Citation
Economic reforms and agricultural supply response in Jamaica

Material Information

Title:
Economic reforms and agricultural supply response in Jamaica
Creator:
Ballayram, 1950-
Publication Date:
Language:
English
Physical Description:
xvi, 192 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Agrarian reform ( jstor )
Agriculture ( jstor )
Coffee industry ( jstor )
Crops ( jstor )
Economic reform ( jstor )
Fertilizers ( jstor )
Mathematical variables ( jstor )
Price shocks ( jstor )
Prices ( jstor )
Supply ( jstor )
Agriculture and state -- Jamaica ( lcsh )
Crops -- Jamaica ( lcsh )
Dissertations, Academic -- Food and Resource Economics -- UF ( lcsh )
Economic policy -- Jamaica ( lcsh )
Food and Resource Economics thesis, Ph.D ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph.D.)--University of Florida, 2001.
Bibliography:
Includes bibliographical references (leaves 183-191).
General Note:
Printout.
General Note:
Vita.
Statement of Responsibility:
by Ballayram.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
026974320 ( ALEPH )
47260198 ( OCLC )

Downloads

This item has the following downloads:


Full Text











ECONOMIC REFORMS AND AGRICULTURAL SUPPLY RESPONSE
IN JAMAICA
















By

BALLAYRAM


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


2001















ACKNOWLEDGEMENTS

First, I would like to express my deepest gratitude to

my wife for all her support throughout my studies here at

the University of Florida. Her support in this process was

overwhelming and unselfish. My daughters were also very

understanding, supportive, and encouraging, and I thank

them sincerely for the sacrifices they had to make so that

I could complete my graduate studies. My parents and in-

laws (all now deceased) gave me encouragement and support,

both financial and otherwise, in my quest to achieve a

higher education. Without their love, confidence and moral

support, none of this would have been possible. I owe them

an eternal debt of gratitude, because their support came at

a time when the cost of higher education was far beyond the

reach which their economic resources could accommodate.

I would also like to express my deepest gratitude to

Dr. Carlton G. Davis, chairman of my supervisory committee.

He opened the door for me at the University of Florida, and

provided me with countless hours of individual attention

during my course of studies. On numerous occasions he went

beyond what would normally be the call of duty of an









academic supervisor, to provide me with the support that

enabled me to continue my studies.

I would also like to acknowledge the contributions

from the other members of my supervisory committee: Dr.

Robert Emerson, Dr. Clyde Kiker, Dr. Richard Kilmer and Dr.

David Denslow, who provided me with valuable suggestions

during the preparation of this dissertation.

I also benefited immensely, in my course of studies at

U.F., from discussions with Dr. Max Langham, Dr. Charles

Moss, Dr. Ronald Ward and Dr. James Seale, Jr.

Financial support from the Food and Resource Economics

Department is greatly acknowledged. Dr. Carlton Davis'

efforts in securing this funding for me are greatly

appreciated.

Finally, I wish to acknowledge the excellent computer

guidance and support services which Alex Heyman, Roger and

Laura Clemons, Ed Howard and Juan Carlos, of the Food and

Resource Economics Support Center, provided throughout my

studies here at U.F.


iii















TABLE OF CONTENTS


page


ACKNOWLEDGEMENTS ...... .................. ii

LIST OF TABLES ....... .................... vi


LIST OF FIGURES ........


ABSTRACT . . . . . .

CHAPTERS

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

1.1 Background to Jamaica's Economic
Reforms ..... .................
1.2 Problem Statement and Study
Justification ... ................
1.3 Objectives of the Study .. ...........
1.4 Hypotheses ...............
1.5 Conceptual Issues ..... ............
1.6 Procedural and Methodological
Considerations ............
1.7 Summary ....... ..................

2. ECONOMIC STRUCTURE, GROWTH, AND POLICY
REFORMS IN THE JAMAICAN ECONOMY, 1962-1999


2.1 Structure of the Economy and
Economic Growth: An Overview ...
2.2 Output Trends and Policies in
Jamaican Agriculture ........
2.3 The Economic Reforms .........
2.4 Summary ....... ................

3. METHODOLOGICAL AND EMPIRICAL ISSUES IN
ECONOMIC REFORMS AND SUPPLY RESPONSE .

3.1 Review of the Economic Reform
Literature . . . .


* 26

* 30
* 38
* 46


. . ix


. xv


2

8
* 10
. 11
* 12.

* 20
* 22









Before-After Approach ....
With-Without Approach ....
Comparison of Simulations


Approach ....


3.2 Preliminary Issues in Modeling Supply
Response in Jamaica .. ...........
3.3 Error Correction Model ........
3.4 An Error Correction Model for Crop
Supply Response in Jamaica .....
3.5 The Data . . . .
3.6 Summary ....... ................

4. EMPIRICAL ESTIMATION OF CROP SUPPLY RESPONSE

4.1 Motivations for Using Cointegration
Analysis . . . .
4.2 Definition of Variables ............
4.3 Testing for Stationarity .......
4.4 Estimating Long-run Supply Response .
4.5 Analysis of Short-run Dynamics ....
4.6 Summary ....... ................

5. SUPPLY RESPONSE AND COUNTERFACTUAL ANALYSIS

5.1 Before-After Analysis ... ..........
5.2 Counterfactual Analysis ............
5.3 Summary ....... ................

6. CONCLUSIONS AND POLICY IMPLICATIONS ...


. 75
. 79
* 87
* 90
* 106
. 136

* 137


137
145
164


. 165


APPENDICES


A. PRICE OF SELECTED CROPS PRODUCED IN JAMAICA,
1962-1999 (J$) ..... ...............


. 174


B. QUANTITY OF SELECTED CROPS PRODUCED IN
JAMAICA, 1962-1999 (METRIC TONES) .. ....... ..175

C. SELECTED ECONOMIC STATISTICS ON JAMAICA ..... ..176

D. DIAGNOSTIC STATISTICS FOR MODEL SPECIFICATION
AND TEST STATISTICS FOR COINTEGRATION ..... ..177

REFERENCES ......... ...................... 183

BIOGRAPHICAL SKETCH ....... .................. .192


3.1.1
3.1.2
3.1.3


. . 54















LIST OF TABLES


page


Table


2.1 Sectoral Contribution to Real Gross
Domestic Product (Period Averages,
Percentages) ....... .................. 27

2.2 GDP and Sectoral Growth (1995=100)
(Percentage) ....... .................. 28

2.3 Growth Rates of Total Agricultural and
Broad Agricultural Aggregate Indexes ...... ..30

2.4 Composition of Agricultural Output
(J$M, 1995=100), Selected Years ... ......... .32

2.5 Agriculture Sub-sectors as a Percentage
of Total Agriculture (Period Averages).... 32

2.6 Growth Rates of Agricultural Components
(Percentages) .................... ...... 34

2.7 Average Rates of Growth of Farmgate
and F.O.B. Prices, Output and the
Nominal Protection Coefficient (NPC)
for the Period 1970-78, Jamaica ... ......... .36

2.8 Net Barter Terms of Trade and Agricultural
Income Terms of Trade (1980=100)
(Period Averages) ...... ................ .38

2.9 Outcomes of Economic Reform Policies
in Jamaica, 1986 vs. 1995 .... ............ .45


3.1 Empirical Studies on Agriculture
Supply Responses in Jamaica ..........

4.1 Alternative (Substitute) Crops in Each ECM

4.2 Unit Root Tests--Prices and Quantities
in Levels ...... .................


. 56

. 84


. 88









4.3 Unit Root Tests--Prices and Quantities
in First Difference ...... ............... ..89

4.4 Diagnostic Statistics for Residual
Tests for the Banana ECM .... ............ 91

4.5 Tests of Cointegration Rank for Banana ECM . 95

4.6 Estimated Long-run and Adjustment
Coefficients, P's, a's--Banana ECM .. ....... ..96

4.7 Estimated Long-run and Adjustment
Coefficients, P's, a's for all Crops ........ ..100

4.8 Endogenous and Exogenous Variables in Crops'
Impulse Response Functions .... ........... .109

4.9 Responses of Banana Quantity to Shocks
in the Banana VAR ...... ............... 111

4.10 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance--Banana VAR ..... .............. .114

4.11 Responses of Quantity to Shocks in Exogenous
Variables ....... ................... 117

4.12 Summary of Forecast Error Variance in
Crop Quantities Explained by Variables
in the VAR (Percentage) .... ............ 119

4.13 Responses of Sugar Quantity to Shocks
in the Sugar VAR ...... ................ ..121

4.14 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance--Sugar VAR ..... .............. 121

4.15 Responses of Coffee Quantity to Shocks
in the Coffee VAR ..... ............... ..122

4.16 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance--Coffee VAR ..... .............. ..122

4.17 Responses of Pimento Quantity to Shocks
in the Pimento VAR ...... ............... ..123

vii









4.18 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance--Pimento VAR .... ............. .123

4.19 Responses of Yam Quantity to Shocks
in the Yam VAR ....... ................. ..124

4.20 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance--Yam VAR ..... ............... ..124

4.21 Responses of Orange Quantity to Shocks
in the Orange VAR ..... ......... ..125

4.22 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance--Orange VAR ..... .............. .125

4.23 Responses of Cocoa-bean Quantity to Shocks
in the Cocoa-bean VAR .... ............. .126

4.24 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance--Cocoa-bean VAR ..... ............ .126


4.25 Responses of Potato Quantity to Shocks
in the Potato VAR ...........

4.26 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance--Potato VAR ... ...........

