise with interpreting the final estimates in Table 5.2 as representative of the derived demand curve.

113
necessarily act in unison. Since the double log formulation was used, the values of the
coefficients also represent the elasticities. The estimate of the own-price elasticity of
HFCS was -1.64 implying that the demand for the commodity was relatively elastic. In
the case of sugar the cross price elasticity was inelastic but not significant at any
reasonable level of confidence. While the price of flour was non significant at the 10%
level the price of soft drink was significant at this level. The last column of Table 5.2
shows the absolute values of the Beta (standardized) coefficients. These values give an
indication of the relative importance of the independent variables in the model and allow
for direct comparison (Pindyck and Rubinfeld, 1991). Thus, with respect to the price of
HFCS the value implies that a one standard deviation change in the price of HFCS leads
to a 2.27 standard deviation change in the quantity of HFCS demanded. The values also
reveal that among the explanatory variables chosen the price of soft drink was the most
important in explaining the relative variation in the quantity of HFCS demanded.
Attention is now turn to an examination of the implications of these estimates on
the time path adjustments for the selected parameter of the trend, own-price elasticity of
HFCS, and the cross-price elasticity with respect to sugar.
Model Implications
The state-space and the Kalman filter allow for the modeling of time-varying
parameters and for stochastic movements in the underlying trend variable. Addressing an
observation made by Engle and Watson (1987) that time-varying parameters when used
in most studies are typically interpreted as tests for the stability equation, the emphasis
here is on the economic importance of such variations. As such the concern is not so
much with whether the estimated parameters are stable and represent long-term

114
relationship, but rather whether any important economic interpretation can be associated
with their movements through time and the implications of such movements. In this
regard, discussion is therefore focused on the trend, own price elasticity of HFCS and
cross-price elasticity of HFCS with respect to sugar. To assist with the interpretation the
respective values of the parameters are plotted against time, ignoring the first few
observations which were used to derive the starting values. Thus, Figures 5.1, 5.2 and 5.3
show time paths of adjustments for the stochastic trend (n,), own price elasticity (pi,), and
the cross-price elasticity (p2t) for the period 1982-1998, using the smoothed estimates
(Appendix). In the case of Figure 5.1 the graph is indexed on the coefficient for 1982.
Intercept Adjustment Over Time (u.)
The dynamic path of the underlying trend in the utilization of HFCS is shown in
Figure 5.1. The path is nearly linear in the earlier stages, followed by an exponential
growth from the late 1980s until about 1996 and then a slow down in the growth pattern
in subsequent years. The observed pattern is very much in keeping with what is known
about the industry and support the hypotheses advanced in Chapter 1. HFCS was
commercialized in the 1970s and despite being sold at a discount to sugar its adoption by
many of the users was gradual. The early slow rate of adoption was due in part to the
general skepticism which would have surrounded the use of a new product, as well as the
existence of switching costs --the cost to the users associated with developing new
recipes and modifying existing ones. A period was needed for users to learn more about
the product, to determine its compatibility with the manufacturing process, and to
formulate new products as well as to assess consumers response.

115
Fig. 5.1. Dynamic path of HFCS intercept over time, 1982-98
The exponential growth beginning toward the end of the eighties and continuing
to about 1996 is consistent with the fact that by the late 1980s all the major soft drink
producers had switched from using liquid sugar to HFCS. In addition, it also reflects
changes in productivity resulting from technical progress and increased capital stock
among the users of the product. It is well known that the producers of HFCS work in
concert with the users to modify the product so as to improve its acceptability. With the
general up turn in the state of the US economy beginning around 1992 there has been an
increase in the HFCS demanders (increased capital stock) resulting from an increase in
the demand for final HFCS-containing commodities. Figure 5.1 reveals however, that

116
since around 1996, there appears to have been a slow down and a leveling off in the
growth pattern of the demand for the product. This finding is important from two
perspectives. First, it corroborates the suggestion made by some industry experts that
within the framework of the current technology, the extent of the substitution between
HFCS and liquid sugar has been completed. Secondly, the finding has wider implications
with regard to the future of the HFCS industry and its expansion within a wider trading
environment occasioned by the proposed formation of the FTAA. It also support the
earlier conjecture with regard to possible rent seeking behavior on the part of the
suppliers within a much more liberalized trading environment. This latter issue is
discussed in greater detail in the next chapter.
Own-Price Elasticity (Bid
Figure 5.2 shows the time path adjustments for the own price elasticity of HFCS.
In particular, the evidence suggests that the elasticity of the demand has been becoming
less elastic at a decreasing rate. In other words, the demand for the product although still
elastic, has become relatively less elastic changing from -2.2 to -1.65. Again, this seems
plausible given the developments that have taken place on the supply side in
differentiating the product from liquid sugar and making it more user-friendly. Recalling
the sugar and HFCS competition within the framework of the Stackelberg model
(Chapter 4), the observed pattern would be consistent with the theory which suggests that
by differentiating a product to such an extent where it is perceived as completely different
from its competition, it is possible to drive the competitor out of the market. This has
been the case to some extent with the HFCS-liquid sugar competition. Judging from
statements made by some of the industry experts, HFCS is now considered to be a

Elasticities
117
superior input to sugar in some applications such as in the coloring of certain baked
products. Moreover, as pointed out by Marks (1993) many of the users, having been
convinced of the attributes of HFCS and its favorable long-term pricing
Years
Fig. 5.2. Dynamic path of adjustment in HFCS own-price elasticity, 1982-98
prospects relative to sugar, have made investments in the necessary machinery and
equipment for using and storing HFCS. Consequently, the responsiveness to a change in
the price of the product is expected to fall since the importance of the product has
increased and the degree of liquid-sugar as a substitute should have declined. Finally, the
decline in the elasticity is consistent with our earlier suggestion of a possible maturity
with regard to substitution possibilities. This is so since if one were to accept the

118
argument, then further reduction in the price of HFCS should have less and less impact
given the limitation of technology.
Notwithstanding the above, the fact that the demand for the product is still elastic
is consistent with the observed recent pricing of the product, whereby the price has been
steadily declining since 1995. If the result had indicated that the demand was inelastic,
then owing to the constancy of the elasticity associated with double log model, it would
imply that the suppliers would be better off increasing their price in order to maximize
their profit. However, with an elastic demand expenditures can be increased and market
shares retained by reducing the price of the product. This is what has been taking place in
the industry and could help to explain why the price of HFCS has been decreasing in
recent years.
Cross-Price Elasticity (B?,)
Table 5.2 reveals that the coefficient for the price of sugar was not significantly different
from zero. The estimated coefficient implies that currently the effect of a change in the
price of sugar is having no impact on the quantity of HFCS demanded. Although this
result might seem somewhat surprising, it is nonetheless plausible if one were to accept
the argument that the suppliers of HFCS have managed to differentiate the product vis--
vis liquid sugar. And, that the extent of the possible substitution between HFCS and
liquid sugar, within the confines of current technology, is complete. Given current
restrictions on the price of sugar, the advances in technology which have facilitated the
substantial reduction in the cost of production of HFCS, and the extent of the product
development, HFCS seems to be operating in its own market as a separate and distinct
product from liquid sugar. As a consequence, a change in the price of sugar is having

Elasticities
119
little or no impact on the quantity HFCS being used. It now seems plausible that a
considerable reduction in the price of sugar would be needed to impact on the quantity of
HFCS currently being demanded
Fig. 5.3. Dynamic path of adjustment of cross-price elasticity of HFCS with respect to
sugar, 1982-98
This latter view is also consistent with the dynamic path observed in relation to
the cross-price elasticity as shown in Figure 5.3. Figure 5.3 shows clearly that the
adjustments over time has been in a manner such that the impact that a change in the
price of sugar was having on the demand for HFCS has been declining. If so, the extent
of the substitution of HFCS for sugar has been largely completed and the use of sugar

120
within such industries can be view more as a complement (joint inputs)21 rather than
substitute commodity. The declining degree of substitutability is perfectly consistent with
a product that eventually established its role as a reliable and safe alternative to liquid
sugar.
21 Recall that for some industrial uses HFCS is inappropriate and sugar must be used. It is in this broad
sense that the two products might be viewed as complements in the production of a representative industrial
product, given current prices and level of technology.

CHAPTER 6
SUMMARY, POLICY RECOMMENDATIONS AND SUGGESTIONS FOR
FURTHER RESEARCH
Summary and Conclusions
This dissertation had two general objectives. The first objective was to
qualitatively assess the potential economic impact that likely changes to the US sugar
program, and an expanded High Fructose Com Syrup (HFCS) industry, might have on
the CARICOM sugar industry within the framework of the proposed Free Trade Area of
the Americas (FTAA). The second objective was to provide quantitative estimates for the
US HFCS demand configuration that can be used in subsequent modeling exercises.
In chapter 1 the problem to be addressed as well as the hypotheses and specific
objectives were elaborated. It was noted that the issues were complex and involved many
players and institutions operating within a changing global policy environment. Only a
subset of the problem was addressed in this study. The approach to the study was
outlined. It involved developing a conceptual model of the US sugar and HFCS markets
and assessing on the basis of this model the trade and welfare implications resulting from
likely changes to the US sugar program and an expanded HFCS industry within the
context of the proposed FTAA. Subsequently, the derived demand for HFCS in the US
was estimated and used to explain likely concerns of the CARICOM sugar producers.
The analysis began in Chapter 2 with the development of a detailed background
on some of the key players and institutions involved. The complex nature of the problem
121