5.1 Descriptive Price and Quantity Statistics
--1962-1979 and 1980-1999 .......

5.2 Descriptive Price and Quantity Statistics
--1975-1979 and 1980-1984 .......

5.3 Diagnostic Statistics for Forecasts of
Fitted Values .............


* 127



* 127


. .. 140


. 141


. 155


5.4 Estimated Long-run and Adjustment
Coefficients, P's, a's for all Crops,
1962-1979 . . . . .

5.5 Estimated Long-run and Adjustment
Coefficients, P's, a's for all Crops,
1980-1999 . . . . .
viii


* 157



159















LIST OF FIGURES


Figure

1.1 Conceptualization of the Jamaican
Agricultural Sector ...........

2.1 GDP and Sectoral Contribution to GDP
in Constant (1995) Dollars ............

2.2 GDP Growth (Percentage Change Over
Previous Year, 1995=100) ... ..........

2.3 Growth of Total Agriculture, Industry
and Services (Percentage Change Over
Previous Year, 1995=100) ... ..........

2.4 Total Agricultural Output Index (1995=100)

2.5 Output Indexes for Broad Agricultural
Aggregates (1995=100) ..........

2.6 Sub-sectors as a Proportion of Agriculture

2.7 Selected Economic Policy Reforms
in Jamaica, 1977-98 ...........

4.1 Residuals for Banana Price in Banana ECM .

4.2 Residuals for Banana Quantity in Banana ECM

4.3 Residuals for Sugar Price in Banana ECM .

4.4 Residuals for Wage in Banana ECM .........

4.5 Cross- and Autocorrelograms in Banana ECM .

4.6 Plot of Eigenvalues for the Banana ECM .

4.7 Impulse Responses to an Innovation in Sugar
Price--Banana VAR ............


page


. 14


. 27


* 29



* 29

. 31


31

* 33


* 43

* 92

* 92

* 93

* 93

* 94

* 94


* 112









4.8 Impulse Responses to an Innovation in
Fertilizer Price--Banana VAR ... ........

4.9 Impulse Responses to an Innovation in Wage
--Banana VAR . . . .

4.10 Impulse Responses to an Innovation in
Banana Price--Banana VAR ... ..........

4.11 Impulse Responses to an Innovation in Banana
Quantity--Banana VAR ..... ...........

4.12 Impulse Responses to an Innovation in Sugar
Quantity--Sugar VAR .... ............

4.13 Impulse Responses to an Innovation in Coffee
Quantity--Coffee VAR .... ............

4.14 Impulse Responses to an Innovation in Pimento
Quantity--Pimento VAR .............

4.15 Impulse Responses to an Innovation in Banana
Price--Sugar VAR ..... ..............


112


. 112


. 112


. 113


. 113


. 113


. 113


. 128


4.16 Impulse Responses to an Innovation in
Fertilizer Price--Sugar VAR .......

4.17 Impulse Responses to an Innovation in Wage
--Sugar VAR ........ ............

4.18 Impulse Responses to an Innovation in
Sugar Price--Sugar VAR .... ...........

4.19 Impulse Responses to an Innovation in Banana
Price--Coffee VAR ............

4.20 Impulse Responses to an Innovation in
Fertilizer Price--Coffee VAR ..........

4.21 Impulse Responses to an Innovation in Wage
--Coffee VAR .............

4.22 Impulse Responses to an Innovation in


Coffee Price--Coffee VAR


* 128


* 128


* 128


* 129


* 129


* 129


. ........ 129


4.23 Impulse Responses to an Innovation in Banana
Price--Pimento VAR ... .............


. 130









4.24 Impulse Responses to an Innovation in
Fertilizer Price--Pimento VAR ... ......... .130

4.25 Impulse Responses to an Innovation in Wage
--Pimento VAR ...... ................. .130

4.26 Impulse Responses to an Innovation in
Pimento Price--Pimento VAR .... ........... .130

4.27 Impulse Responses to an Innovation in Cassava
Price--Yam VAR ....... ................. .131

4.28 Impulse Responses to an Innovation in
Fertilizer Price--Yam VAR .... ........... .131

4.29 Impulse Responses to an Innovation in Wage
--Yam VAR ....... ................... .131

4.30 Impulse Responses to an Innovation in
Yam Price--Yam VAR ...... ............... ..131

4.31 Impulse Responses to an Innovation in Yam
Quantity--Yam VAR ..... ............... ..132

4.32 Impulse Responses to an Innovation in Orange
Quantity--Orange VAR ...... ........... 132

4.33 Impulse Responses to an Innovation in Cocoa-
bean Quantity--Cocoa-bean VAR ... ......... .132

4.34 Impulse Responses to an Innovation in Potato
Quantity--Potato VAR ......... ......... 132

4.35 Impulse Responses to an Innovation in
Grapefruit Price--Orange VAR .... .......... .133

4.36 Impulse Responses to an Innovation in
Fertilizer Price--Orange VAR .... .......... .133

4.37 Impulse Responses to an Innovation in Wage
--Orange VAR ....... ................. 133

4.38 Impulse Responses to an Innovation in
Orange Price--Orange VAR .... ............ .133

4.39 Impulse Responses to an Innovation in Banana
Price--Cocoa-bean VAR ..... ............. 134

xi









4.40 Impulse Responses to an Innovation in
Fertilizer Price--Cocoa-bean VAR ... ........ ..134

4.41 Impulse Responses to an Innovation in Wage
--Cocoa-bean VAR ...... ................ ..134

4.42 Impulse Responses to an Innovation in
Cocoa-bean Price--Cocoa-bean VAR ... ........ ..134

4.43 Impulse Responses to an Innovation in Cassava
Price--Potato VAR ..... ............... ..135

4.44 Impulse Responses to an Innovation in
Fertilizer Price--Potato VAR .... .......... .135

4.45 Impulse Responses to an Innovation in Wage
--Potato VAR ....... .................. .135

4.46 Impulse Responses to an Innovation in
Potato Price--Potato VAR ..... ............ .135

5.1 Nominal Price Changes--Banana (BANP)
and Sugar (SUGP) ...... ................ .143

5.2 Nominal Price Changes--Coffee (COFP)
and Pimento (PIMP) ...... ............... .143

5.3 Nominal Price Changes--Yam (YAMP)
and Orange (ORP) ...... ................ 143

5.4 Nominal Price Changes--Cocoa-bean (COBP)
and Potato (POTP) ..... ............... .143

5.5 Real Price Changes--Banana and Sugar
(1980=100) ....... ................... ..144

5.6 Real Price Changes--Coffee and Pimento
(1980=100) ....... ................... ..144

5.7 Real Price Changes--Yam and Orange
(1980=100) ....... ................... ..144

5.8 Real Price Changes--Cocoa-bean and Potato
(1980=100) ....... ................... ..144

5.9 Banana Quantity--Actual (LBANQ) and
Forecasted (FOLBANQ) ..... .............. .148

xii









5.10 Banana Price--Actual (LBANPR) and
Forecasted (FOLBANPR) ..... ............. 148

5.11 Sugar Price--Actual (LSUGPR) and
Forecasted (FOLSUGPR) ..... ............. 148

5.12 Wage--Actual (LWAGE) and Forecasted
(FOLWAGE) ........ ................... 148

5.13 Fertilizer Price--Actual (LFERP) and
Forecasted (FOLFERP) ..... .............. .149

5.14 Sugar Quantity--Actual (LSUGQ) and
Forecasted (FOLSUGQ) ..... .............. .149

5.15 Coffee Quantity--Actual (LCOFQ) and
Forecasted (FOLCOFQ) ..... .............. .149

5.16 Coffee Price--Actual (LCOFPR) and
Forecasted (FOLCOFPR) ..... ............ 149

5.17 Pimento Quantity--Actual (LPIMQ) and
Forecasted (FOLPIMQ) ..... .............. .150

5.18 Pimento Price--Actual (LPIMPR) and
Forecasted (FOLPIMPR) ..... ............. 150

5.19 Yam Quantity--Actual (LYAMQ) and
Forecasted (FOLYAMQ) ..... .............. .150

5.20 Yam Price--Actual (LYAMPR) and
Forecasted (FOLYAMPR) ..... ............. 150

5.21 Cassava Price--Actual (LCASPR) and
Forecasted (FOLCASPR) ..... ............. 151

5.22 Orange Quantity--Actual (LORQ) and
Forecasted (FOLORQ) ..... .............. 151


5.23 Orange Price--Actual (LORPR) and
Forecasted (FOLORPR) ... ..........

5.24 Grapefruit Price--Actual (LGRPR) and
Forecasted (FOLGRPR) ... ..........

5.25 Cocoa-bean Quantity--Actual (LCOBQ) and
Forecasted (FOLCOBQ) ... ..........

xiii


. . 151


. . 151


. . 152









5.26 Cocoa-bean Price--Actual (LCOBPR) and
Forecasted (FOLCOBPR) ..... ............. 152

5.27 Potato Quantity--Actual (LPOTQ) and
Forecasted (FOLPOTQ) ..... .............. .152

5.28 Potato Price--Actual (LPOTPR) and
Forecasted (FOLPOTPR) ...... .............. .152

5.29 Banana--Fitted Output for Reform Period
(BANFITA) and Counterfactual (BANFITC) ...... .161

5.30 Sugar--Fitted Output for Reform Period
(SUGFITA) and Counterfactual (SUGFITC) ...... ..161

5.31 Coffee--Fitted Output for Reform Period
(COFFITA) and Counterfactual (COFFITC) ...... ..161

5.32 Pimento--Fitted Output for Reform Period
(PIMFITA) and Counterfactual (PIMFITC) ...... .161

5.33 Yam--Fitted Output for Reform Period
(YAMFITA) and Counterfactual (YAMFITC) ...... .162

5.34 Orange--Fitted Output for Reform Period
(ORFITA) and Counterfactual (ORFITC) ........ ..162

5.35 Cocoa-bean--Fitted Output for Reform Period
(COBFITA) and Counterfactual (COBFITC) ...... ..162

5.36 Potato--Fitted Output for Reform Period
(POTFITA) and Counterfactual (POTFITC) ...... .162


xiv














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

ECONOMIC REFORMS AND AGRICULTURAL SUPPLY RESPONSE
IN JAMAICA

By

Ballayram

May 2001

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

A number of economic reform programs have been

undertaken in Jamaica, over the past two decades. Designed

largely by the International Monetary Fund (IMF) and the

World Bank, these reforms focused on correcting internal

policy weaknesses and creating an environment conducive to

sustained growth. The reforms emphasized liberal trade and

exchange rate regimes, a less intrusive and smaller public

sector, and reliance on market forces to determine

agricultural prices and quantities.

Against this background, this study investigates the

impact of these recent economic reforms on agricultural

crop supply responses in Jamaica. The estimation technique

used an error correction modeling framework based on


xv









cointegration theory, within an estimation framework

developed by Johansen. The results of the crop supply

response estimation confirm that there is a long-run

relationship between agricultural crop output and price

incentives. Most of the estimated crop price elasticities

are low, statistically significant, and fall within the

range estimated by other studies on Jamaica. The adjustment

process of the short-run to the long-run was found to be

slow for some crops and higher for others.

Using a counterfactual, which assumed no change in

policy regime, fitted series of supply response functions

from the pre-reform period were forecasted within a

univariate ARMA(p,q) framework and compared to fitted

series of supply responses from the reform period. The

results are mixed. It was found that the impacts of the

economic reforms in Jamaica are crop and time specific.

Mean output was higher in the reform period for four of the

eight crops analyzed. Higher real price shifts were

observed in the reform period for some crops but these

price shifts were also accompanied by higher price-

variability. This suggests that the pro-competitive effects

that were expected to accompany the reforms may have

outweighed the stability impulses of administered prices in

the pre-reform era.


xvi















CHAPTER 1
INTRODUCTION


The past two decades have seen a shift in economic

policies throughout Latin American and Caribbean countries.

Generally, these policy changes have been in response to

debt crises, persistent payments-imbalances, and years of

negative and slow growth. By the 1990s a new orthodoxy on

development thinking could be discerned. This is manifested

in reductions in public sector activities and greater

reliance on private sector initiatives and market forces, a

departure from the policy regimes of the 1960s and 1970s.

In Jamaica, as elsewhere, the new orthodoxy has

influenced economic policies in agriculture. In particular,

policies have shifted away from far-reaching government

interventions and control, to increased reliance on market

signals to allocate resources and form prices. In addition,

economy-wide macro-economic policy changes have been made

which can potentially influence agricultural incentives,

income, and output. Designed largely by the International

Monetary Fund (IMF) and the World Bank in consultation with







2

the Government of Jamaica, these reforms' are expected to

favor Jamaican agriculture under the presumption by the

Bretton Woods institutions that previous policies severely

discriminated against the sector. (This point is elaborated

in Chapter 2, section 2.4). This study pertains to the

impacts that these economic policy reforms have on

agricultural supply responses in Jamaica, as the

agricultural sector emerges from a heavily regulated sector

to one that is more open2 and liberal. This chapter

discusses the problems that this study addresses.

1.1 Background to Jamaica's Economic Reforms

For purposes of this study, the reform period is

defined as the years 1980-1999, while the pre-reform period

is 1962-1979. The year 1962 is sufficiently far back to

facilitate a meaningful analysis of economic policies and

conditions that led up to the dramatic changes in policies

in the reform period. The reform period marks a re-

orientation of economic policy. This change arose from the

stabilization programs of the IMF in Jamaica in the late

1970s. It gathered momentum in the 1980s and 1990s in the



I In this study, "policy reforms," "economic policy reforms," "economic
reforms," and "structural reforms," are used interchangeably.
2 Openness is defined here in terms of trade/GDP ratio. A high
trade/GDP ratio can coexist with high import tariffs, export subsidies,
and non-tariff barriers, signaling a non-liberal trade regime.







3

form of structural adjustment programs cum reforms for

economic liberalization, under the direction of the IMF,

the World Bank and other leading external funding agencies,

notably the Inter-American Development Bank (IDB) and the

United States Agency for International Development (USAID).

In effect, therefore, for the purposes of this study,

economic reforms refer to the policies stipulated in the

stabilization and structural adjustment programs which

Jamaica has been pursuing since the late 1970s.

When the People's National Party (PNP) took office in

1972 under Prime Minister Michael Manley, it was under

pressure to keep its election campaign promises to increase

real wages and government spending, and to reduce social

inequalities. The responses of the new administration to

this situation, coupled with subsequent external and

domestic events, were to plunge the economy into a

sustained recession and major disequilibrium. During the

previous administration under the Jamaica Labor Party

(JLP), the performance of the economy was impressive. GDP

growth rates were positive and relatively high (5.1 percent

per annum over the 1960-1970 period), inflation was less

than five percent per annum and the balance of payments

showed modest surpluses (Jefferson, 1972; IMF, 1998). Over

the 1970-1980 period, however, real GDP declined by -0.9







4

percent. In fact, with the exception of a negligible 0.7

percent GDP growth in 1978, negative growth rates were

recorded for all other six years between 1974 and 1980.

Similarly, in the productive sectors, agricultural value-

added growth declined from 6.7 percent in 1970-1972 to -0.4

percent in 1973-1980. For these same periods, declines in

growth rates were recorded for services from 9.2 percent to

-0.08 percent, manufacturing from 6.7 percent to -4.2

percent, construction from 6.0 to -10.6 percent, and mining

from 13.2 percent to 1.3 percent, respectively (Singh,

1995; IMF, 1998).

The external account reflected the production crisis

described above. Between 1972 and 1980, the accumulated

deficit on the balance of payments was US$679.2 million,

compared to an accumulated surplus of US$95 million between

1960 and 1971 (Jamaica, Bank of Jamaica, 1985). Further,

the deficit on the current account increased almost three-

fold from US$570.2 million in 1962-1970 to US$1,620.0

million in 1972-1980, and the capital account became

negative in 1977, signaling capital flight. At the same

time, the dwindling net capital inflows could not match the

current account deficit. This depleted international

reserves, which plummeted sharply from US$103.1 million in

1972 to minus US$461 million in 1980 (Sharpley, 1984). The







5

foreign exchange shortage harmed the economy given its

relative openness and dependence on foreign intermediate

imports.

Finally, the inflation rate, which was 4.2 percent per

annum over the 1960-1970 period, recorded double-digit

rates throughout the following decade. Domestic prices, as

measured by the consumer price index, rose sharply,

recording an annual increase of 25 percent and 47 percent

over the 1974-1975 and 1978-1979 periods respectively

(Sharpley, 1984). By the late 1970s economic conditions had

deteriorated to such an extent that, when combined with

manifest signs of social distress (such as a dramatic

increase in violent crimes and emigration of skilled

workers), they generated a sense of instability and crises

(Stephens and Stephens, 1986; Kaufman, 1985).

The crises indicated a need for policy changes, and it

appears that the government was thinking along those lines

as reflected in a Stand-by Agreement it negotiated with the

IMF just before the December 1976 elections. The Agreement

included, inter alia, a wage freeze, fiscal restraint and a

currency devaluation of 20 to 40 percent (Boyd, 1988).

However, in the December 1976 election the PNP was returned

to office and the new administration promptly rejected the

IMF Stand-by Agreement as being inconsistent with the







6

mandate obtained in the recent election. Instead, and as a

way of avoiding agreements with the IMF (Boyd, 1988), and

after significant slippages in the external and domestic

accounts (Sharpley, 1984), the government implemented its

own adjustment policy measures in January 1977. These

included (Boyd, 1988)

(1) The reversal of the previous wage indexation policy;

(2) Implementation of import and exchange controls, a

dual exchange rate system and foreign exchange

rationing;

(3) Suspension of foreign debts for 18 months; and,

(4) A search for funding from sympathetic governments.

However, it became increasingly clear to the

government that in order to attract foreign capital some

formal agreement with the IMF was necessary. In this

regard, the government signed three agreements with the IMF

between August 1977 and June 1979. The first was suspended

after four months for failure to meet the stipulated

quarterly tests. The second, signed in May 1978, was

pursued in all its aspects by the government for a year.

Although economic performance under this agreement was poor

(Sharpley, 1984; Boyd, 1988), it was re-negotiated in June

1979. However, the agreement collapsed in December 1979







7

after the government failed to meet the program's

performance criteria.

In 1980, in light of increasing social unrest,

continued slippages in the fiscal deficits and external

accounts, severe foreign exchange shortages, and rising

inflation (Sharpley, 1984; Boyd, 1988; Thomas, 1999), Prime

Minister Manley called a general election so that the

nation could decide on a path of economic development and

"what part the IMF should play; or whether it should play

any part at all" (Jamaica, API, Manley, 1980). Dubbed "the

IMF election", the October 1980 election was won by the

strongly pro-IMF JLP, led by Edward Seaga. Expectedly,

therefore, both the IMF and the government began

negotiations on funding and policies aimed at ameliorating

the country's economic and social crises.

The new administration relied heavily on funding from

external sources. Several loan agreements were negotiated

with the leading collaborating lending institutions, the

IMF, World Bank, IDB and USAID which have developed a

system of cross-conditionalities on their loans, given

their basic set of shared objectives on economic policies

and on Jamaica (Anderson and Witter, 1994).

The prescription by the IMF and the World Bank

(Fund/Bank) on how best to deal with the economic crisis in







8

Jamaica was a series of economic policy reforms. These

reforms went beyond the expenditure-switching and

expenditure-reducing policies which macroeconomic policy

would normally prescribe under these circumstances, to

include a built-in bias in favor of a more liberal economic

system. The latter aimed at, inter alia, fiscal discipline,

liberalizing the domestic market and the external trade

sector, privatizing of State Owned Enterprises (SOEs) and

other social services, and generally, greater reliance on

market signals to allocate resources. Over all, between

1981 and 1992, IMF financial support totaled US$1,036.9

million (under 10 agreements), and that of the World Bank

US$360.4 million (under 8 agreements).

1.2 Problem Statement and Study Justification

The cumulative effect of the reforms mentioned above

is that the Jamaican agricultural sector is now expected to

operate within an economic framework that differs vastly

from that of the 1960s and 1970s. Historically, the sector

has played an important role in the economy. Agricultural

output to GDP ratio is about eight percent, agricultural

food export in total export is approximately 20 percent,

and agriculture's share of the total labor force exceeds 26

percent (Davis et al., 1999). Given the importance of

agriculture in the economy, how the sector performs in this







9

new policy environment demands urgent attention. In this

study, agricultural sector performance is evaluated in

terms of crop supply responses. Quantitative analyses of

the impact of these reforms on Jamaican agricultural supply

responses are both scarce and generally not particularly

robust. In addition, the literature on similar economic

reforms in other countries suggests that the outcomes of

these reforms are ambiguous, which justifies more empirical

work (Khan, 1990).

Additionally, this study is justified on the following

grounds. First, the economic reforms have generated major

changes in the long-term prospects for the national economy

within which agriculture will operate. After two decades a

good set of data on these policy reforms is now available

which should facilitate empirical research on the impact of

these reforms on agricultural crop supply responses. This

evaluation can throw light on courses of action which

policy makers in Jamaica and elsewhere can consider as they

seek to reactivate and sustain long-term growth in the

agricultural sector.

Second, while a few studies have analyzed specific

aspects of the impact of the economic policy reforms on

Jamaican agriculture (Singh, 1995; Anderson and Witter,

1994; Newman and Le Franc, 1994; and Brown, 1994), what is







10

lacking is an over-all analytically rigorous assessment of

agricultural supply responses within the context of the new

policy framework as defined by the recent economic reforms.

Finally, analysis of the impact of these reforms on

agricultural supply responses is both timely and urgent in

light of an on-going debate. A number of writers in the

Caribbean and elsewhere have been advancing the view that

the kinds of economic reforms implemented in Jamaica and in

other countries have harmed agriculture (Anderson and

Witter, 1994; Green, 1989; Bates, 1989). In contrast a

recent series of studies sponsored by the World Bank

suggests that policy reforms similar to Jamaica's should

boost agriculture in countries which have traditionally

discriminated against the sector (Schiff and Vald6s, 1992a;

Krueger, 1992).

1.3 Objectives of the study

The general objective of this research is to evaluate

the impact of economic reforms on selected agricultural

crop supply responses in Jamaica over the 1980-1999 period.

The specific objectives are to provide

(1) A statistical evaluation of the performance of the

agricultural sector in the pre-reform (1962-1979)

and reform (1980-1999) periods.







11

(2) A descriptive, diagnostic and statistical analysis

of the economic reforms that have taken place over

the past two decades within the agricultural sector.

(3) An econometric (modeling) evaluation of selected

agricultural crop supply responses over the 1962-

1999 period, with the number and types of crops

chosen for study to be determined by data

availability.

(4) A comparative analysis of agricultural crop supply

responses in the reform period with a counter-

factual, the latter defined as crop supply responses

that would have resulted in the absence of the

reforms.

1.4 Hypotheses

The following hypotheses are advanced in the study:

(1) Structural reforms significantly improve agri-

cultural supply responses.

(2) Sustained improvements in agricultural supply

responses require further broadening and deepening

of the reform process.

1.5 Conceptual Issues

Agriculture features prominently in policy discussions

when (1) it is believed that agricultural policies and

institutions help precipitate an economic problem which







12

policy makers are addressing; and (2) when agriculture is

an important contributor to employment, GDP, export

earnings, domestic food supply, and revenue for the

government (Binswanger and Deininger, 1997). Various

publications by the government of Jamaica suggest that both

of these factors have featured prominently in the recent

reforms, while (2) was a major consideration in the pre-

reform (1962-1979) period. Consequently, the conceptual-

ization is based on two perspectives that are believed to

have influenced the design of the policy reforms:

(1) The Jamaican government perspective. From numerous

government publications, it would appear that this

perspective is clustered around three main goals:

(a) increasing food supply; (b) developing rural

areas; and (c) increasing agriculture's contribution

to the over-all economy, through, inter alia,

employment, contribution to GDP, and export earn-

ings; and,

(2) The IMF/World Bank perspective, whose focus on the

agricultural sector clusters around (a) getting

prices right; (b) improving efficiency; and (c)

rationalizing public expenditure in the agricultural

sector.







13

These two perspectives are not mutually exclusive. On

the contrary, the choice of policies in a particular

program package results from extensive discussions between

officials of the IMF/World Bank and the Jamaican

government. The program policy mix therefore reflects the

particular economic situation in the country as well as the

preferences of the government (Khan, 1990). Despite this, a

review of the literature on Jamaican agricultural policy

formation suggests that since the late 1970s the influence

of the IMF/World Bank perspective has exceeded the Jamaican

government's.

Given the two perspectives mentioned above, and a

review of the content of the reforms since the late 1970s,

the evidence suggests that the reforms have focused on

three inter-related issues of relevance to the agricultural

sector: (i) agricultural (inter-sectoral) terms of trade;

(ii) agricultural growth, its adjustment, and supply

response; and (iii) efficiency in the agricultural sector

(Nallari, 1992; Singh, 1995; World Bank, 1996).

Figure 1.1 conceptualizes the Jamaican agricultural

sector, dividing the factors impacting agriculture into

external (e.g., world prices) and internal. The latter can

be separated into those that are exogenous (weather,

terrain) and those that are policy induced. Another useful












































Policy Instruets:
P1. Agriculture Projects:
infrastructure, land
development, mechanization,
crop improvement
P2. Rural industries
P3. Monetary Policies: interest
rates, exchange rates.
P4. Tax Policies: income,
indirect, customs duties.
P5. Public consumption, rural
invest.
P6. Rural health, education,
welfare.
P7. Import/Export controls,
tariffs, subsidies.
P8. Price controls, input
subsidies.
P9. Agriculture credit
PlO.Marketing: infrastructure,
information.


Agriculture Sector
Performance Indicators:
1. Economic Outcomes
(a) quantity flows
supply response
agri. in GDP
agri. growth rate
(b) agri. multi-factor
productivity
(c) relative prices
(TOT, RER).
2. Accumulation
Indicators
coefficient of
investment
capital formation
agri. employment
agri. tax revenues
rural nutrition
farm incomes


Figure 1.1:
Sector.


Conceptualization of the Jamaican Agricultural


T I I







15

distinction is between prices and non-price factors

(including exogenous shocks) that influence agricultural

output and income. At the micro level, the price variables

include prices of the particular crops, prices of

alternative crops, input prices, and the general price

level. Government policies exercise both direct and

indirect influence on agricultural prices and thus on

agricultural performance. Direct (sectoral) policies such

as price ceilings, guaranteed prices and trade

taxes/subsidies provide incentives to shift resources among

crops and sectors of the economy (Binswanger, 1989).