122
on hand within a rapidly changing international agricultural policy arena was
highlighted. Initiatives such as the WTO and FTAA were explored within the context of
trade liberalization and an improvement of the global/regional benefits as well as their
implied costs for some sectors. The CARICOM sugar industry was identified as one
likely to suffer adverse consequences of the rapid changes occurring in the international
agricultural policy arena.
It was noted in chapter 2 that the proposed FTAA comprises a wide diversity of
countriessome less developed such as those in the Caribbean, others developing such as
Brazil and Chile and others which are developed such as Canada and the US. A likely
consequence of this is that while the overall benefits to the region will improve as a
result of integration, some countries will benefit more than others. Indeed, given the
features of some of the smaller CARICOM countries, they could be among the losersat
least in the short-run.
A detailed review was made of the trends in the world, and Western Hemisphere
sugar and HFCS prices, production, consumption and trade, and the importance of the
sugar industry to the CARICOM sugar producing countries discussed. In the 1980s, there
have been major structural changes in the world sugar market as a consequence of
HFCS. Developing countries now dominated the import trade and their lower income,
and higher price elasticities suggest fewer fluctuations in world market price (Hannah and
Spence, 1997). A further implication is that the current relatively low world market prices
of sugar can be expected to continue into the future. This change in the status of the
market is attributed largely to the developed countries implementing policies, which
fostered the growth of their domestic sweetener industry. A consequence has been the

123
growth of large HFCS industries in the US, Japan and the EU. The US is the largest sugar
importing country within the hemisphere and the world lowest-cost producer of
sweeteners because of its highly efficient HFCS industry and relative abundance of com.
Currently, the US produces almost 75% of the world supply of HFCS. Even so,
production capacities remain underutilized. Since 1980s there has been a gradual
penetration of HFCS into developing countries, a trend which is likely to continue
(Hannah and Spence, 1997). There exist a real possibility that the product could pose a
formidable challenge to higher cost sugar markets in the region.
The sugar industry of the CARICOM sugar producing countries is of great
importance to the subregion. It contributes to foreign exchange, employment and
domestic activities and plays a vital role in the development of the countries through the
maintenance and upgrading of rural infrastructure, health, recreational and education
facilities.
In Chapter 3, studies of the US sugar market that took into consideration HFCS in
their analysis were reviewed and important issues with respect to time series analysis
discussed. All of the studies used time series analysis, and there are always concerns
when time series data are used to estimate economic parameters of interest. The impact of
HFCS was considered indirectly through the substitution effect in previous studies and
none included HFCS directly into their trade analysis. In general, all studies concluded
that the world market price of sugar would increase if the US partially or completely
liberalized its sugar policy. In the final sections of the chapter a description of the state-
space methodology and the Kalman filter was presented This framework was used to
estimate the derived demand for HFCS. Advantages of this approach were discussed and

124
a justification given for use of the methodology in this study. This methodology: 1)
allows for modeling of time-varying parameters, if the situation so warrants and provides
estimates at any point within the sample permitting one to estimate the path parameters
have taken over time; 2) deals fairly easily with nonstationary time series ; and 3) allows
modeling to be done in level versus differenced form which facilitates interpretation. The
flexibility of the state-space model makes it appropriate for estimation in situations, such
as in the case of derived demand for HFCS, where there are indications that the industry
is still evolving. The chapter was concluded with a brief theoretical discussion of the
concept of a derived demand.
A major part of the dissertation was concerned with constructing the conceptual
model of the HFCS and sugar markets used in assessing the implications. This was done
in Chapter 4. For convenience, the US demand for sugar was separated into the demand
for crystallized sugar (CS) and liquid sugar (LS). The CS represents the demand for
tabletop sugar plus sugar used in the industrial sector in processes in which it was not
technically feasible to use HFCS. The LS represents the demand for sugar for which
HFCS provides an almost perfect substitute. This kind of classification results in a kinked
demand for sugar. At prices above the kink there is no effective demand for liquid sugar.
The model is essentially static and therefore only provides a first approximation of the
likely occurrence. Dynamics were inserted through a comparative static framework with
regard to the interrelationship between the sugar and HFCS price. In modeling the HFCS
market, the imperfectly competitive framework within which the industry operates was
recognized. Consequently, it was argued that a supply function for HFCS did not exist
and that the supply could be better explained in terms of a reaction function, somewhat

125
similar to the situation in the Stackelberg model in which one player was sophisticated
(HFCS producers) and the other was naive. The users of HFCS were assumed to be price
takers. Suppliers established the best-response price, with consideration to the situation in
the sugar market and their objectives of maximizing profit and/or maintaining market
shares. Industry experts believe the long-term prospects of the HFCS industry appear
more favorable than its sugar substitute. This belief played a central role in modeling the
long-run situation in the HFCS industry. This view stemmed from the fact that the
product is the outcome of biotechnology with prospects of further significant
improvements in technologies, compared with sugar that is essentially a chemical
product. A likely implication is that the cost of producing HFCS could decrease
considerably as external economies are realized, providing real justification on the part of
the com refiners to actively seek expansion of the market for HFCS products.
Theoretical Findings
The impact of a partial liberalization of the US sugar program and a complete
liberalization of the program within the framework of the FTAA were also considered in
Chapter 4. The conceptual trade model utilized the typical large country trade
assumptions and assumed that the effect of an expanded HFCS industry could be
captured through its impact on the excess sugar supply and demand curves. The results of
the first investigation revealed the importance of the kink in the sugar demand
configuration. If in a situation of partial liberalization of the US sugar program, the US
domestic support price were to be reduced to a level at which it was still above the kink
in the sugar demand then, there would be an increase in the US demand for sugar and
most likely an increase in the levels of imports. However, the increase in the demand for

126
sugar would be marginal, since it would not affect the demand for HFCSi.e., the
increase in the demand would only be in relation to crystallized sugar. If on the other
hand, the price were lowered to a level that initially fell below the kink, then this would
set off a chain of events. Given declining cost in the production of HFCS and the current
view that the HFCS producers earn economic rent, then to maintain market shares the
price of HFCS could be lowered to such an extent to cause the effective demand for
liquid sugar to fall below the kink. Any increase in demand for sugar in the US market
would again have to be accommodated under the demand for tabletop (CS) sugar.
Implications are that structural changes have occurred within the US market and sugar
demand in the future will be limited further.
When the trade considerations are imposed on the model, a somewhat startling
result is obtained. The model suggests that within the context of a partial liberalizing of
the US sugar program and the prospects of HFCS penetrating overseas markets, the
world market price of sugar could fall rather than increase. This is due mainly as a result
of a portion of the rest of the world (ROW) domestic sugar demand being displaced with
HFCS and surpluses are forced on the world market. This revelation represents a radical
departure from the general findings of most studies. The discrepancy is due to a failure
of these studies to directly take into consideration the prospect of HFCS entering into
trade in their analysis. For example, HFCS could easily displace a substantial quantity of
the sugar used in Mexico's soft drink industry, with consequences of an increase in the
world surplus of sugar.
The welfare analysis associated with the first scenario suggests that whereas
consumers in general and the industrial users of sweeteners, in particular would benefit a

127
great deal, the higher cost sugar producers would suffer losses. Sugar producers in the US
would lose as a result of the decline in the support price. However, the HFCS producers
would gain as a result of a combination of increased sales and further decline in
production costs. The CARICOM producers with their relatively high cost of production
would be negatively impacted in competition with some of the more efficient sugar
producers, such as Brazil and Guatemala, for a share of the US market. The losses they
would experience could come as a result of a: 1) loss of US import quota rent; 2)
possible loss of lucrative preferential EU/ACP sugar quotas; 3 ) loss of a portion of their
domestic market; and 4) loss in earnings if the anticipated fall in world market price
prevails.
The theoretical findings in relation to the second scenario, the full liberalization of
the US sugar program, proved inconclusive. They did suggest however, that the prospect
of the world sugar price increasing substantially as a result of this decision was highly
unlikely. At best one could expect a slight to moderate increase in the world price of
sugar. Again, these results differ from those of most of the sugar studies reviewed, which
indicated an appreciable rise in the world market price of sugar as result of such action.
This apparent discrepancy is due to the failure of such studies to take into consideration
the potential impact of trade in HFCS in a much freer environment. The theoretical model
also suggests that with the complete liberalization of the US sugar program, relatively
cheap imported sugar might be able to reclaim some of its lost market shares. Again, this
conclusion rests on speculation, since it is known that the HFCS product is being
differentiated to the extent that it is being regarded in some circles as a superior product
to sugar. If indeed this were the situation, it would imply that the price of sugar would

128
have to be at a considerable discount to HFCS to cause a reversal in the current sweetener
pattern in the US.
The welfare implications of the full liberalization scenario were somewhat similar
to that of the partial liberalization, with consumers in general benefiting and the relatively
high cost sugar producers being the losers. Producers of HFCS again are expected to
benefit although to a lesser extent than in the case in which there is only a partial
liberalization of the program. And, as in the previous scenario the relatively high cost
CARICOM sugar producers would be the major losers.
Empirical Findings
The estimation of the US derived demand for HFCS was the only aspect of the
conceptual model investigated empirically and was the subject of Chapter 5. The ongoing
product development that has occurred in the market for HFCS, suggests that the
coefficients of the derived demand should be estimated within a framework that allows
for the estimation of time-varying parameters. The particular framework chosen was the
state-space model and the Kalman filters which permits one to observe the dynamic path
of adjustments taken by the varying parameter. In applying the technique it was necessary
to utilize a two-step approach. The first step essentially involved placing the model in the
state-space form and ensuring that all the explanatory variables were exogenous. The
second step involved carrying out the actual estimation of the equation using the Kalman
filter. The double log form was chosen to represent the form of the derived demand
equation so the estimated parameters are the respective elasticities. In keeping with the
focus of the dissertation, attention was only given to the dynamic path of adjustments
with respect to the intercept, and the own- and cross-price elasticities.

129
The empirical results clearly showed that structural drift has been an important
component of the derived demand for HFCS over time. The intercept coefficient was
significant, and indicated an asymptotic growth pattern which supported the view of
some industry experts that the industry is now in a mature stage and has reached the limit
of its technological applicability to various uses as far as the US market is concerned. The
implications for the sugar interest and the CARICOM sugar producers suggest that
producers of the product, barring any significant break-through in the crystallized market,
now face a problem of actively seeking external markets for their product. This becomes
even more crucial with the additional gains to be had from the realization of external
I
economies within such an expanded market. All this could spark a iurther round of rent
seeking on the part of the com refiners as they actively seek to expand their market, with
fewer barriers to trade. It also implies a iurther weakening of the US sugar-corn coalition
lobby as the com processors appear no longer interested in the protection offered by the
sugar program at the expense of penetrating overseas markets.
For the sugar producers in the US, the findings represent somewhat of a comfort
to the extent that there might not be any further erosion of the market shares in the US
sweeteners within the confines of the current technology. In this regard, it allows one to
get an indication of the likely size of the sugar market in the US and by extension the size
of the sugar import market. However, the tabletop (crystallized sugar) market still
represents a substantial portion of the sweeteners market and could result in more
aggressive research on the part of the HFCS producers to perfect their effort to produce a
competitive crystallized HFCS product. With respect to the non-US sugar producers,
such as CARICOM, the main concern of the findings and its implications will be the

130
extent to which HFCS could penetrate the non-US market and displaces a portion of the
existing sugar trade levels.
As in the case of the intercept coefficient, the own-price elasticity was significant
and elastic. The dynamic path of adjustments showed that the absolute elasticity had been
decreasing over time, implying that the demand for the commodity has become relatively
less elastic. This finding is consistent with the views of some industry experts that in
light of the considerable amount of HFCS product development over the years, the
product is establishing its own market and is being regarded in certain uses as a superior
product to liquid sugar. An implication of this is that the shift away from liquid sugar
which has occurred in the sweetener market in some uses, might be structural and that a
simple reduction in the price of sugar might not cause a major reversal of the current
pattern of sweetener used in the US.
Unlike the other two coefficients considered, the cross-price elasticity of HFCS
with respect to sugar was non significant. This may suggest that at present the change in
the sugar price no longer has a significant impact on the demand for sugar. This however
does not necessarily imply that at an earlier time period the impact of the sugar variable
was not significant. Again this finding is consistent with the view that the HFCS is
establishing its own market. The dynamic path of adjustments of this parameter estimate
showed that the relative cross-price elasticity had decreased considerably over time. This
suggests that over time the impact of a one-percent change in the price of sugar is having
a smaller effect on the relative quantity of HFCS demanded.
The above findings make the US sweetener market and how it will react less
predictable with the liberalization of sugar and with the price of sugar falling to a level

131
that is within the vicinity of the price of HFCS. The empirical findings therefore lend
supports to some of the earlier findings of the conceptual model, with regard to
uncertainty surrounding the possible outcome in the event that the US completely
abolishes its sugar program.
Policy Implications and Recommendations
In general, the findings of this study confirm inter alia, our hypotheses (See
Chapter 1). In particular, they supported the hypothesis that unqualified membership of
CARICOM countries in the proposed FTAA, could in the short to medium term have
potentially negative economic impacts on the subregion's sugar industry. Also, that an
expanded US HFCS industry could pose a threat to the subregional sugar industry. It
should be noted however, that the study did not specifically consider the competitiveness
of the CARICOM sugar industry vis-a-vis other sugar industries in the region. This was
not undertaken since it is generally accepted that despite the best efforts to modernize the
CARICOM sugar industry it can never be as competitive as segments of the US sugar
industry or those in countries such as Columbia, Brazil and Guatemala. As pointed out by
Kennedy et al (1998), the limitation of land and other resources are severe constraints in
CARICOM countries. A study conducted by LMC and reported by Haley (1998a)
suggests the field cost of production in the US is comparable to that of some of the
CARICOM countries. However, the US enjoys in some areas as much as a 40%
advantage over non-US countries factory costs. There are however, significant areas in
which the CARICOM sugar industry can improve its productivity and efficiency
considerably (Caribbean Update, 1999, James, {personal interview, 1999)).

132
As pointed out in Chapter 2 the industry is of paramount importance to the
maintenance of social and economic stability. It provides valuable foreign exchange for
which the shadow price exceeds the official rate and the earnings used to purchase
valuable capital goods that the subregion must acquire from outside sources. The industry
has a history dating back to the seventeen-century when the subregion produced the bulk
of the world output (Hannah and Spence, 1997). A consequence of this tradition is that
the subregion has acquired knowledge and skill in the cultivation and processing of the
crop. Moreover, a substantial amount of investment has been amassed in the industry and
fixity of these assets make it difficult to shift resources without creating serious social
and economic dislocation (Haraksingh, 1997).
The importance of the sugar industry to the CARICOM countries and its
limitations in terms of competitiveness must therefore be taken into consideration in any
discussion about integrating the subregion into a wider hemispheric free trade area.
Negotiations cannot be based solely on overall efficiency and competitiveness. Rather,
such considerations will have to be based on objectives such as those considered in the
political economy literature. Such a body of literature (Helpman, 1995; Rausser et al,
1982) emphasizes distributional considerations as a means of explaining among other
things, various trade polices of countries. This body of literature for example, views trade
policy as a device for income transfers to preferred interest groups in a society or group
of societies. Although this theme and approach was not explored in the main body of the
dissertation, it is clear that this dissertation has a political economy aspect.
As pointed out in Chapter 1, a dilemma facing CARICOM is whether to integrate
with the Americas within the framework of the FTAA or with the EU. The dilemma

133
stems from the fact that the CARICOM sugar producing countries run the risk of losing
their lucrative sugar market in the EU should they proceed with the decision to become
part of the proposed FTAA. While it has been argued (Gonzales, 1995) that the benefits
of participating in the EU/ACP program have not brought about the expected level of
development to CARICOM and its sugar industry, the benefits realized are quite
substantial. Also, while it is true that the accelerated pace of globalization and changes to
the EU policy are reducing the magnitude of such benefits, their contribution to the social
and economic stability is still of consequence.
Moreover, such benefits when viewed in the context of the findings of this study
assume an even greater level of importance. As the findings suggest, the CARICOM
countries within the framework of an FTAA would be hard pressed to compete with the
more efficient sugar producers for the US market. Furthemore, such competition would
be carried out for a sugar market that has undergone considerable structural changes
resulting in a decrease in its effective size. The situation becomes even more uncertain
when cognizance is taken of the fact that in accordance with the provision of the NAFTA
agreement, the US and Mexicos sugar markets are to become fully integrated by the
year 2008. In addition, there is the potential threat from the possible penetration of HFCS
in non-US markets in the region that would make it unlikely for the CARICOM sugar
producers to successfully compete in those markets. For CARICOM, the situation could
worsen especially if the domestic CARICOM sugar market also shrinks as a result of
successful penetration by HFCS.
Finally, even if the CARICOM countries were to modernize their industry to the
extent possible and be guaranteed a portion of the regional market within the framework

134
of the FTAA, it is clear that the price that they would receive would be less than that
which they are likely to receive in the EU market. This is so because there are several low
cost sweetener (sugar and HFCS) producers in the region, whereas in the EU the cost of
beet production is relatively high.
Consequent on the above, the CARICOM sugar producing countries need to push
ahead with their current restructuring aimed at streamlining the industry to take
advantage of production economies where such possibilities exist. This can be achieved
to some extent by closing down unnecessary mills and cutting back production carried
out on land deemed marginal for sugar cane production. In an interview with Mr. James,
the chairman of the Sugar Association of the Caribbean (SAC), it was pointed out that
this was an area in which there could be improvement. In some countries there are too
many mills which are not being operated at full capacity.
Of greater importance, the Lome/FTAA trade dilemma must be resolved through
a tripartite discussion throughout the various stages of the new Lome and FTAA
negotiation. It is clear that for the CARICOM sugar industry to survive, the retention of
its EU/ACP sugar quota is essential. And, while undoubtedly the overall benefits to the
region resulting from the formation of an FTAA will be increased, it must be borne in
mind that the benefits and costs will be distributed disproportionately. Furthermore, the
FTAA represents a group of countries at various stages of the development continuum.
On this point Balassa (1969) states quite clearly that regional agreements which involve
countries with dissimilar development levels usually offer fewer benefits to some
members, since differences in productivity and efficiency are usually of major
importance. It is known that the benefits of such agreements are usually greater where the

135
trade flows are mutually complementary. While to some extent there is some
complementarity of trade flows within the proposed FTAA, the majority of the
prospective members are major producers and exporters of sweeteners. Any negotiated
agreement to which CARICOM is a part should therefore include special provisions for
the CARICOM sugar industry and accommodate the sugar arrangement embodied in the
EU/ACP sugar protocol.
The HFCS in the future, whether within an FTAA framework or not, is going to
pose a formidable challenge to sugar as the dynamics of the sweetener market continues
to unfold. Consequently, to the extent possible a third recommendation is that the
subregional sugar industry must pursue an active program aimed at diversifying its end
product as well as the uses of sugar cane.
Limitations of the Study and Suggestions for Further Research
The research undertaken centered on assessing the dynamics of the US sugar and
HFCS sugar markets and on drawing of implications for the CARICOM sugar producers
within the framework of the proposed FTAA. The following limitations of this study
suggest areas for future investigation. First, although some attempts were made to explore
the dynamics of the situation in the theoretical model and more so in the empirical
estimation of the demand curve, the setting was essentially static. As pointed out
elsewhere, static models can only provide useful summary statistics and indicate the
direction of the outcome. Thus, while being a useful first step in an attempt to understand
complex issues, fully dynamic models are required to capture the full effects and better
represent the reality of the assumed changes. A Computable General Equilibrium (CGE)
model which takes into consideration some of the issues raised in this study might offer a

136
worthwhile approach in the future as more data become available. Consideration might
also be given to developing a dynamic partial equilibrium net trade sugar and HFCS
simulation model for the FTAA countries and the rest of the world (ROW). This study
focused on only the sweetener sector. The inclusion of other sectors in an analysis could
serve to better improve the understanding of the issues confronted. Such an approach
requires additional data and time not available within the confines of this study.
Only one aspect of the theoretical model was investigated empirically. Although
this is in keeping with the specific objectives identified in chapter 1, a more complete
model which facilitates testing of some of the theoretical findings is needed. This study
did not estimate the kink in the sugar demand curve that played an important role in
modeling the impact of the likely changes of the US sugar program. This too needs to be
studied. The location of this kink is a function of the dynamics of the rapid economic and
technological changes occurring with the sweetener market and should be studied in a
dynamic framework. A game theoretical analysis of the behavior of the HFCS suppliers,
may provide a clearer understanding of how suppliers might respond to a freer trading
environment.
A third shortcoming had to do with the exclusion of artificial sweeteners
noncaloric sweeteners such as saccharin, aspartame, acesulfame-K. The rationale given
for their exclusion was that there is a certain amount of health related issues surrounding
the use of these products. While to some extent this is true, within recent years various
health authorities have cleared most of the products. Such sweeteners are increasingly
being used in the so-called diet foods and soft drinks and are usually 200 to 300 times
sweeter than sugar. While these sweeteners have not yet made significant inroads into the

137
traditional sweetener market (Hannah and Spence, 1997), it is the view of these experts
that as more research is undertaken with respect to these products, they will begin to
seriously encroach on sucrose and HFCS, rather than remain confined to the diet market.
At such time it will not be easy to exclude them from the analysis of the dynamics of the
sweetener market.
Finally, this analysis was severely hampered by the lack of data on the HFCS
industry from two perspectives. Firstly, the industry is relatively new being in existence
for less than three decades. A consequence is that there are only a limited number of
annual sample observations available that severely constrain the methodology one can
employ. Secondly, in view of the competitiveness of the sweetener market and the
oligopolistic nature of the structure of the HFCS industry, important sensitive cost of
production data and prices are proprietary and remain difficult if not impossible to access
for public research purposes.