Government policies also can influence output indirectly

through macroeconomic (or economy-wide) policies. The most

critical components in this regard are fiscal policies and

exchange rate policies (Mamingi, 1997).

The terms of trade as a policy variable turns on the

hypothesis that if all prices were determined in markets,

and effective rates of taxation were equalized across

commodities, then agriculture's supply responses, growth,

income, and contribution to GDP would be higher compared to

situations of negative effective protection of the sector

(Schiff and Vald~s, 1992b). A negative effective protection

of agriculture arises when:









(1.1)i Pf)+ 5 C,] < o
P

where P is a general price deflator, Pi denotes the producer

price of crop i, Pf is the free-market price for crop i,

and 8 is a subsidy to producers as a proportion of cost, Ci.

For many developing countries, agricultural pricing

policies have consistently kept Pi below P and subsidies

have not sufficiently counterbalanced this disparity

(Schiff and Vald6s, 1992b). Given severe budgetary

constraints, to achieve at least zero effective protection,

the policy implication of (1.1) is to raise the real


P


eliminating the disparity between Pi and P. In effect,

improving agriculture's terms of trade.

The exchange rate sets an upper limit on agricultural

export earnings and, when combined with input taxes and

subsidies, affects input prices and competing agricultural

imports. Changes in the real exchange rate can affect

agricultural output and growth by altering the terms of

trade between agriculture and non-agricultural sectors.

Further, policies that lead to over-valuation of the

exchange rate can adversely affect export crops, encourage







17

rent-seeking activities and generate unproductive uses of

resources (Jaeger, 1991).

In addition to the price policies, non-price factors

constitute an important influence on agricultural output.

These factors reflect the material conditions of production

in the agricultural sector. Some of these factors include,

inter alia, government expenditure and investment in the

agricultural sector, construction of rural infrastructure

(roads, drainage and irrigation works), extension services,

rural credit institutions, and dissemination of relevant

scientific information to farmers. The aim of agricultural

sector-specific programs is to find a mix of policies for

increasing efficiency and productivity in the sector. It is

assumed that these two outcomes are necessary (if not

sufficient) conditions for increasing agricultural output

and growth by reducing costs of production and/or

increasing product prices. These policies focus on using

public investment in agriculture efficiently, reducing

marketing and transport bottlenecks, improving agricultural

extension services, health and education (capital formation

in agriculture), rationalizing input prices, and enhancing

the efficiency of parastatals in the agriculture sector.

Behrman (1990) has advanced an important perspective

on the channels through which policies affect performance







18

indicators. He argues that analysis of the impact of policy

reforms on performance indicators must explicitly consider

the conduits through which the effects of the policies are

transmitted to the observed (or desired) outcome. These

"meso-level" variables, identified as markets (product,

input and financial) and infrastructure (economic and

social), are the interface between the policies (sector-

specific and economy-wide) and targets. In Figure 1.1, for

example, various policies, Pi, i = 1,2,3,...10, are

identified. Some are macroeconomic (e.g., monetary policies

and exchange rate adjustments), whereas others are sector

specific (e.g., agriculture credit, tariff reductions in

the trade sector and elimination of price controls in

agriculture). These policies are then combined to achieve

specific targets, either at the sectoral or macroeconomic

levels.

Behrman (1990) emphasizes, however, that the meso-

level variables condition the effectiveness of policies on

the target variables and that analyses of policy impacts

must also evaluate how policies have influenced these meso-

level variables. For example, if poor transportation, lack

of effective irrigation, or in-efficient research and

extension services restrain farmers' responses to higher







19

prices, then improving these meso-level variables may do

more for the farmer than price increases (Chhibber, 1989).

The influence of the factors mentioned above on

agricultural supply responses has been well documented in

the empirical literature. However, the recent economic

reforms have had an enduring effect on these factors. This

points to an important issue raised in the literature,

viz., how to measure the magnitude of reforms (IDB, 1997).

Economic statistics such as exchange rate differentials,

inflation rates, tax changes and so on deal with outcomes

rather than the policies that gave rise to them. In order

to address this issue, structural policy indexes have been

constructed by Lora (1997) and others.

The index by Lora (1997) was constructed for twenty

countries in Latin America and the Caribbean. This index

seeks to measure the extent of market freedom accorded to

economic policies in areas of trade, tax, finance,

privatization and labor. In each of these areas indices of

market freedom are identified. For example, in trade policy

the indices are average tariffs and tariff spreads; in tax

policy the indices include, inter alia, tax rates on

companies and on individuals; and on financial policy,

indices include freedom of interest rates on deposits,

loans, reserves on bank deposits, etc. (IDB, 1997). The







20
structural policy index is a simple average of the indices

in the five areas. The index can range from 0 to 1, based

on the worst and best observations respectively, on market

freedom in the country. Further, the index is an important

indicator of the extent to which countries are departing

from past ways of operating their economies. Jamaica's

structural policy index has shown continuous movement

towards market freedom, increasing from 0.426 in 1985 to

0.684 in 1995. These index values are higher than the

average reported for the twenty Latin American and

Caribbean countries in the sample (Lora, 1997).

1.6 Procedural and Methodological Considerations

In approaching the general and specific objectives of

the study, critical analysis of the Jamaican agricultural

sector's performance is first undertaken. To achieve the

general objectives and, more specifically, objectives (2),

(3) and (4), requires four specific tasks. The first task

is to identify and evaluate the structural reforms

undertaken since the late 1970s. Both the stabilization and

the structural reform policies are evaluated according to

their (1) theoretical bases; (2) breath of vision; and (3)

logical consistency. Second, supply response models are

developed and estimated. Finally, specific objective (4) is







21

addressed within the context of simulating an alternative

path to that of the reforms.

Any appraisal of the reforms must compare their

impacts not to the pre-reform indicators but rather to some

specified, hypothetical alternative. This requires some

simulation exercises. The idea here is to generate

simulated time series over the reform period for the

performance indicators. The simulated series then

constitute the counterfactual to which the actual outcomes

in the reform period are compared. For the counterfactuals,

at least three scenarios appear to be logical extensions.

Scenario 1 assumes that the policies pursued over the pre-

reform period continued into the reform period, i.e., that

the reforms over the 1980-1999 period were not instituted.

Scenario 2 assumes policies based on the Jamaican

government's critique of the IMF/World Bank programs were

implemented. Finally, Scenario 3 assumes that IMF/World

Bank assistance came with no conditionalities. These

components form the basis of the data series for the

simulation exercise.

The three scenarios appear logical and plausible

options given (1) the initial defiance by the government of

the mandated reforms in the late 1970s; (2) the frequent

failures to meet conditionality tests because the reforms







22

were not implemented; and (3) government's own protest-

ations over the years of what are considered "acceptable"

reforms. While numerous instances of disagreements exist

between the Jamaican government and the IMF/World bank with

respect to appropriate reform policies, the clearest

statement to this effect is captured in the following

quotation:

Even though lender and borrower therefore shared the
common concern regarding the need for economic reform,
there is still much debate about appropriate social
policies and about the specific effects of
manipulating major economic variables such as exchange
rates and interest rates. This was illustrated by the
lengthy and tortuous negotiations between the
government of Jamaica and the IDB and World Bank on
the 1989-90 Agricultural Sector Adjustment Loan.
(Anderson and Witter, 1994, p.14)


1.7 Summary

It is generally believed that developing countries

have historically discriminated against their agricultural

sectors in favor of industrial development. It is further

felt that policies that reverse that discrimination should

boost agricultural output and income. Over the past two

decades far-reaching economic reforms have been undertaken

in Jamaica leading to a new economic framework within which

the agricultural sector must now operate. This study seeks

to assess the impact of these reforms on agriculture crop

supply responses, an undertaking that is both timely and







23

urgent given the importance of the sector in the economy

and the absence of rigorous analytical studies on the

Jamaican experience with these reforms.















CHAPTER 2
ECONOMIC STRUCTURE, GROWTH, AND POLICY REFORMS IN THE
JAMAICAN ECONOMY, 1962-1999


It is useful, at the outset, to situate the discussions

in this chapter against the backdrop of three distinct

phases of economic policy that characterize the economic

history of Jamaica over the period 1962-1999. In the first

phase, 1962-1972, the Jamaica Labor Party (JLP) emphasized

free markets. However, Bonnick (1984) argues that despite

the rhetoric of economic liberalism, in reality the

government pursued an import-substitution (IS) strategy

based on protectionism, trade restrictions, and price

controls.

In the second phase, 1972-1980, the People's National

Party (PNP) under Prime Minister Michael Manley pursued

economic populism and state directed control (dirigisme) in

an effort to build democratic socialism. This phase was

marked by extensive government intervention, which included

nationalization of major industries, price controls, and

subsidization of basic foods and some agricultural imported

inputs. These policies made significant demands on national

resources and created a large bureaucratic economic

24







25

structure. Consequently, the government was unable to

respond to the crisis in the world economy in the mid-

1970s, and was forced to approach the IMF for stabilization

funding in 1977.

The third phase, 1980-present, is directly linked to

the policies of the previous phases, particularly the 1977

IMF stabilization funding. Phase three can be viewed as

having two sub-phases. The first is the return of the JLP

to government in 1980-1989 and the implementation of

various Fund/Bank stabilization and structural adjustment

programs. The second, 1989-present, marks the return to

government by the PNP, the continuation of the market led

policies of the preceding JLP administration, but more

importantly, the intensification of the country's commit-

ment to liberalization of the economy. The rest of this

chapter is organized as follows: Section 2.1 provides an

overview of the structure of the economy and its growth

performance. Section 2.2 highlights trends and policies in

the agriculture sector. Sector 2.3 describes the economic

reforms, and Section 2.4 provides a summary of issues

raised in this chapter.









2.1 Structure of the Economy and Economic Growth:
An Overview

The relative contribution of economic sectors to GDP in

Jamaica has not changed much over the past two decades. The

data presented in Table 2.1 and Figure 2.1 show that

agriculture's share has been fairly stable, ranging from an

average of 8.4 percent over the period 1969-1979 to 8.0

percent over 1990-1998. Although agricultural contribution

to GDP is lower than that of the other sectors, these

comparatively small percentage contributions are deceptive.

Davis et al. (1999) have shown that the sector plays an

important part in the country's employment, food

production, and foreign exchange earnings.

The industrial sector, which includes mining,

manufacturing and construction, contributed on average 39.1

percent to GDP in the period 1969-1979 but declined to 37.3

percent in 1990-1998. The manufacturing sub-sector also

shows a decline from 18.4 percent in 1969-1979 to 16.3

percent in 1990-1998. Finally, services constitute the

largest sector of the Jamaican economy, accounting for over

50 percent of GDP over the period 1969-1998. Its

contribution to real GDP has increased from 48.5 percent in

1969-1979 to 54.7 percent in 1990-1998.









Table 2.1: Sectoral Contribution to Real Gross Domestic
Product (Period Averages, Percentage).


1969-79 1980-89 1990-98 1969-98
Agriculture 8.4 6.4 8.0 7.6
Industry 39.1 34.5 37.3 37.0
Manufacturing 18.4 20.9 16.3 17.1
Services 48.5 59.1 54.7 53.9


Source: Computed using data from Jamaica, Pannlng
Institute of Jamaica (various issues).
Note: The 1998 data used were preliminary estimates.


1973 1977 1981 1985
Years 1969-1998


1989 1993 1997


--Real GDP ---Agri. -Indus. -*-Serv.

Figure 2.1: GDP and Sectoral Contribution to GDP in
Constant 1995 Dollars.



Real sectoral and GDP growth rates are estimated and

shown in Table 2.2. The estimation uses a log-linear model,

Yt = ao0 + P*Time + Et and corrected for autocorrelation

whenever it exists. Over the 1969-1998 period, growth rates

were positive but consistently low, while sub-period growth

reveal mixed results. GDP growth rates were just over two


2000

1600

1200

800

400

0
19


69









Table 2.2: GDP and Sectoral Growth (1995=100)
(Percentage).
1969-79 1980-89 1990-98 1969-98
GDP 2.3a 2.4b -0.2 I. 'M
Agriculture -0.3 1.1 2.3b 0.6b
Industry 0.9 4_._3_ 3 -3.8a 0.8a
Services 4.6a 0.1 2.4b 1.6a
Source: Computed using data from Jamaica, Planning
Institute of Jamaica (various issues).
Note: The 1998 data used were preliminary estimates.
aI b indicate statistical significance at the five
and 10 percent levels, respectively.


percent in the 1969-1979 and 1980-1989 periods. Services

growth was reduced from 4.6 percent over 1969-1979 to 2.4

percent over 1990-1998, and was insignificant over the

1980-1989 period. Industrial growth, which was 4.3 percent

in the 1980-1989 period, declined significantly to -3.8

percent over the 1990-1998 period. Agriculture growth over

the 1990-1998 period was 2.3 percent, compared to its

insignificant growth in the two previous sub-periods. Over

the entire 1969-1998 period agriculture and industry

recorded less than one percent growth, while GDP and

services grew just over one percent, respectively.

These periodic growth rates mask the highly volatile

annual growth rates in these economic series as shown in

Figures 2.2 and 2.3. Over the period 1969-1998, annual

growth rates for GDP ranged between -9.0 and 13.0 percent,

and were negative in 12 of the 30 years. For the











15

10

tit 5

e 0
0
54 -5

10

-15
1970 1974 1978 1982 1986 1990 1994 1998
Years 1969-98

---GDP Growth

Figure 2.2: GDP Growth (Percentage Change Over Previous
Year, 1995-100).


40
30 A
20 A

4 10
00
54 -10
0 -20

-30
-40

1970 1974 1978 1982 1986 1990 1994 1998
Years 1969-98
----Agri. Growth mInd. Growth ----Ser. Growth

Figure 2.3: Growth of Total Agriculture, Industry and
Services (Percentage Change over Previous Year, 1995=100).




1969-1998 period, agriculture growth rates ranged between

-12.6 and 29.3 percent, and were negative in 15 years.







30

Similar annual volatilities characterize the industrial and

services growth rate series.

2.2 Output Trends and Policies in
Jamaican Agriculture

The growth rates of volume indexes for total

agriculture and broad aggregates (food, crops, livestock

and cereals) and graphs for these series are shown in Table

2.3 and Figures 2.4 and 2.5. With the exception of the

cereals index, which declined over 1971-1998 at an annual

rate of -3.2 percent, the other indexes grew but at less

than two percent over the 1971-1998 period.



Table 2.3: Growth Rates of Total Agricultural and
Broad Agricultural Aggregate Indexes.
1971-79 1980-89 1990-98 1971-98
Total Agriculture 1.6a 1.7 1.9a 1.7a
Food 1.6a 1.6 1. 9a 1.7a
Crops 2.6lb 1.5a -3.6 1.7a
Livestock 1.6a -16.5 0.7 1.3a
Cereals 16.9a -9.0 -4.5- -3.2]
Source: Computed using data from Food and Agriculture
Organization, FAO Production Yearbook (various issues).
aI b, indicate statistical significance at five and ten
percent levels, respectively.



The composition of agricultural output is shown in

Table 2.4. Export agriculture, domestic agriculture, and

livestock, forestry and fishing constitute the aggregative

components of agriculture in GDP. The data in Table 2.5 and

the graph in Figure 2.6 show that domestic agriculture











o 120

90
I i


60
60

30



0 1971 1976 1981 1986 1991 1996
0
Years 1971-98

-*-Total Agriculture

Figure 2.4: Total Agricultural Output Index (1995=100).


Figure 2.5: Output
Aggregates (1995=100).


Indexes


for Broad Agricultural


250
0
200

150

100

H 50
4J

0
o 1971 1976 1981 1986 1991 1996
Years 1971-1998

-Crops a.Livestock -nCereal --Food









Table 2.4: Composition of Agricultural Output
1995=100). Selected Years.


Source: uamaica, Pannlng
issues).


instLIUte 01 Jamaica


Table 2.5: Agriculture Sub-sectors as a Percentage of Total
Agriculture (Period Averaqes).


Source: Computed using data from Jamaica,
Institute of Jamaica (various issues).
Note: Figures in parentheses are standard deviation.


Planning


(J$M,


kVdiOUs


Agriculture Sub-sectors 1969-79 1980-89 1990-98 1969-98
23.47 15.95 11.21 7.29
Export Agriculture (5.50) (2.24) (1.46) (6.26)
44.78 57.21 68.79 56.13
Domestic Agriculture (5.32) (6.35) (3.62) (11.17)
Livestock, Forestry & 31.73 26.85 20.03 26.59
Fishing (1.35) (4.49) (2.50) (5.65)


Agriculture
and Sub-
sectors 1969 1975 1980 1985 1990 1995 1998
Total 95.9 105.9 101.3 72.2 104.4 154.4 120.4
Agriculture
Export 34.3 25.9 14.8 13.9 13.6 15.4 12.3
Agriculture
Sugar 22.6 17.7 9.9 7.8 9.2 9.4 7.9
Cane
Other 11.6 8.2 4.8 6.1 4.4 5.9 4.4
Main
Exports
Domestic 33.0 47.1 56.9 38.0 66.2 114.5 82.8
Agriculture
Root 14.4 26.9 25.3 14.6 35.1 62.8 42.7
Crops
Other 18.6 20.2 31.7 23.4 31.1 51.7 40.1
Primary
Products
Livestock, 28.6 32.9 29.5 20.3 24.6 24.5 25.3
Forestry &
Fishing






















0 1 I- i . 1 -
1969 1973 1977 1981 1985 1989 1993 1997
Years 1969-98
s-Export Agriculture
-n-Domestic Agriculture
---Livestock, Fishing & Forestry

Figure 2.6: Sub-sectors as a Proportion of Agriculture.



constitutes the largest proportion of total agriculture in

the 1969-1998 period and other sub-periods. In 1969-1979

the domestic agriculture sub-sector averaged 44.8 percent

in total agriculture and increased to 68.8 percent in 1990-

1998. Both of the other two sub-sectors, export agriculture

and livestock, forestry and fishing, have declined over the

period 1990-1998 compared to the two previous decades. In

terms of growth rates, Table 2.6 shows that domestic

agriculture and its sub-groups are the only sub-sectors

within the agri-sub-sector that have positive growth rates

over the 1969-1998 period. This has compensated somewhat

for the negative and low growth in the other agriculture

sub-sectors.


80


60


S40


20







34

Table 2.6: Growth Rates of Agricultural Components
(Percentage).
Agriculture Sub-sectors 1969-79 1980-89 1990-98 1969-98
Export Agriculture -4.4a -I. 3D -T. 1 -2. a
Sugar -4 2a -2.6a -5.4a -2.0a
Other main exports -4.6a 1.3 1.0 -2.7a
Domestic Agriculture 1.8 11.2a 25.6a 2.6a
Root crops -2.7 19.7a -1.2b 2.6b
Other domestic crops 4.0a 3.4b -145 2.7a
Livestock, forestry, & -1.5 -3.3a 0.1 -1.7a
fishing II I
Source: Computed using data from Jamaica, Planning
Institute of Jamaica (various issues).
a, b, indicate statistical significance at the five and ten
percent levels, respectively.



The stagnation of agriculture in Jamaica, which can be

inferred from the growth rates in Table 2.6 reflects, to a

large degree, the state of the agricultural sectors in many

less developing countries (LDCs). A plausible explanation

for this, is that the development literature in the 1950s

and 1960s viewed agriculture as a static sector, from which

resources could be shifted to promote industry, considered

as the dynamic sector. To a large extent, this view has

influenced the kinds of agricultural policies that have

been pursued in the past by LDCs. In this regard, Schiff

and Vald~s (1992a, p.59) state:

In many developing countries, the high rate of
agricultural taxation has been part of an explicit or
implicit policy of industrialization-led growth,
justified in part by the belief that industry was the
dynamic sector while agriculture was static and not
very responsive to incentives. That means that
economic growth could be accelerated by shifting







35

resources from agriculture to industry, by taxing
agriculture directly, and by protecting the industrial
sector.

Many governments in the developing countries have

intervened in their agricultural sectors, both directly

through agricultural sector policies, and indirectly,

through economy-wide policies such as industrial

protection. Direct interventions take numerous forms. Some

of these include procurement measures (e.g., government

marketing boards as sole buyers of agricultural output, and

suppliers of major agricultural input); quotas and direct

taxation on various agricultural export crops; subsidies on

farm credit and farm inputs; and quantitative restrictions

and tariffs on imported agricultural imports. While some of

the direct interventions have benefited agricultural

producers, some are tantamount to an implicit tax on

agriculture, depressing farmgate prices and farm incomes

below levels that would otherwise prevail (Schiff and

Vald6s, 1992a). Indirect forms of interventions affect

agricultural production incentives via macroeconomic

policies (e.g., overvaluation of the exchange rate) and

industrial protection policies (Krueger, 1992).

Various measures have been employed by the Jamaican

government to extract surpluses from the agricultural

sector. These include taxing agricultural exports,







36

controlling farmgate prices by the state marketing boards,

over valuating exchange rates and reducing internal

agricultural terms of trade relative to manufacturing.

Studies by Gafar (1980, 1997) and Pollard and Graham (1985)

strongly support this proposition. Table 2.7 shows that



Table 2.7: Average Rates of Growth of Farmgate and F.O.B.
Prices, Output and the Nominal Protection Coefficient (NPC)
for the Period 1970-1978, Jamaica.
Growth Rates (%)1970-78 a
Farmgate F.O.B. Nominal Protection
Commodities Prices Prices Output Coefficient b
Sugar Cane -1.25 4.13 -2.90 0.82
Banana 5.06 7.77 -7.43 0.77
Cocoa 1.64 14.06 -0.54 0.72
Coffee 9.80 8.34 2.02 0.61
Coconut 7.06 4.67 -21.00 n.a.
a Adapted from Pollard and Graham (1985), Table 2.
b Based on the data in Pollard and Graham (1985), Table 4
for 1970-1979. The NPC is defined as the ratio of farmgate
prices (PF) to F.O.B. prices (Pw) received minus marketing
and processing costs (C): NPC = PF / (Pw C).



over the 1970-1978 period, the nominal protection

coefficient (NPC), for Jamaican agriculture was less than

one, indicating the extent to which the sector was taxed.

For example, a NPC of 0.82 means that the commodity is

taxed at a rate of 18 percent. Subtracting one from the NPC

gives the nominal protection rate, NPR, which, according to

Table 2.7, is negative, suggesting that producers of the

crops reported were not supported, but were taxed instead.







37

It should be emphasized also, that developing countries

have a tendency to overvalue their domestic currencies, so

that the official exchange rates used to calculate the NPC

overstate internal prices. Hence government taxation of

agriculture is even greater than is usually captured by the

NPC (Gardner, 1987).

More recent data compiled for this study also support

the proposition that Jamaican agriculture has been taxed.

Table 2.8 reports on the net barter and income terms of

trade between agriculture and manufacturing for Jamaica.

The net barter terms of trade, N, is defined here as


N = L, where PA and PM are the price indexes of agriculture


and manufacturing, respectively. Values of N greater than

one indicate that prices of agricultural commodities have

risen relative to those in manufacturing (Tsakok, 1990).

The estimates in Table 2.8 reveal that over the 1966-1978

and 1989-1999 periods, the net barter terms of trade were

unfavorable to Jamaican agriculture, but were favorable

over the decade 1979-1988.

Although the net barter terms of trade may move

against agriculture, the sector can still increase its

purchasing power if agricultural output increases pro-

portionately more than the decrease in agricultural prices.