YEAR
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
APPENDIX
STATE AND SMOOTEHED ESTIMATES
STATE ESTIMATES
INTERCEPT
HFCS
SUGAR
FLOUR
SOFT DRINKS
2.35060
-1.15039
0.54531
0.36625
4.37817
2.35060
-1.15039
0.54531
0.36625
4.37817
2.35060
-1.15039
0.54531
0.36625
4.37817
2.35060
-1.15039
0.54531
0.36625
4.37817
2.35060
-1.15039
0.54531
0.36625
4.37817
1.77204
-0.75027
0.31236
1.01283
3.21563
2.19728
-1.04408
0.47241
0.69941
3.84027
2.69823
-1.58493
0.87955
-0.54212
5.23126
2.93150
-1.76100
1.05752
-2.34914
6.68421
2.71913
-1.89913
1.26641
-2.91761
6.98483
2.71488
-1.90004
1.26872
-2.92442
6.98753
3.39063
-2.30996
1.45474
-2.84288
6.89725
3.54696
-3.14583
2.15802
-2.66301
6.80810
0.40229
-2.37257
2.45879
-4.31041
6.26375
0.43293
-2.36084
2.43532
-4.19048
6.18130
1.41812
-2.33124
2.12672
-4.10239
6.47785
1.94168
-1.76743
1.48854
-4.29419
6.43354
2.58921
-1.66569
1.21797
-4.34319
6.19456
4.72430
-1.40368
0.35185
-3.49925
5.94294
7.56107
-2.34393
0.32677
-1.66712
5.41541
7.50306
-1.93869
-0.01940
-1.48805
5.35799
7.97624
-1.63964
-0.43357
-0.98789
5.34171
138

YEAR
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
139
INTERCEPT
2.35060
2.35060
2.35060
2.35060
2.35060
4.46769
4.55585
4.75492
4.97237
5.18549
5.40252
5.58922
5.76038
5.94295
6.32377
6.69173
7.04554
7.38946
7.71357
7.93952
7.96047
7.97624
SMOOTHED ESTIMATES
HFCS
SUGAR
-1.15039
0.54531
-1.15039
0.54531
-1.15039
0.54531
-1.15039
0.54531
-1.15039
0.54531
-2.22674
1.10953
-2.14419
1.01180
-2.09098
0.90724
-2.04130
0.80051
-1.99177
0.69514
-1.94281
0.59014
-1.89038
0.49291
-1.83814
0.39991
-1.77051
0.28948
-1.74007
0.14978
-1.70745
0.01258
-1.67621
-0.11948
-1.66450
-0.23255
-1.66559
-0.32941
-1.68225
-0.38388
-1.66338
-0.40732
-1.63964
-0.43357
FLOUR
SOFT DRINKS
0.36625
4.37817
0.36625
4.37817
0.36625
4.37817
0.36625
4.37817
0.36625
4.37817
-2.91161
6.54083
-2.82031
6.28829
-2.67334
6.13299
-2.51831
5.99348
-2.36385
5.85107
-2.23221
5.74120
-2.15580
5.67193
-2.08330
5.60115
-2.02052
5.53505
-1.85404
5.50256
-1.70599
5.48071
-1.55618
5.45034
-1.38316
5.41328
-1.19581
5.37920
-1.03380
5.34817
-1.01011
5.34521
-0.98789
5.34171

LIST OF REFERENCES
American Sugar Alliance. Initiation of Section 302 Investigation on Mexican Practices
Affecting HFCS, (Office of the United States Trade Representative) June 19,
1998.
Anderson, B and J Moore Optimal Filtering. Englewood Cliffs, NJ: Prentice Hall, 1979
Aoki, M. State Space Modeling of Time Series. 2nd edn. New York: Springer-Ver lag
1990.
Atlas, M. The Impact of Trade Liberalization by Major Importing and Exporting
Countries on World Sugar Trade. Ph.D. Dissertation, University of Colorado,
1996.
Balassa, B. The Theory of Economic Integration. London: George Allen & Unwin Ltd,
1969
Barros, A. R. Sugar Prices and HFCS Consumption in the United States. Journal of
Agricultural Economics, 43 (1992):64 -73.
Bernal, R. The Integration of Small Economies in the Free Trade Area of the Americas.
Paper prepared under the aegis of the Center for Strategic and International
Studies (CSIS), Washington: Volume IX Study 1, February 1998.
Bemirschka, M., W. Koo, and J. Lou. World Sugar Policy Simulation Model: Description
and Computer Program Documentation. Fargo department of Agricultural
Economics, Agricultural Experiment Station, North Dakota State University,
1996.
Blanco, M. ACP-EU Trade: Into the New Millennium. The Courier. 166(November-
December 1997): 64-67.
Borrell, B. and R. Duncan. A Survey of World Policies, in The Economics and Politics
of World Sugar Policies S. V. Marks and K. E. Maskus eds., p. 15-48 Ann
Arbor: The University of Michigan Press, 1993.
Bryan, A. Coping with the New Dynamics in The Caribbean: New Dynamics in Trade
and Political Economy, A. T. Bryan, ed., North-South Center, 1995.
Buzzanell, P. U.S. Sugar Trade Relations with Latin America. Paper presented at the
Third World Sugar and Sweeteners Conference, Sao Paulo, Brazil, October 1998.
140

141
Caribbean Update. Kal Wagenheim: Maplewood. Vol 14 #11 Dec. 1998.
Caribbean Update. Kal Wagenheim: Maplewood. Vol 14 #11 July 1999.
Carman, F. "A Trend Projection of High Fructose Com Syrup Substitution for Sugar."
American Journal of Agricultural Economics, 64(1982):625-633.
Cheung, Y. and K. Lai. Finite-Sample Sizes of Johansens Likelihood Ratio Tests for
Cointegration. Oxford Bulletin of Econometrics and Statistics, 55(1993):313-28.
Congress of US, The 1997 Economic Report of the President: Economic and Budget
Outlook: Hearing Before the Joint Economic Committee, Congress of the US,
One Hundred Fifth Congress, First Session. Washington DC: US GPO, 1997
Contrearas, A. Agricultural Issues Relating to the 1999 World Trade Organization
Multilateral Trade Negotiation. House Agriculture Committee Field Hearing,
January 22, 1998.
Cooley, T. and E. Prescott "An Adaptive Regression Model," International Economic
Review, 14(1973):364-71.
Darnell, A. A Dictionary of Econometrics. Aldershot: Edward Elgar, 1994
Engle, R. F. and M. Watson A Time Domain Approach to Dynamic Factor Analysis
and Mimic Models, Les Cahiers du Seminaire dEconometric, 22,(1980): 109-22.
Engle, R. F. and M. Watson The Kalman Filter: Applications to Forecasting and Rational
Expectations Model In T.F. Bewley (ed,)Advances in Econometrics: Fifth
World Congress. Cambridge: Cambridge University Press, 1(1987): 245-84.
Flemming, J. Inflation, Oxford: Oxford University Press, 1976
Friedman, M. Optimal Expectations and the Extreme Information Assumptions
of Rational Expectations Macro Models, Journal of Monetary Economics, 5
(1979): 23-41.
Gemmill, G. The World Sugar Economy : An Econometric Analysis of Production and
Policies. Michigan State University. Agricultural Economics Report No. 313,
1976.
Gonzales, A. Europe and the Caribbean: Toward a Post-Lome Strategy," in The
Caribbean: New Dynamics in Trade and Political Economy, A. T. Bryan, ed.,
North-South Center, 1995.
Group of Experts, Overcoming Obstacles and Maximizing Opportunities: Smaller
Economies and Western Hemispheric Integration, http://ncfap.org/ftaa-
wtol.htm. 1998.