38

The purchasing power of agriculture is captured in the net


income terms of trade, I, estimated as: I = PA x QAI where QA
PM

is the agricultural output index. The index I is therefore

the net barter terms of trade adjusted for marketed

amounts. Increases in I indicate a rising purchasing power

of agriculture for buying manufactured goods (Tsakok,

1990). The data in Table 2.8 suggest that agriculture's

ability to purchase manufactured goods have peaked in the

1979-1988 period.



Table 2.8: Not Barter Terms of Trade and Agricultural
Income Terms of Trade (1980=100; Period Averages).
1966-78 1979-88 1989-98
Net Barter Terms of Trade 0.60 1.39 0.86
Income Terms of Trade 58.20 159.18 134.03
Source: Compiled by author.



2.3 The Economic Reforms

After the late 1970s, Jamaica undertook major changes

in economic policies, both sector-specific and economy-

wide. These policy changes have been an integral part of

the conditionalities usually attached to program packages

by external funding agencies. The reforms vary in

intensity, both inter-temporally and in terms of the areas

in which policy reforms were enacted, such as foreign

trade, taxation, and liberalization measures. Underlying







39

the reforms has been a conviction by funding agencies that

freer markets allow for a more productive and efficient use

of scarce resources, a condition not necessarily accepted

uncritically by the Jamaican government. Nevertheless, a

number of market restrictions were eliminated, regulations

deemed necessary were simplified and made more transparent,

and private enterprises were encouraged. The most

conspicuous reforms were the liberalization efforts to

facilitate foreign trade and financing activities through

exchange rate and tariff adjustments.

The reform period, 1980-1999, has been punctuated by

successive IMF/World Bank stabilization and structural

adjustment programs, as well as programs financed by other

international funding agencies. The stabilization and

structural adjustment programs of the IMF and the World

Bank for Jamaica were developed as programs of policy

reforms, aimed at reversing, within a few years, the

economic crisis in the country. Three facets of this crisis

emphasized in the IMF/World Bank literature are structural

factors, external factors and poor domestic policies,

including an anti-agricultural bias in pricing policies.

However, it is a matter of public record that these

IMF/World Bank reforms have been resisted by the Jamaican

government at various times, and have been hotly debated in







40

the popular media regarding the social impacts of

externally imposed structural adjustment programs.

Despite this apparent dissatisfaction, the government

has utilized operational policies along the lines of those

prescribed by the Bretton Woods institutions. While in

principle the early reforms were aimed at stabilizing the

economy, from the outset there was a built-in bias in the

policies for a more liberalized economic system. For

example, the three year Extended Fund Facility (EFF), which

the Jamaican government signed with the IMF in May 1978,

contained specific conditions for exchange rate devaluation

as well as the liberalization of domestic prices (Boyd,

1988). More extensive economic reforms were advanced by the

new administration under Prime Minister Edward Seaga who

held office from 1980-1989.

Under the 1982 and 1983 structural adjustment

programs, quantitative restrictions (QR), on trade were

replaced by tariff equivalents, and by 1984 all QRs were

eliminated. Exchange rate devaluation was pursued in 1984-

1985, followed by the unification of the official and

parallel rates. At the same time, the government began to

reduce the public sector via the divestment of state owned

enterprises (SOEs).







41

The 1990s witnessed the most protracted reforms aimed

at deregulation and liberalization of the Jamaican economy.

The PNP, which took office in 1989 under Prime Minister

Michael Manley, had campaigned in the general elections on

a platform similar to the JLP, vowing to continue the JLP's

market oriented policies. This was an opportunity for a

fresh departure for the PNP, which was noted for its

previous support for democratic socialism and extensive

government intervention. Garrity (1996, p.52) argues that

...the Manley-led government made the decision in 1990
to truly embrace the market and proceed with economic
liberalization reforms. For Manley... [t]o continue the
reforms of the previous JLP government would not be
sufficient to bring about the needed structural
changes, nor were there sufficient resources to
continue the previous reforms....

Based on the limited available options, Manley's

commitment to the liberalization process represented a

turning point in state-society relations in Jamaica. In

particular, the administration sought an accommodative

pattern of governance between state and social actors by

energizing the private sector. In this sense, the

liberalization process can be viewed as "state-sponsored

disengagement from a command role in the economy ... and the

creation of an enabling environment for private sector-led

growth and development." (Garrity, 1996, p.52)







42

Against this background, the sectoral adjustment loan,

which the government signed with the World Bank in 1989,

focused on liberalizing the agriculture sector along the

following lines:

(1) Deregulation of the coffee and cocoa boards by 1990;

(2) Shifts in government focus from production and

markets to support services and infrastructure;

(3) Continuation of the divestment of agro-enterprises

(e.g., sugar);

(4) Elimination of subsidies on agricultural credit;

(5) Reduction of tariffs; and

(6) Elimination of the Generalized Food Subsidies;

As a result, by 1991 the generalized food subsidies

were eliminated, and complete liberalization of the

exchange market and credit interest rate were achieved.

Consequently, the overall liberalization process, since the

first agreement with the IMF in 1977, opened markets to

more competition and reduced the role of the public sector.

As would be expected, the conditions that accompanied

the loans from the Bretton Woods Institutions involved

reforms at both the macroeconomic and the sectoral levels.

Figure 2.7 is a schematic presentation of the kinds of

economic reforms and a partial list of specific tasks

undertaken since the late 1970s. For convenience, the





























Domestic Fiscal Monetary/Fin. Trade Other
Deregulation Reforms Reforms Liberalization Reforms


UL goVL. 01 SUDSl].1es k ]rea.lon
assets on interest, Agriculti
New pricing loans and Credit Bi
system for food and Nati
coffee, Introduction DevelopmE
cocoa, of Bank
citrus and Generalized
pimento and Consumption
more Tax
appropriate
sharing of
export
earnings





Figure 2.7: Selected Economic
1977-98.


Policy Reforms in Jamaica,







44

policy reforms are organized into five blocks: (1) Domestic

Deregulation; (2) Fiscal Reforms; (3) Monetary/Financial

Reforms; (4) Trade Liberalization and (5) Other Reforms.

Within each block, specific tasks are listed, reflecting

implicitly and explicitly policy changes, which seek to

enhance the role of market forces, encourage private sector

initiatives and reduce government regulation and

intervention in the economy. It is interesting to note that

in spite of intense criticisms regarding the reforms' lack

of success in producing the economic recovery predicted by

its proponents, the trend of economic reforms has not been

reversed in periods of economic stress. Such periods have

instead been met with the broadening and deepening of

reform efforts.

Table 2.9 shows selected economic statistics in the

pre-reform and reform periods in Jamaica. In the area of

trade, average tariffs in 1986 were 56 percent but fell to

11 percent in 1995. Exchange rate systems have been an

important policy instrument in Jamaica to establish

restrictions on capital outflows and restrictions for

repatriating export revenues and foreign exchange.

Following the exchange rate liberalization, many of these

restrictions have been dismantled. As evidence of the

process of exchange rate unification and deregulation,







45

Table 2.9 shows that the exchange rate differential (i.e.,

the difference between the average market price for foreign

exchange--inclusive of transaction costs and exchange rate

taxes--and the official rate) was 25 percent in 1986 but

fell to 5 percent in 1995.



Table 2.9: Outcomes of Economic Reform Policies in Jamaica,
1986 vs. 1995.
Structural
Maximum Tax Rate Policy
Tariff Exchange
Reduction Differential Companies Individuals Index, I
Net aver.
Tariffs % Percent Percent Percent 0 198611995 1986 1995 1986 11995 1986 1995 1985 1995
56 JF 25 5 45 32 50 25 0.402 0.684
Source: Adapted from IDB (1997).



In the area of tax reforms, Jamaica adopted various

changes aimed at administrative simplicity and ease of tax

collection. Taxes for companies were reduced from 45

percent in 1986 to 32 percent in 1995. The maximum taxes

for individuals were reduced from 50 percent to 25 percent

over the same period. Finally, the structural policy index,

which is a summary statistic of the extent to which market

forces are allowed to operate in the economy, increased

from 0.402 in 1985 to 0.684 in 1995.









2.4 Summary

The evidence presented in this chapter suggests that

the direction of change of economic structure of the

Jamaican economy that began around 1980 continued over the

past two decades. Over the period 1966-1998, growth rates

of GDP and economic sectors were low but positive. With few

exceptions, sectoral growth rates in the reform period,

1980-1999, were higher than in the pre-reform (1962-1979)

period.

Some structural change within the agricultural sector

was discerned from the data. In particular, export

agriculture declined relative to other agri-sub-sectors.

Domestic agriculture recorded impressive growth, and it is

this agri-sub-sector's overall growth performance that

partially compensated for the negative and low growth in

the other agri-sub-sectors. Significant economic reforms

were implemented, and analysis of their impact on

agriculture will be undertaken in a later chapter.















CHAPTER 3
METHODOLOGICAL AND EMPIRICAL ISSUES IN ECONOMIC
REFORMS AND SUPPLY RESPONSE


The purpose of this chapter is four-fold. First, it

reviews the methods that have been used in the literature

to analyze the impact of economic reforms in developing

countries. Second, it reviews the literature on supply

response analysis. Third, it develops a crop supply

response model for Jamaica; and finally, it presents the

data sources for the analysis.

3.1 Review of the Economic Reform Literature

In this review of the economic reform literature two

sets of issues are addressed. First, attention is directed

at the way the economic reforms are conceptualized. Second,

the analytical methods used to evaluate the impact of the

reforms are examined.

One of the most challenging problems in capturing the

impact of economic reforms on target variables is how to

isolate the effect of each of the reforms undertaken. Since

the reforms were undertaken in various areas (trade

reforms, fiscal/monetary reforms, etc), at different points

in time, and with varying levels of intensities, their

47







48

effects on particular variables in the economy become

compounding and difficult to isolate. Khan (1990) argues

that it is theoretically and empirically difficult to link

all the policy reform measures to the ultimate targets of

the policies. Hence most studies have attempted to assess

the effects of the overall policy package on particular

target outcomes. This is the thrust of the studies by

Chadha et al. (1998), and some of the studies in Tshibaka

(1998), Campbell and Stein (1991), La Guerre (1994), Vald6s

and Muir-Leresche (1993), Commander (1989) and Weeks

(1995). In these studies, neither the precise nature of the

underlying economic relationships, nor the specific

policies adopted are made explicit. Instead, the attention

is on whether or not the program package, which in effect

gives rise to a particular policy environment, has been

"effective" in the sense of achieving broad macroeconomic

objectives (Khan, 1990).

Most of the studies reviewed proceeded from the

premise that the economic reforms originated from the

conditionalities that accompanied Fund/Bank-supported

packages. In all country experiences the problem of

economic instability provided the initial imperative to

seek IMF assistance, which, when it came took the form of

stabilization policies. However, in most of the studies







49

reviewed the stabilization policies are treated en passant,

or not at all, and instead attention is focused exclusively
3
on economic outcomes associated with structural adjustment

For most developing countries that have been pursuing

the Fund/Bank economic programs, the last decade has been

one in which the reforms have intensified the process of

liberalizing the economic system. A frequently raised

question with respect to Fund/Bank-supported structural

adjustment programs cum structural reforms for economic

liberalization, has been whether these programs are

effective in achieving stated economic objectives. A second

and related question is how to measure the effects of these

reform policies on the target variables identified in the

programs. With respect to this latter question, Guitidn

(1981) has argued that economic performance under a policy

package should be compared to a counterfactual. The latter

is defined as economic outcomes that would have taken place

in the absence of the program package. The concept of a

counterfactual is intuitively appealing and is a standard

yardstick widely used in economics to measure the impact of

policy interventions. Since the counterfactual is



3 Stabilization policies are designed to put the economy back on its
equilibrium path, whereas, structural adjustment policies aim at
putting the economy on a new (higher) equilibrium.







50

unobservable, it has to be estimated, hence alternative

methodologies used to evaluate the program effects are

judged in terms of their estimates of the counterfactual.

Although the preceding two questions have generated a

large body of literature over the past two decades, there

is little consensus in the economics profession either

about the impact of past programs on target variables, or

about how to estimate the effects of program packages

(Khan, 1990). With respect to measuring the impact of

reforms, there are three main approaches that have been

applied in the empirical literature. These are

(1) The before-after approach, which compares the

behavior of key macro-economic variables before and

during, or after, any particular reform period, or

policy package.

(2) The with-without approach compares the performance

of macro-economic variables of non-program countries

(the control group) with those from program

countries. A modified version of this approach is a

reduced form regression estimate that controls for

initial conditions in program and non-program

countries.

(3) The comparison-of-simulations approach. This

approach simulates performance of Fund/Bank-type







51

programs then compares them with simulated outcomes

from another set of policies.

Approaches (2) and (3) have been used extensively in cross-

country studies to assess the impact of Fund/Bank-supported

programs. The before-after approach has been the most

prevalent in country case studies using time series data.

3.1.1 Before-After Approach

The before-after approach has been the most popular in

the literature on Fund/Bank-support programs. Reichmann and

Stillson (1978) were the first to use this approach in an

examination of 79 Fund/Bank-supported programs over the

period 1963-1972. They compared growth, balance of payments

and inflation in the two years prior to and after the

implementation of the program. Connors (1979) also used the

before-after approach to evaluate 31 programs in 23

countries over the period 1973-1977. More recent studies

include Singh (1995) and the papers in Le Franc (1994), on

Jamaica. Similar studies on other developing countries

include most of the papers in Tshibaka (1998), Campbell and

Stein (1991), La Guerre (1994), Vald~s and Muir-Leresche

(1993), Commander (1989) and Weeks (1995).

The before-after approach is viewed by some analysts

as providing a relatively poor estimate of the counter-

factual to program effects by assuming that non-program







52
determinants of economic variables remain constant between

non-program and program periods. This assumption has been

questioned in the literature in light of the fact that non-

program determinants of economic outcomes (e.g., terms of

trade variations, changes in international interest rates,

weather, etc.) typically change year after year. Conse-

quently, Goldstein and Montiel (1986) have demonstrated

that the before-after estimates of program effects are

biased, since all economic changes in program period are

incorrectly attributed to program factors. These authors

suggested further that the before-after estimates are

unsystematic over time, since in any particular year

program effects will often be dominated by non-program

factors. For example, if a hurricane damages infra-

structural works and agricultural crops, then agricultural

growth may decline, causing all programs in that year to

appear to have performed poorly.

3.1.2 With-Without Approach

The with-without approach seeks to overcome the

shortcomings of the before-after approach by comparing

economic outcomes between program and non-program

countries. Since both groups of countries are subjected to

the similar external environments, it is argued that this

comparison cancels out the non-program determinants, so







53

that any observable differences in outcomes in the two

groups of countries are attributable to the Fund/Bank-

supported programs. In other words, the economic outcomes

observed in non-program countries are taken as the

counterfactual of what would have occurred in program

countries in the absence of Fund/Bank-supported programs.

Donovan (1981, 1982) was the first to use this

approach on a sample of Fund/Bank-supported programs

implemented between 1970-1980. Loxley (1984) also applied

this approach to 38 less developed economies (income of

$690 or less) with Fund/Bank program during 1971-1982.

Other similar studies include Gylfason (1987) and Pastor

(1987).

A major problem with the with-without approach is that

countries in the sample of non-program and program

countries are not randomly selected. They are program

countries because of poor economic conditions prior to

entering into a Fund/Bank-support program. Consequently

there is a systematic difference between these two groups

of countries and the non-random selection of program

countries produces a biased estimate of program effects.

This is the result of attributing observable differences in

economic outcomes between program and non-program countries

to program status, when in fact the initial economic







54
position of the two groups of countries is an important

determinant of economic performance.

To overcome this problem requires identifying the

specific differences between program and non-program

countries in the pre-program period and controlling for

these differences prior to comparing economic outcomes.

This is the idea underlying a modified version of the with-

without approach. In this regard, Goldstein and Montiel

(1986) proposed a generalized evaluation estimator to

control for pre-program differences between program and

non-program countries. This estimator is a reduced-form

relationship that links changes in macroeconomic outcome

(program target variables) to lagged values of the target

variables, lagged values of policy variables and variables

that represent exogenous effects on the target variables.

These authors applied this approach to a sample of 58

developing countries in which 68 programs were implemented

over the period 1974-1981. The approach was extended by

Serieux (1999) to include the effect of democratization in

Fund/Bank-supported program countries.

3.1.3 Comparison of Simulations Approach

Finally, the simulation approach relies on simulations

of economic models to make inferences on hypothetical

outcomes of Fund/Bank-type policy packages. Khan and Knight







55

(1981), created a simulation using panel data for 29

developing countries in a dynamic econometric model. The

objective was to investigate the hypothetical effects of

pursuing a stabilization program with similar policy

characteristics as a Fund/Bank-supported one. The authors

later (Khan and Knight, 1985) extended their simulation

exercise to include comparisons of alternative packages.

More recent studies include Robinson and Gehlhar (1996),

and Chadha et al. (1998), who use computable general

equilibrium (CGE) models for the Egyptian and Indian

experiences with recent economic reforms, respectively.

3.2 Preliminary Issues in Modeling Supply
Response in Jamaica

A number of empirical studies have been conducted in

the past to estimate crop supply response in Jamaica. A

partial list of these studies is shown in Table 3.1. With

the exception of Gafar (1997), all of the studies were

conducted prior to the period 1980-1999. In addition, none

of the studies addressed the issue of the impact of

economic reforms on agricultural supply response.

One approach to capturing the long-run and short-run

changes in agriculture supply response is to use the

Nerlovian-type partial adjustment models. Both quantity and

prices can be modeled to adjust to their long-run or

equilibrium path, and the model is capable of estimating










Table 3.1: Empirical Studies on Agriculture
mme i,% Tnniminsk


56

Supply


both long-run and short-run parameters as well as the speed

of adjustment towards the long run equilibrium. A potential

complication in such an analysis of long-run and short-run

changes is that most economic time series data are non-

stationary and usually characterized by a unit root.

This means that the linear properties of the series

such as its mean and variance, are not constant over the

sample but change over time (Greene, 1993; Gujarati, 1995).

Nelson and Plosser (1982) have shown that the economic

implications of an economic time series that is


Crop/ Funct. Price Elasticity
Category Period Author & Year Form Short-run Long-run
Banana 1954-1972 Gafar (1980) Linear 0.16 0.57
1961-1979 Pollard & Graham log 0.49 -2.72
(1985)
Cocoa 1954-1972 Gafar (1980) Linear 0.41 2.56
1961-1979 Pollard & Graham Log 0.74 0.76
(1985)
Coffee 1953-1968 Williams (1972) Log 0.70 -0.80
1954-1972 Gafar (1980) Linear 0.92 1.15
1961-1979 Pollard & Graham Log 0.10 0.07
(1985) 1
Citrus 1961-1979 Pollard & Graham Log 0.24 -1.33
(1985)
Sugar 1954-1972 Gafar (1980) Linear 0.17-0.29 0.31-0.7
1961-1979 Pollard & Graham log 0.24 1.41
(1985)
Broad Agri.
Aggregates 1964-1990 Gafar (1997) log
Export 0.20 0.35
Domestic 0.15 1.08
Livestock 0.15 0.21
Forestry &
Fishing 0.02 0.21
Total Agri. 0.12 0.23
Source: Adapted from Gafar (1997, p.213).







57

characterized by a unit root are different from those of a

stationary process. In particular, an economic time series

with a unit root will have a permanent response consequent

upon any shock in the system. In contrast, a stationary

series will reflect only a transitory response. Therefore,

the response to economic reform policies is not simply a

"policy on", "policy off" or one-shot choice. Even if the

reforms were temporary, so long as the economic series

possesses a unit root, there will be a permanent response.

Non-stationarity of variables poses problems of

estimation of functional relationships using conventional

econometric methods. In the first place, Fuller (1976) has

shown that under non-stationarity the limiting distribution

of the asymptotic variance of the parameter estimates is

not finitely defined, hence the conventional t and F tests

are inappropriate. Secondly, non-stationarity gives rise to

spurious correlation among variables (Greene, 1993). In

macroeconomic time series it is not unusual to find that a

variable is non-stationary in its level4 but stationary in

first differences. In technical terms, if Yt is non-

stationary, but AYt (the first difference) is stationary,

then Yt is integrated of order one, i.e., Yt~I(1). If two


4 Data that have not been transformed in any way (such as logarithmic
transformation, first differences, etc.), are said to be in 'level'
form.







58

series Yt and Xt are 1(1) then a linear combination of them,

e.g., Zt = Yt aXt may also be 1(1). However, there may also

exist, a value for a that ensures that Zt is stationary. In

such an event, the two series are said to be cointegrated,

and the cointegrating vector is denoted (1, a).

It is tempting to conclude that in order to estimate

meaningful relationships among non-stationary variables,

all that is necessary is to difference the variables until

they achieve stationarity and then estimate the

relationship. However, Johansen and Juselius (1990) argue

that unless the difference operator is also explicitly

applied to the error process, such differencing results in

loss of information. In this event, resorting only to

estimating the relationship in difference form captures

only the short-term effects, while the long-term

relationship among the variables is left undetected

(Nickell, 1985). Finally, differencing the economic series

may not be appropriate, such as when economic theory

postulates a relationship among variables in levels, not in

difference forms.

To overcome these problems, econometricians have

developed an approach known as error correction models

based on cointegration. Hylleberg and Mizon (1989, p.124)

claim that







59

...when estimating structural models it is our
experience from practical applications that the error
correction formulation provides an excellent framework
within which, it is possible to apply both the data
information and the information obtainable from
economic theory.

Error correction modeling requires two conditions

(1) All variables must be integrated to the same order,

i.e., Yi-I(d), where d is the number of times Y has

to be differenced to achieve stationarity; and,

(2) All variables must be cointegrated of order (d b),

where b>O.

The idea underlying cointegration is that one or more

linear combinations of non-stationary variables are

stationary. That is, cointegration approaches the station-

arity issue as linear combinations of economic series,

rather than by differencing the series. The implication of

this is that if a set of variables is cointegrated then,

following the Granger Representation Theorem (Fuller, 1985)

a valid error correction representation of the data exists.

In effect, then, cointegration is a test of existence of a

long-run relationship of variables that are integrated of

the same order (Greene, 1993; Gujarati, 1995). However, an

important feature of error correction models based on

cointegration is that the data in both levels and

differences are included, thereby facilitating investi-

gation of both short-run and long-run effects in the data.









3.3 Error Correction Model

The error correction model (ECM) can be derived from a

re-parameterization of an Autoregressive Distributed Lagged

(ADL) model (Hendry et al., 1984). Alternatively, the ECM

can be derived from the dynamic optimizing behavior of

economic agents. This latter approach is presented here.

Following Nickell (1985), suppose economic agents

optimize their behavior with respect to an inter-temporal

quadratic loss function:


(3.1) L = t Y Y 8
s-0

(Y- tsy*s- Yt+S-1 A

where Y* is the desired or long-run equilibrium value of Y

that the economic agent can control to minimize L,

conditional on information at time t, and subject to

movements in Y*. The discounting and weighting factors are

a, (OO), respectively.

Minimizing (3.1) with respect to Yt+s gives a second

order difference equation whose solution at time t is:

(3.2) Ayt = 02AYt + ( Pl)


[( + (1 X- (2)0 (aq1)N+ -
EI J t







61
where pi is the stable root (i.e., the root that is < 1)

from solving the characteristic equation of the general

Euler equation which was derived from minimizing (3.1).

Equation (3.2) is an optimal rule and has an error

correction term in brackets [.]; the coefficient (I-Ii) is

the speed of adjustment, i.e., the speed of closure to any

discrepancy between desired and actual values of Y. In the

error correction term in (3.2) the long-run target is a

convex combination of all target values from Y-, onwards.