142
Haley, S. US and World Sugar and HFCS Production Costs. Sugar and Sweetener
Situation and Outlook Report, USD A, Economic Research Service, SSS 223
May, 1998a.
Haley, S. Modeling the US Sweetener Sector: An Application to the Analysis of Policy
Reform. International Agricultural Trade Research Consortium (IATRC),
Working Paper, # 98-5, 1998b.
Hall, S., K. Cuthbertson, and M. Taylor. Applied Econometric Techniques. Hemel
Hempstead, UK: Philip Allan, 1992.
Hannah, A. and D. Spence. The International Sugar Trade. New York: John Wiley &
Sons, 1997.
Haraksingh, K. The Future of Preferential Trade and Sugar. Caricom Perspective.
67(June 1997): 10-13.
Harvey, A. Application of the Kalman Filter in Econometrics, in T.F. Bewley, (ed.),
Advances in Econometrics: Fifth World Congress. Cambridge: Cambridge
University Press, 1(1987): 285-313.
Harvey, A. Time Series Models (2nd ed.), Cambridge, Massachusets: The MIT Press,
1993.
Harvey, A. Trends, Cycles and Autoregressions. Economic Journal 107(1997): 192-
201.
Harvey, A., S. Henry, S. Peters, and S. Wren-Lewis. Stochastic Trends in Dynamic
Regression Models: An Application to the Employment-Output Equation. The
Economic Journal, 96(1986): 975-85.
Helpman, E. "Politics and Trade Policy." NBER Working Paper No. 5309, Cambridge,
MA, 1995.
Hildreth, C. and J.P. Houck, Some Estimators for a Linear Model with Random
Coefficients, Journal of the American Statistical Association, 63(1968): 584-
595.
International Sugar Organization (ISO). Quarterly Review. London, Various Issues
James, K. Prospects for the Caribbean Sugar Agro-Industry. Paper presented at the
UWI Ag.-50 Workshop on Agriculture in the Caribbean Issues and Challenges,
Trinidad and Tobago 16-23 August 1998.
Johansen, S. Statistical Analysis of Cointegrating Vectors. Journal of Economic
Dyaamics and Control, 13(1988): .231-254.

143
Johnston, J. and J. Dinardo. Econometric Methods, 4th edn. New York: Mcgraw-Hill,
1997.
Kalman, R.E. A New Approach to Linear Filtering and Prediction Problems,
Transactions ASME, Journal of Basic Engineering, 82(1960): 35-45.
Kalman, R.E. and R. S. Buey, New Results in Linear Filtering and Prediction Theory,
Transactions ASME, Journal of Basic Engineering, 83(1961): 95-103.
Kennedy, P. A Guide to Econometrics. 4th edn. Cambridge, Mass.: The MIT Press, 1998.
Kennedy, P. L., R. Harrison, M. Piedra. "Analyzing Agribusiness Competitivenes: The
Case of the United States Sugar Industry" International Food and Agribusiness
Management Review, l(1998):245-57.
Krugman, P. R. Increasing Returns, Monopolistic Competition and International Trade.
Journal of International Economics, 9 (1979)469-79.
Kwaitowski, O., P. Phillips, P. Schmidt, and J. Shin. Testing the Null Hypothesis of
Stationarity Against the Alternative of a Unit Root Journal of Econometics,
54(1992):. 159-78.
Landell Mills Commodities (LMC) Annual Review of the World HFCS Industry, in
Sweetener Analysis, New York, NY: LMC International, December, 1998.
Langham, M. and M. Mara. More Realistic Single Equation Models Through
Specification of Random Coefficients. Southern Journal of Agricultural
Economics, 5(1973)161-66.
Leu, G., A. Schmitz, and R. Knutson. "Gains and Losses of Sugar Program Policy
Options." American Journal of Agricultural Economics, 69(1987):591-602.
Lindert, P. H. and T. A. Pugel. International Economics. 19th edn., Chicago: Erwin, 1996.
Lopez, R. A., and J. L. Sepulveda. Changes in the US Demand of Sugar and
Implications for Import Policies. Northeast Journal of Agricultural Resource and
Economics. 14 (1985): 177 82.
Lord, R. Sugar: Background for 1995 Farm Legislation US Department of
Agriculture, Economic Research Service, Commodity Economics Division.
Report #711 June, 1995.
Lucas, R. Econometric Policy Evaluation: A Critique," in The Phillips Curve and Labor
Markets, K. Brunner and A Meltzer, eds., Carnegie Rochester Conference Series
on Public Policy, Amsterdam: North-Holland, 1 (1976): 19-46.
Mankiw, N. G. Principles of Economics. New York: The Dryden Press, 1997.

144
Marks, S. A Reassessment of Empirical Evidence on the US Sugar Program, in The
Economics and Politics of World Sugar Policies, S. V. Marks and K. E. Maskus ,
eds., Ann Arbor: The University of Michigan Press, p. 79 108 ,1993.
McFadden, D. Cost, Revenue nd Profit Funcions, in Production Economics: A Dual
Approach to Theory and Applications, M. Fuss and D. McFadden, eds.,
Amsterdam: North-Holland Publishing Co., 1978
Njchols, D. F. and A.R. Pagan, Varying Coefficient Regression. in Handbook of
Statistics, Vol. 5 E. J. Hannan, P.R. Krishnaiah, and M. M. Rao, eds., Amsterdam
:North-Holland, p. 413-49, 1983.
Nicholson, W. Microeconomic Theory: Basic Principles and Extensions, 6th edn., New
York: The Dryden Press, 1995.
Pindyck, R. and D. Rubinfeld. Econometric Models and Economic Forecasts. 3rd edn.,
New York: McGraw-Hill Inc., 1991.
Polopolus, L. and J. Alvarez. Marketing Sugar and Other Sweeteners. Amsterdam:
Elsevier Science Publishers, 1991.
Rausser, G., E. Lichtenberg, and R. Lattimore. "Development in Theory and Empirical
Applications of Endogenous Governmental Behavior," in New Directions in
Econometric Modeling and Forecasting in US Agriculture, Gordon C. Rausser,
ed., Amsterdam: North Holland Publishing Co., 1982.
Rendleman, M. and T. Hertel. Economy Wide Effects of the Sugar Program. Paper
presented at American Agricultural Economics Association (AAEA), 1989
Rosenberg, B."The Analysis of a Cross-Section of Time Series by Stochastically
Convergent Parameter Regression," Annals of Economic asn Social Measurement,
2(1973):461 -4.
Schmitz, A. The Free Trade Myth and the Reality of European Subsidies. Paper
presented at the 12th International Sweetener Symposium, US Sweetener Policy in
the Farm Bill: Charting a Course for the Future, Washington DC(27 June), 1995.
Schmitz, A., and L. Polopolus. Alcohol Fuel Tax Credit Policy: Sugar, Com and the
Environment in F. Casey, A. Schmitz, S. Swinton and D. Zilberman, eds.,
Environmental Regulation and Technology Adoption in Agriculture. Boston, MA:
Kluwer Academic Publishers, (Forthcoming).
Schmitz, A. and D. Christian. The Economics and Politics of US Sugar Policy. in The
Economics and Politics of World Sugar Policies, S. V. Marks and K. E. Maskus,
eds., Ann Arbor: The University of Michigan Press, p. 49 -78, 1993.
Shen, C., D. Hakes, and K. Brown. Time-Varying Response of Monetary Policy to
Macroeconomic Conditions Southern Economic Journal 65(1999): 584-93

145
Shim, J., and J. Siegel. Dictionary of Economics. New York :John Willey and Sons,
1995.
Sudaryanto, T. The Potential Impact of Liberalized Trade Policies in the United States
and in the European Economic Community on International Markets for Sugar.
Ph.D. Dissertation, North Carolina State University, 1987.
Tanyeri-Abur, A. An Agricultural Sector Analysis of the United States Sugar Import
Policy. Ph.D. Dissertation, Texas A&M University, 1990.
Thomas, C. Y. Sugar : Threat or Challenge? Ottawa: International Development
Research Center, 1985.
Thompson, R. A Survey of Recent US Development in International Agricultural Trade
Models. Bibliographies and Literature of Agriculture, No. 21, Economic Research
Service, USDA, 1981.
US Department of Agriculture (USDA). Economic Research Service, Commodity
Economic Division, Sugar and Sweeteners Situation and Outlook Report.
Washington, DC, Various Issues.
US Department of Agriculture (USDA), Economic Research Service, Commodity
Economic Division, Sugar and Sweeteners Situation and Outlook Yearbook.
Washington, DC, Various Issues.
US Department of Agriculture. Economic Research Service (USDA), Commodity
Economic Division, Free Trade in the Americas. Situation and Outlook Series,
WRS-98-1, Washington, DC, November 1998.
Ward R. and L. Myers. Advertising Effectiveness and Coefficient Variation Over
Time. Agricultural Economics Research, 31(1979): 1-11.
Ward R. and D. Tilley. Time-varying Parameters With Random Components: The
Orange Juice Industry. Southern Journal of Agricultural Economics 12(1980):
5-13.
West Indian Commission. Time for Action: Report of the West Indian Commission
(Second Edition). Kingston : University of West Indies Press, 1993.
Wint, A. South-South Integration. Paper presented a Seminar on New Global Trading
Arrangement, organized by the UWI and Sponsored by the European Union.
Kingston, Jamaica November 19, 1997.
Yule, G. U. Why Do W e Sometimes Get Nonsense Correlation Between Time Series?
Journal of the Royal Statistical Society, 89(1926): 1-64

146
Zietz, J., and A, Valdes. The Potential Benefits to LDCs of Trade Liberalization Sugar
and Beef by Industrialized Countries. Weltwirtschaftliches Archiv 22 (1986): 439
-464.
Zitt, M. A Complex Invention Process: Industrial, Technological and Scientific
Dynamics in HFCS History. Paper presented at the 12th International Economic
History Congress, Spain, 1998.

BIOGRAPHICAL SKETCH
Edward A. Evans, alias Gilly, was born on November 3, 1956, in Jamaica W.I. to
Evroy and Elizabeth Evans. He graduated from St. Georges College, Kingston, Jamaica
W.I., in 1974. He enrolled in the University of West Indies (UWI), St. Augustine,
Trinidad and Tobago and graduated in 1979 with a Bachelor of Science (Hons.) in
agriculture. In 1979, he was awarded a UWI scholarship to pursue a master in
agricultural economics at the same university in the Department of Agricultural
Economics and Farm Management (DAEFM). He graduated with his Master of Science
(M.Sc.) degree in 1982 and taught for a short time at the UWI Extra Mural Studies
Department and in the DAEFM.
In 1983, he joined the staff of the Caribbean Community (CARICOM)
Secretariat, Guyana, as an economist and worked there for a period of eight and one-half
years conducting economic analyses in the area of international trade and trading
arrangements. In 1991, he joined the staff of the Caribbean Agricultural Research and
Development Institute (CARDI), Trinidad W.I., as a senior economist responsible for
performing all economic analysis pertaining to the CARDI executed, and Canadian
International Development Agency (CIDA)-funded Sheep and Goat Production and
Marketing Project.
In the fall of 1995, he enrolled in the Ph.D. program of the Food and Resource
Economics Department at the University of Florida. He completed his course work with a
GPA of 3.96 and was awarded the doctor of philosophy (Ph.D.) in 1999 with
specialization in international trade and policy analysis. He has written extensively and
has authored or coauthored more than 30 publications/presentations, many in the area of
international trade.
147

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosopl
Carlton G. Da
Distinguished Professor of Food and
Resource Economics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
ax R. Langham
Professor of Food and Resource
Economics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Ronald W. Ward
Professor of Food and Resource
Economics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation andjsjiilly-a^equate, in scppeand quality,
as a dissertation for the degree of Doctor o
2rt D. Emerson
Professor of Food and Resource
Economics
I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
Andrew Schmitz
Eminent Scholar in Food and Resource
Economics

I certify that I have read this study and that in my opinion it conforms to
acceptable standards of scholarly presentation and is fully adequate, in scope and quality,
as a dissertation for the degree of Doctor of Philosophy.
)avid A. Denslow
Distinguished Professor of Economics
This dissertation was submitted to the Graduate Faculty of the College of
Agriculture and Life Sciences and to the Graduate School and was accepted as partial
fulfillment of the requirements for the degree of Doctqrf>f Philosophy.
December, 1999
Den, College of Agriculture and Life
Sciences
Dean, Graduate School



59
applied economics and the model became a natural generalization of latent variable
models to a full dynamic framework A brief taxonomy of empirical applications of such
modeling in economics can be found in Engel and Watson (1980).
To illustrate the state-space model, first consider the following standard single
equation regression:
m
y, = a++ Yj 8ix, (3-6)
i=0
where yt is a scalar observable data series, x,, is the single observable explanatory
variable with the structure of the lagged coefficients being 60...6m, t is the trend variable
with level and slope coefficients a and P, respectively, and et is the stochastic disturbance
term distributed such that st ~ NID(0,oe2). Harvey (1987) asserts that while equation
(3.6) is typical of many regression equations and that while many economic time series
show a trend, it is equally apparent that unless the time period is fairly short the trends
cannot be adequately captured by a deterministic (straight line) representation. He
observed that a great deal of applied economic work fall prey to the above error, since
they start off by detrending the data by regressing on time, thus making all that follows
invalid. He further points out that there is no reason in principle, why the parameters of
the explanatory variables of time series should not also be allowed to vary over time. He
suggests that a better approach would be to formulate the model in a state-space
framework. Such a representation could account for any unobservable components as
well as time-varying parameters.
Following along the lines suggested by Harvey (1987), reparametizing equation
(3.6) in terms of a simplified version of the state-space model, appropriate for time-
varying parameters yields the following:


ACKNOWLEDGMENTS
I would like to express my profound gratitude to Dr. Carlton Davis, chairman of
my supervisory committee, for all his help and encouragement throughout the course of
this study. Dr. Davis' patience, understanding, guidance, assistance and constructive
criticisms given to me during the preparation of this dissertation are greatly appreciated.
His desire for excellence, and steadfast refusal to lower the bar kept my shoulders to the
wheel and has contributed immensely to my education and self-development.
Drs. Max Langham and Ron Ward, members of my committee, deserve special
mention. I have benefited a great deal from Dr. Langham's criticisms and vast experience,
and even though we did not always agree, I am often reminded that the beauty of a choir
comes from the fact that not everyone sings the same melody. Dr. Ward has provided
useful suggestions and has welcomed me into his home on several occasions during the
night.
To the other members of my committee, Drs. R. Emerson, A. Schmitz and D.
Denslow I am eternally grateful. Without their criticisms and useful suggestions this
effort would not have been possible. Special thanks is also due to Drs C. Moss and R.
Kilmer who made worthwhile contributions in the initial stages of the dissertation.
I gratefully acknowledge the financial support I received from the Department of
Food and Resource Economics. In this regard I must again say thanks to Dr. Davis and
will always remember the words he spoke which jolted me into finally accepting his
offer"there are more dogs than bones. I would also like to acknowledge the stipend I
iii


47
each country. Such models are usually less efficient as a means of examining the effects
of changes in policy variables, such as tariffs and transport costs. The main advantage of
these models is that they are usually much easier to solve (Thompson, 1981). Both spatial
and non-spatial models are based on a competitive framework. However, as shown by
authors such as Krugman (1979), models with imperfect competition features might be
more realistic when modeling international trade. Although it is recognized by this author
that of necessity any model is an abstraction of the real world, an attempt will be made to
address a few of the shortcomings alluded to.
Time Series Analysis
Aoki (1990) states that the main aims of time series analysis are one or a
combination of the following: 1) to describe the data behavior succinctly; 2) to explain
the behavior of the time series in terms of exogenous variables; and 3) to forecast and in
some cases to control. Harvey (1987) makes a distinction between a pure time series
model and a structural time series model. The distinguishing feature of the two is that no
attempt is made in the former to formulate behavioral relationships among variables, in
contrast to the latter. Rather, the movements in the variable are 'explained' solely in terms
of its own past, or by its position in relation to time.
In utilizing time series data to carry out analysis several issues must be confronted
including a determination of whether the data are: 1) serially correlated; 2) stationary; 3)
possess a deterministic or stochastic trend; and 4) the stability of the parameters (time-
varying parameters). In cases where two or more time series are to be regressed on each
other there is the added concern of cointegration and guarding against spurious
regression. The following discussion provides brief insight into each of these issues.


29
reduction in the quota levelsimplies a considerable loss of real earnings for the quota
holders.
A stated objective of the US Sugar Program is to ensure a reliable supply of sugar to
American consumers at competitive prices, while providing some stability for American
sugar interests (growers and processors). Key components of the program include: 1)
loan rates; (2) an implicit market stabilization price (MSP); and 3) Tariff-Rate Quota
(TRQ). These components include:
1. Loan rates-a loan rate of US $0.18 per pound of raw sugar is made available to
processors. The loan rate of 22.9 cents per pound for refined beet sugar was
historically set in relation to raw sugar with a prescribed formula. Loans were
previously non-recourse in that, at any time, the processors could refuse to repay the
loan and the government would accept the sugar in lieu of cash. However, with the
implementation of the Federal Agricultural Improvement and Reform Act (FAIR) of
1996 current loan rate programs are to operate on a recourse basis under certain
conditions. Whenever the tariff-rate quota (TRQ, discussed below) is 1.36 MMT or
less, the USDA can demand repayment of the loan with interest at maturity,
regardless of the price of sugar. The loans are considered to be non-recourse only
when the TRQ is above the stipulated 1.36 MMT. In addition, loan rates were fixed at
the 1995 levels of 18.0 and 22.9 cents per pound for raw and refined sugar,
respectively.
2. Market stabilization price (MSP)a major stipulation of the program in the past was
that it should run at no cost to the government. To ensure that such provision was
met, an MSP was explicitly established at a level to ensure that commercial market


BIOGRAPHICAL SKETCH
Edward A. Evans, alias Gilly, was born on November 3, 1956, in Jamaica W.I. to
Evroy and Elizabeth Evans. He graduated from St. Georges College, Kingston, Jamaica
W.I., in 1974. He enrolled in the University of West Indies (UWI), St. Augustine,
Trinidad and Tobago and graduated in 1979 with a Bachelor of Science (Hons.) in
agriculture. In 1979, he was awarded a UWI scholarship to pursue a master in
agricultural economics at the same university in the Department of Agricultural
Economics and Farm Management (DAEFM). He graduated with his Master of Science
(M.Sc.) degree in 1982 and taught for a short time at the UWI Extra Mural Studies
Department and in the DAEFM.
In 1983, he joined the staff of the Caribbean Community (CARICOM)
Secretariat, Guyana, as an economist and worked there for a period of eight and one-half
years conducting economic analyses in the area of international trade and trading
arrangements. In 1991, he joined the staff of the Caribbean Agricultural Research and
Development Institute (CARDI), Trinidad W.I., as a senior economist responsible for
performing all economic analysis pertaining to the CARDI executed, and Canadian
International Development Agency (CIDA)-funded Sheep and Goat Production and
Marketing Project.
In the fall of 1995, he enrolled in the Ph.D. program of the Food and Resource
Economics Department at the University of Florida. He completed his course work with a
GPA of 3.96 and was awarded the doctor of philosophy (Ph.D.) in 1999 with
specialization in international trade and policy analysis. He has written extensively and
has authored or coauthored more than 30 publications/presentations, many in the area of
international trade.
147


62
equation. Also, it is assumed that et ~ NID (0, oe2), and cot ~ NID (0, oj ). Although x,,
Gt, ,oe2 and oj may ultimately depend on a set of unknown parameters they are for the
purpose of the Kalman filter regarded as being fixed and known. Gt, oe and are
defined as the hyperparameters and are normally supplied by the researcher. In cases
where such information is not provided each defaults to the identity matrix. The
parameter of interest in the model is the state variable 5t, which evolves over time.
The Kalman filter uses a recursive estimation procedure to update the state as new
observations become available. In order to begin the procedure one must specify values
for ,oe2, Oa,2, xt, G, and the starting value for 5o. However, since such values are not
usually available the usual procedure is to construct starting values from the observations
themselves. Thus, in the case of a classical linear regression model with k explanatory
variables the starting values can be obtained by applying ordinary least squares (OLS) to
the first k observations with G = I and o(02 = 0.
The operation of the Kalman filter is easier understood when interpreted from a
Bayesian point of view. In this regard the goal of estimating 5t can be though of as being
carried out in two stages (Shen et al, 1999). In the first stage the 5t is estimated prior to
observing yt through equation 3.1 la. This can be represented as:
8t=GtSul (3.11b)
where, 8X is regarded as the best guess of the true value based on all information
available up to time t-1. In the second stage when the actual value of yt becomes
available, 8t is updated as follows:
t = Gtt., +A,e,
(3.11c)


43
estimated that free trade would result in the US prices for sugarcane, sugar beet and corn
sweeteners declining by 34%, 31% and 10%, respectively. Com sweetener byproducts
would decline by 2% but prices would increase by 12 %. In terms of the changes to
output, they estimated that US sugarcane production would decline by 77%, sugar beet
production by 47%, and com sweetener production by 2%. This study was an
improvement over many of the previous studies in that an attempt was made to seriously
analyze the interrelationship among the various segments of the sweetener market.
Tanyeri-Abur (1990) analyzed the implications of current and potential changes in
the US sugar policy with a price endogenous agricultural sector model. The sugar and
HFCS markets and the industries, which used these two sweeteners, were modeled.
Several policy runs were made under varying assumptions pertaining to the level of
imports, target prices and substitution relationship between sugar and HFCS. The results
were compared with the based year (1996) in which imports of raw sugar was restricted
to 1.6 million short tons. Her work indicated that: 1) the removal of US sugar quotas
would benefit US consumers by $657 million and increase net foreign surplus by $350
million; and 2) that US producers surplus would decline by $288 million leaving a net
welfare gain of about $718 million. However, when the agricultural sector with all
commodities was considered, there was a reduction in net social welfare for the sector of
$1.2 billion dollars despite an increase in total consumer surplus of $1.4 billion. She
explained that this result was due mainly to the increase in government payments that
would result as the price of cotton and rice fall as land was shifted from sugarcane to
these crops.


135
trade flows are mutually complementary. While to some extent there is some
complementarity of trade flows within the proposed FTAA, the majority of the
prospective members are major producers and exporters of sweeteners. Any negotiated
agreement to which CARICOM is a part should therefore include special provisions for
the CARICOM sugar industry and accommodate the sugar arrangement embodied in the
EU/ACP sugar protocol.
The HFCS in the future, whether within an FTAA framework or not, is going to
pose a formidable challenge to sugar as the dynamics of the sweetener market continues
to unfold. Consequently, to the extent possible a third recommendation is that the
subregional sugar industry must pursue an active program aimed at diversifying its end
product as well as the uses of sugar cane.
Limitations of the Study and Suggestions for Further Research
The research undertaken centered on assessing the dynamics of the US sugar and
HFCS sugar markets and on drawing of implications for the CARICOM sugar producers
within the framework of the proposed FTAA. The following limitations of this study
suggest areas for future investigation. First, although some attempts were made to explore
the dynamics of the situation in the theoretical model and more so in the empirical
estimation of the demand curve, the setting was essentially static. As pointed out
elsewhere, static models can only provide useful summary statistics and indicate the
direction of the outcome. Thus, while being a useful first step in an attempt to understand
complex issues, fully dynamic models are required to capture the full effects and better
represent the reality of the assumed changes. A Computable General Equilibrium (CGE)
model which takes into consideration some of the issues raised in this study might offer a


To Sabrina, Latoya, Akeem, Attli and ma dodo


compete with larger and more efficient sugar-producing FTAA member countries-such
as Brazil, Colombia and Guatemalafor a share of the regional sugar market. Third, and
the focus of this dissertation, a substantial portion of the regional sugar market might be
eroded as a result of the successful penetration of high fructose com syrup (HFCS)
produced in the United States.
A qualitative assessment of the potential economic impact of changes to the US
sugar program and an expanded HFCS industry on the CARICOM sugar industry was
conducted. First, a conceptual model of the US sugar and HFCS markets was developed,
and potential trade and economic impacts were discussed. Second, an empirical model of
the demand for HFCS in the United States was constructed. To capture the dynamics
within the industry a state-space representation utilizing the Kalman filter was employed.
The maturity of the HFCS industry was assessed, and implications for the various sugar
concerns were drawn.
The findings substantiated the view that CARICOM membership in an FTAA
could have potential negative economic impact on the CARICOM sugar industry and that
an FTAA could further depress, rather than increase, the world market sugar price.
Finally, it was concluded that the US HFCS industry is entering a mature phase, forcing
the need for com refiners to actively seek external markets for their products.
xi


73
The US Sugar Subsector
Focusing first on sugar, it should be noted that several studies have attempted to
model the US sugar market (Haley 1998b; Atlas, 1996; Bemirschka, Koo and Lou, 1996;
Marks 1993; Schmitz and Christian, 1993 ; Barros, 1992; Sudaryanto, 1987; Leu et al,
1987; Zietz and Valdez, 1986; Gemmill, 1976). Sudaryanto pointed out that many of the
studies fail to distinguish between direct and indirect consumption. They omit variables
explaining the derived demand characteristics of the consumption pattern. This study
recognizes the point made by Sudaryanto and others (Haley, 1998b; Barros, 1992) on the
need to disaggregate the market for sugars into derived and direct demand components.
However, a slightly different approach was taken in separating the market demand
components.
For convenience, a distinction is made between crystallized sugar (CS) and liquid
16
sugar (LS) In this study, crystallized (tabletop) sugar include sugars used both by
households and those segments of the industrial sugar users, in which it is currently
technically impossible to substitute sugar with liquid HFCS. The importance of this
distinction will become evident later in the analysis. However, for the time being there is
some evidence suggesting that at this juncture, it is in respect of the demand for liquid
sugar that a structural shift in demand has taken place. Only the liquid HFCS competes
with sugar. As pointed out earlier, liquid sugar and liquid HFCS are regarded as almost
perfect substitutes (Marks, 1993; Barros, 1992). Further supporting evidence for this
16 Refined sugar is available in three main product categories: white granulated sugar, liquid sugar and
specialty sugars. Granulated sugar is the most common form of sugar used in the households. Liquid sugar
is a mixture of water and sugar that is preferred by some manufacturers and specialty sugars include icing
sugar, brown or yellow sugars. Chemically, there is no difference between liquid sugar and crystallized
(granulated and specialty).


22
On the consumption side the US and Brazil were the main consumers each
accounting for about 28% of regional consumption. Next in line were Mexico, Argentina,
Columbia and Canada, respectively. The fourteen CARICOM countries consumed 0.32
MMT or about 1% of regional consumption.
Table 2.4. Leading Hemispheric Sugar Producers and Consumers, 1997-98
Producers
Consumers
Country
Quantity
%of
Country
Quantity
%of
(MMT)
Hemispheric
(MMT)
Hemispheric
Production
Consumption
Brazil
15.70
39.6
US
8.91
28.4
US
7.28
18.4
Brazil
8.80
28.0
Mexico
5.49
13.9
Mexico
4.24
13.5
Columbia
2.15
5.4
Argentina
1.45
4.6
Guatemala
1.79
4.5
Colombia
1.37
4.4
Argentina
1.75
4.4
Canada
1.27
4.0
CARICOM8
0.75
1.9
Peru
0.91
2.9
a] Represents the total output of six CARICOM countries
Source: International Sugar Organization
With respect to the hemispheric trade, in the 97/98 period total exports amounted
to 12.77 MMT representing about 35.9% of world exports, estimated at 35.59 MMT. In
contrast, total imports amounted to just 4.46 MMT or about 12.5% of world imports.
Consequently, the region is a net exporter of 8.3 MMT (excluding net exports from Cuba
of 2.3 MMT). Table 2.5 shows the leading sugar exporters and importers in the region.
The Table also shows exports as a percentage of domestic production and imports as a
percentage of consumption. The leading sugar exporters within the region are Brazil,
Guatemala, Mexico and Columbia, respectively. Of interest however, is the fact that the
CARICOM exports, while accounting for only 5.1% of total regional exports, represent
more than 85% of their domestic production--an indication of the importance of this
commodity in trade for these countries.


144
Marks, S. A Reassessment of Empirical Evidence on the US Sugar Program, in The
Economics and Politics of World Sugar Policies, S. V. Marks and K. E. Maskus ,
eds., Ann Arbor: The University of Michigan Press, p. 79 108 ,1993.
McFadden, D. Cost, Revenue nd Profit Funcions, in Production Economics: A Dual
Approach to Theory and Applications, M. Fuss and D. McFadden, eds.,
Amsterdam: North-Holland Publishing Co., 1978
Njchols, D. F. and A.R. Pagan, Varying Coefficient Regression. in Handbook of
Statistics, Vol. 5 E. J. Hannan, P.R. Krishnaiah, and M. M. Rao, eds., Amsterdam
:North-Holland, p. 413-49, 1983.
Nicholson, W. Microeconomic Theory: Basic Principles and Extensions, 6th edn., New
York: The Dryden Press, 1995.
Pindyck, R. and D. Rubinfeld. Econometric Models and Economic Forecasts. 3rd edn.,
New York: McGraw-Hill Inc., 1991.
Polopolus, L. and J. Alvarez. Marketing Sugar and Other Sweeteners. Amsterdam:
Elsevier Science Publishers, 1991.
Rausser, G., E. Lichtenberg, and R. Lattimore. "Development in Theory and Empirical
Applications of Endogenous Governmental Behavior," in New Directions in
Econometric Modeling and Forecasting in US Agriculture, Gordon C. Rausser,
ed., Amsterdam: North Holland Publishing Co., 1982.
Rendleman, M. and T. Hertel. Economy Wide Effects of the Sugar Program. Paper
presented at American Agricultural Economics Association (AAEA), 1989
Rosenberg, B."The Analysis of a Cross-Section of Time Series by Stochastically
Convergent Parameter Regression," Annals of Economic asn Social Measurement,
2(1973):461 -4.
Schmitz, A. The Free Trade Myth and the Reality of European Subsidies. Paper
presented at the 12th International Sweetener Symposium, US Sweetener Policy in
the Farm Bill: Charting a Course for the Future, Washington DC(27 June), 1995.
Schmitz, A., and L. Polopolus. Alcohol Fuel Tax Credit Policy: Sugar, Com and the
Environment in F. Casey, A. Schmitz, S. Swinton and D. Zilberman, eds.,
Environmental Regulation and Technology Adoption in Agriculture. Boston, MA:
Kluwer Academic Publishers, (Forthcoming).
Schmitz, A. and D. Christian. The Economics and Politics of US Sugar Policy. in The
Economics and Politics of World Sugar Policies, S. V. Marks and K. E. Maskus,
eds., Ann Arbor: The University of Michigan Press, p. 49 -78, 1993.
Shen, C., D. Hakes, and K. Brown. Time-Varying Response of Monetary Policy to
Macroeconomic Conditions Southern Economic Journal 65(1999): 584-93


82
number of firms that use HFCS as input into their production process. With respect to
the second assumption, this again seems plausible given that, the industry is highly
capital intensive and that there currently exist a considerable amount of excess capacity
within the HFCS industry In addition, the amount of com used in the industry is less
than 10% of the overall US com production, and consequently a moderate increase in
demand for HFCS is not expected to have any major impact on the market price of com.
Finally, the industry is highly capital intensive and as such one might expect that the cost
of an additional unit energy will remain fairly constant. On this basis, and in the absence
of concrete cost of production data, the short-run marginal cost curve of the industry (the
sum of the marginal cost over the variable cost) is conceptualized as shown in Figure 4.6.
The curve is drawn showing a section which is nearly perfectly elastic (constant marginal
cost) over some quantity range and becoming vertical as output capacity is reached.
Price
Fig. 4.6. Shift in hypothetical HFCS short-run marginal cost curve
In Figure 4.6, SMChi represents the industry short-run marginal cost function in
period one and SMCh2 is the short-run marginal cost curve in period two. As drawn the
shift in the marginal cost curve represents an overall reduction in the cost of production.


18
consumers in 1997-98 are shown in Table 2.2. The European Union, followed by Brazil,
continues to dominate production. The United States is listed as the fifth largest sugar
producer, reflecting growth in domestic sugar beet production. In terms of consumption,
the leading country is India, followed by the European Union and China. The United
States occupies the fourth position in spite of its massive consumption of HFCS. As seen
from the Table, both production and consumption are relatively concentrated, with
Fig. 2.1. World and US refined and raw real sugar prices, 1985-98 (1992=100)
Source: Compiled with USDA sugar data and GDP deflator
the top five producing countries (out of 120 sugar-producing countries) and top five
consuming countries (out of all countries) accounting for 53 percent and 45 percent of
global production and consumption, respectively. Finally, it should be observed that
developing countries occupy three of the top five positions in both cases, reinforcing the
importance of this commodity to these economies.


CHAPTER 1
INTRODUCTION
The global economic environment is in a state of rapid transition. The new
dispensation in both the developed and developing countries is the opening up of national
and regional economies to market forces. In the developed countries, the principal forces
leading to trade liberalization stemmed from: 1) the growing budgetary pressure of
having to maintain protective regimes; 2) growing regionalism; and 3) multilateral
negotiations such as the General Agreement on Tariff and Trade (GATT/WTO).
Developing countries, on the other hand, have tended to liberalize their trading regimes
either unilaterally or as a result of following the dictates of multilateral financial and
developmental institutions (Wint, 1997). Whatever the reason, the result has been a
dramatic reduction in trade barriers.
For developing countries, the wide spread process of trade liberalization and
globalization is only one element of the new global dispensation to which they must
respond. An important related element is the erosion of non-reciprocal preferential
treatment under the Generalized System of Preferences (GSP) or other preferential
regimes such as Lome1, which in many cases has been a mainstay of several of the
economies. At the same time, LDCs excluded from regional blocsas in the case for the
1 Lome Convention is a contractual non-reciprocal trading arrangement between African, Caribbean and
Pacific (ACP) countries and The European Union (EU). The first Lome Convention (Lome 1) was
established in 1975 and the Convention is now in its fourth round (Lome 4). It expires in the year 2000 and
is currently being re-negotiated albeit, under new philosophical circumstances.
1


5
matters iirther, a considerable amount of the land used in the subregion to produce sugar
cane is marginal with respect to sugar cane production but has few viable alternatives.
Also, from a wider socioeconomic point of view, labor and capital could become socially
disruptive in these small undiversified economies if other alternatives were pursued too
quickly (Haraksingh, 1997).
Notwithstanding these drawbacks, the industry remains of considerable import to
the subregion, provides much of the valuable foreign exchange earned, and is a major
contributor to the employment of labor. For example, the industry provides over one half
of the foreign exchange earnings in St. Kitts and Nevis and employs about 97 thousand
persons or about 5 % of the active working population within CARICOM (James, 1998).
Moreover, the socioeconomic importance of sugar production to these countries goes
much further than the above figures might suggest as many of these industries do play a
vital role in the maintenance and upgrading of rural infrastructure, health, recreational
and educational facilities.
With the general consensus that CARICOM should seek to become part of the
proposed FTAA, there has been a genuine concern with regards to the future of its sugar
industry (Group of Experts, 1998). Such a move toward integrating the economies brings
with it the challenge of competing in a liberalized trading environment without the
benefit of preferences to which the subregion has grown accustomed.
The following three interrelated issues with respect to the subregional sugar
industry are of importance. First, there is the possibility that should the subregion pursue
such a path it could be in violation of the terms and conditions of its Lome Agreement.
Specifically, article 174 section 2a of the Agreement forbid recipients to enter into any


3
will change how the subregion conducts its business with the rest of the world and how it
earns its foreign exchange.
Faced with a set of perplexing choices, the CARICOM countries have begun to
systematically weigh the benefits and costs of embarking on one of the following three
strategies: 1) becoming a part of a Free Trade Area of the Americas (FTAA) with the US
as the central figure; 2) seeking to form a Free Trade Area with the EU as part of the new
Lome arrangement; and 3) remaining outside of the formation of any free trade area with
the US or the EU. The emerging consensus favors (1). Irrespective of which strategy is
chosen, there will be serious internal economic conflicts because of gains and losses at
the sectoral and industrial levels. Moreover, the decisions made could have irreversible
economic consequences.
The CARICOM sugar industry, a major contributor to the economic and social
stability of the subregion, is undoubtedly one of the more important industries that will be
affected as the new economic arrangements emerge. Within the framework of the
proposed FTAA, for example, the industry is likely to be faced with the additional
economic challenges associated with significant structural changes that are occurring
within the US-one of its major sugar markets. In particular, this challenge could come as
a result of the rapid development of the US High Fructose Com Syrup (HFCS) Industry
and its expansion in a much freer trading environment.
Statement of Problem
Problematic Situation
The CARICOM countries are characterized as having certain features which,
when taken in combination, makes them extremely vulnerable. These features include


91
barrier members. A Customs Union on the other hand, is where, in addition to the
removal of all barriers to trade among the members, there is the adoption of common set
of external barriers thus avoiding the need for custom inspection at the internal borders
(Lindert and Pugel, 1996). However, in practice it is not uncommon to find a hybrid of
the two integration forms such as is the case of NAFTA, where most but not all of the
barriers to trade are removed and there is an agreement on phasing and harmonization of
barriers to third countries. While details of the specific form that the proposed FTAA will
take are not yet available, one can nonetheless advance a number of scenarios and
examine the consequences for the sugar and HFCS industries.
In advancing these scenarios cognizance is take of the changes to the US sugar
program as a result of the FAIR 1996 Act. In particular, note is taken of the attempt by
the US authority to make the program much more market oriented by the introduction of
the recourse loan policy, which essentially weakens the domestic guaranteed price. At the
same time, the elimination of the no-net-cost provision of the program makes the drawing
of any definite conclusion with regard to future state of the program ambivalent. In this
regard consideration is given to the view of Schmitz and Polopolus (forthcoming) that the
changes to the program as reflective of the FAIR 1996 are minor. However, as the com
refiners seek to expand their markets overseas, in light of: 1) the importance of sugar to
the prospective members of the FTAA; 2) developments within the WTO/GATT
negotiations: 3) internal domestic pressure from other commodity groups for which the
changes under FAIR were more severe; and 4) the weakened (although still powerful) US
sugar lobby it is postulated that the current US sugar program will need to undergo
substantial changes within the FTAA framework. In this connection, the following


130
extent to which HFCS could penetrate the non-US market and displaces a portion of the
existing sugar trade levels.
As in the case of the intercept coefficient, the own-price elasticity was significant
and elastic. The dynamic path of adjustments showed that the absolute elasticity had been
decreasing over time, implying that the demand for the commodity has become relatively
less elastic. This finding is consistent with the views of some industry experts that in
light of the considerable amount of HFCS product development over the years, the
product is establishing its own market and is being regarded in certain uses as a superior
product to liquid sugar. An implication of this is that the shift away from liquid sugar
which has occurred in the sweetener market in some uses, might be structural and that a
simple reduction in the price of sugar might not cause a major reversal of the current
pattern of sweetener used in the US.
Unlike the other two coefficients considered, the cross-price elasticity of HFCS
with respect to sugar was non significant. This may suggest that at present the change in
the sugar price no longer has a significant impact on the demand for sugar. This however
does not necessarily imply that at an earlier time period the impact of the sugar variable
was not significant. Again this finding is consistent with the view that the HFCS is
establishing its own market. The dynamic path of adjustments of this parameter estimate
showed that the relative cross-price elasticity had decreased considerably over time. This
suggests that over time the impact of a one-percent change in the price of sugar is having
a smaller effect on the relative quantity of HFCS demanded.
The above findings make the US sweetener market and how it will react less
predictable with the liberalization of sugar and with the price of sugar falling to a level


34
Owing to the problems alluded to earlier with regards to storing the commodity,
very little is held in stock. Of particular interest is the fact that, between 1990 and 1997,
world consumption of HFCS increased at an annual rate of 4.0% while consumption of
sugar grew at a lackluster rate of 1.5% per annum. With the demand growth for sugar
failing to keep pace, HFCS has captured an increasing share of the combined sugar
market for these sweeteners, increasing from 7.2% in 1990 to an estimated 8.5% in 1997
(LMC International, 1998).
Table 2.9. World HFCS Production, 1989-97 ('000 metric tons dry basis)
Country
1989
1990
1991
1992
1993
1994
1995
1996
1997
US
5,345
5,677
5,852
6,038
6,459
6,813
7,143
7,408
7,711
Japan
744
783
794
747
727
806<