Nickell (1985) has shown that when 02 = 0 in the loss

function (3.1) then equation (3.2) nests the forward-

looking partial adjustment model (PAM):


(3.3) AY = ( I %i ( tiNY$s }Yt-
s-0II

It is important to note that the dynamic equation

(3.3) does not necessarily have to be of a standard PAM

type. To estimate (3.3) empirically requires the para-

meterization of Yj ,. One way is to model this sequence of

expected target values as a stochastic process such that

Yt.S are expressed in terms of current and lagged values and

substituted into (3.3). The resulting equation may not

necessarily result in a PAM. Indeed, Nickell demonstrates

that when







62

...we allow the target to follow anything more
complex than a first order autoregression, the
structural equation [3.3], which is fundamentally a
partial adjustment model, will reduce to an error
correction mechanism in terms of observable variables
(Nickell, 1985, p.124).

Suppose that the actual values of the series follow a

second order autoregressive scheme with a unit root and a

drift:

(3.4) Yt+s= g + PYts-1 + (1 P t+S-2 + Et+S

where s>O, et+, is white noise, and g is the drift. Denoting

the expected value with an (*), the expectation of (3.4)

is:

(3.5) Y s = g + PYt+s-1 + (1 P t s-2

Following the derivations by Nickell (1985) and

Alogoskoufis and Smith (1991) the solution to the second-

order difference equation (3.5) is:

*.= yt-+1 0 4 A4
(3.6) Y = S2-- gs + ( Y


Substituting (3.6) into (3.2) yields the decision rule in

the form:


(3.7) =y C + + +y + iii1 PIAY -y- Y1)
(1 + cxp1( -F t3

where c =g( PX2 P0101 P1X1 (2) glcX1 -
(2 P)[1 + ai(1 P)] (2 PX plp)

Equation (3.7) is written in the form of an ECM. The

parameterization of Y* in terms of exogenous variables would







63
reflect the long-run cointegrating relationship. According

to Nickell (1985, p.124),

since it is almost a stylized fact that aggregate
quantity variables in economics follow second order
autoregression with a root close to unity, we may find
the error correction mechanism appearing in many
different contexts.

3.4 An Error Correction Model for Crop Supply
Response in Jamaica

The dynamically unrestricted version of the ECM in

(3.7) can be expressed as:

(3.8) = *0 + 1AQ + 2(Qtl Qt

where Qt is output in logarithm and Vo, *1, *2 are straight-

forward analogues of the intercept and coefficients in

(3.7). Generally, Q* is parameterized in terms of other

(weakly) exogenous variables. This parameterization shows

the long-run cointegrating relationship between the

exogenous variables and the dependent variable, Qt. Given

the unrestricted nature of (3.8) a wide range of possible

processes that describe the law of motion of Qt can be

accommodated. Following previous agricultural supply

response models, quantity supplied is postulated as a

function of the expected values of a set of variables that

are believed to capture agricultural incentives. These

exogenous variables are, the price of the crop, Pc, the

price of substitute crops, Psi, i=1,2... and the prices of







64

inputs. In this study, two inputs are considered,

fertilizer and labor, whose prices are denoted as F and W,

respectively. With these specifications of the exogenous

variables, the parameterization of the long-run equilibrium

supply function of the cth crop is:

(3.9) Q. + O + 02iPit + O + 04F +

c = l...n; i=1,2,...; t=l...T; et-iid(0,a).

where all variables are measured in logarithms (to

facilitate interpretation of estimated coefficients), and

have been previously defined. The superscript e denotes

expected value of the variables. Given the parameterization

in 3.9, the general error correction model, with all

variables as previously defined, can be written as:

(3.10) AQt = X0 + XIAQtI + X2Apt_1 + \31AP1,t + X4AWe_

+ X5AF t1 + X6(Q +
+ es I+A t- Qt-J +


To give empirical content to these models, the

specification of expected values must be addressed. Clearly

if economic agents had full information about the current

variables at the time they are set, then actual values of

the variables would be substituted for their expected

values. When this is not the case, some process of

expectations formation has to be assumed. The applied

literature on ECM has largely ignored issues of







65

expectations. Because of its simplicity in practice, actual

values are usually substituted for expected values. An

alternative to this procedure is to assume that crop and

input prices are determined by policy changes for one year

in advance. With this assumption, expectations about

current economic variables have to be based on information

that is available up to the end of the previous period,

(t- I). This can be expressed as:

(3.11) x=e =(XtjZj =

where Xt denotes the economic variable of interest, and 0 is

the information set available to the economic agent.

Previous studies of crop supply response in Jamaica

assumed naive expectation, based on past (simple) lagged or

polynomial lagged prices. Under rational expectations,

expected current and future prices of crops, wages and

inputs will reflect the generation process for these

explanatory-forcing variables. As noted previously, a

second order auto-regressive process with a unit root with

drift seems adequate to describe the process followed by

many economic series. The data on quantities, prices and

inputs for this study have been analyzed and found to

follow this process. Consequently, the expected prices of

crops and inputs are characterized as follows:

(3.12) <:"I-, = ao + a, '-, + 0 a, ,-









(3.13) P-, = bo + bIP.1 + (I bj)P. ,-

(3.14) Ft= co +cF-, +(1-c )F,

(3.15) Wt =do +dW.-1 +(-dl)W,.2

Using (3.8)-(3.15), the general error correction model for

the cth crop is:

2 2
(3.16) AQ ct = a0c + .,t6Q.,t-j + Y a2j c,t-J + E a3ik APi,t-k
jal k-1

2 2
+ a.lawt 1 + E (mA -. + ajQc 51pc (21psi
1==1

-83W, -840),, ;


where, i=1,2, all variables as previously defined, and the

random error term is suppressed.

The last term in (3.16) is the error correction term. In

the empirical estimation of the ECM (3.16), the error

correction term is usually specified as the residual from

the cointegrating relationship.

The Engle-Granger (1987) two-step method has been used

extensively in the applied literature to estimate the ECM

(3.16). However this method assumes that the cointegration

vector is unique. Except in the bivariate model, this

assumption may be violated in multivariate models. To test

for, and estimate multiple cointegrating vectors, Johansen

(1988) and Johansen and Juselius (1990) have devised an

appropriate method within the following framework. Define a







67

standard vector autoregressive (VAR) model with lag length

k as:

(3.17) Xt = i1x + H2Xt_2 + ... + [kXt-k + Et

t=l1..., T

where X is an Nxl vector of N endogenous variables, et-iid


(0,A) with dimension NxN. The long-run, or cointegrating

matrix is:
(3.18) I nI n2 nk = n

The number of distinct cointegrating vectors, r, which

exists between the variables of X, is given by Rank (n).

Most economic time series appear to be integrated to

the order of one, in which case, r N-1, where N is the

number of variables in the vector X. In the case of a

bivariate model, N=2, and therefore if the variables are

cointegrated, then there is a unique cointegrating vector.

The matrix n is then decomposed as:

(3.19) H = ap'

where I represents the matrix containing the r

cointegrating vector, and a is the matrix of weights with

which each cointegrating vector enters each of the

differenced X equations. A large a value implies that the

system will respond to any deviation from the long-run

equilibrium path with a rapid adjustment. If a's are zero







68

in some equations, this is a sign of a weak exogeneity,

implying that the variable does not respond to the

disequilibrium in the system. The parameters a and P form

an over-parameterization of the model. However, the space

spanned by P, sp(P), can be estimated, and shown to be the

empirical canonical variates of Xt-k with respect to AXt.

This is in effect the following theorem advanced by

Johansen:

The maximum likelihood estimator of the space spanned
by P is the space spanned by the r canonical variates
corresponding to the r largest squared canonical
correlations between the residuals of Xt-k and AXt
corrected for the effect of the lagged differences of
the X process. (Johansen, 1988, p.233)

Implementation of this theorem begins by re-para-

meterizing (3.17) (detailed derivations are in Johansen,

(1988) and Enders, (1995)), into the following error

correction model:

(3.20) AXt = FAXt- + --+ Fk-AXt-k+1 + FkXt-k + St

where F = -I + R1 + H2 + ... + Ili, i=l ...k

Without any loss of information, the ECM in (3.20) is

therefore a transformation of the VAR(k) model in equation

(3.17), and is expressed in first differences and augmented

by the error correction term, fkXt-k. The long-run equi-

librium or impact matrix is the matrix F, and is equivalent

to rI = a' 0 in (3.19). The rank of fl is the basis of







69

determining the number of cointegrating relationship

between the variables in the ECM (3.20). Johansen (1988)

identifies three possibilities with regards to Rank(l),

(1) rank(I) = 0. This means that the variables are not

cointegrated and the model is basically a VAR in

first differences.

(2) 0 < Rank(11) < p. In this case, the variables are

cointegrated and the number of cointegrating

relationship(s) is less than the number of

variables, p, in the model.

(3) rank() = p. This means that all variables are

stationary and the model is in effect a VAR in

levels.

The loglikelihood representation of (3.20) is:

(3.21) L ) =


I (Rot + P'R + a'R)


Johansen's procedure begins by regressing AXt on the lagged

differences of AXt and generating fitted residuals Rot, then

regressing Xt-k on the lagged differences and generating

fitted residuals, Rkt. These fitted residuals are then used

to construct the following product moment matrices:

1
(3.22) S= T 1 Rjt i, j = 0, k
Tj Rit
i-i







70

The product moment matrices (3.22) are then used to find

the cointegrating vectors by solving the determinant:

(3.23) 1xSkk SkOSOOSOkj = 0

This will yield the estimated eigenvalues (xi, ..., n) and

eigenvectors (vl,...,vn), which are normalized such that:

(3.24) V'SkkV = I

where V is the matrix of eigenvectors. The most significant

eigenvectors then constitute the r cointegrating vectors,

i.e.,

(3.25) 1 = V '"" "Vr)

Using (3.25), a is then estimated from (3.19).

The critical issue in all of this is to determine

which, and how many, of the eigenvectors in (3.24)

represent significant cointegrating relationships. First,

the 0 vectors that have the largest partial correlation

with AXt, conditional on the lags of AXt, are identified.

Second, the eigen vectors that correspond to the r largest

eigen values are chosen. Finally, to determine the value of

r the following test statistics suggested by Johansen

(1988) are employed:

n
(3.26) 1(q, n) = -T : ln(l X)
i=q+l

(3.27) 2(q, q + 1) = -Tln( q+1)







71

The null hypothesis HO: r~q is tested with (3.26), while

HO: r = q is tested against HI: r = q + 1 with (3.27). The

critical values for these tests are taken from Osterwald-

Lenum (1992). The critical values from this source

recalculates and extends those critical values from

Johansen (1988) and Johansen and Juselius (1990), to handle

a full test sequence from full rank (r = p, i.e., Xt is

stationary) to zero rank (r = 0, i.e., all linear

combinations of X are I(1)), for at most 11-dimensional

systems.

The cointegration technique is used to determine the

long-run relationships among variables. This co-movement of

variables in a long-run supply function has not been

explored for Jamaica. Cointegration analysis is appropriate

in this regard. It will also suggest which variables in the

supply function are in the long-run equilibrium.

3.5 The Data

The principal sources of annual time series data,

which are used in this study, are the Food and Agriculture

Organization (FAO) agricultural database (FAOSTAT), avail-

able on the internet, and annual publications of various

government agencies in Jamaica. These include, Economic and

Social Survey of Jamaica (Jamaica, Planning Institute of

Jamaica); Production Statistics (Jamaica, Statistical







72

Institute of Jamaica (STATIN)); Statistical Digest

(Jamaica, Bank of Jamaica), Statistical Year Book of

Jamaica (Jamaica, Statistical Institute of Jamaica

(STATIN)); Census of Agriculture, 1968, 1978, 1996

(Jamaica, Statistical Institute of Jamaica (STATIN));

International Financial Statistics (IMF); and data from

published and unpublished documents from the Ministry of

Agriculture in Jamaica.

It should be noted that the data on Jamaica contained

in the FAO database, FAOSTAT, IMF and World Bank sources

are based on publications and data supplied by STATIN,

Ministry of Agriculture, and other agencies of the

Government of Jamaica. The data used to estimate the supply

functions in this study are taken from the FAO FAOSTAT

database. This is the most comprehensive data set on crop

output and prices in Jamaica. However, in several areas the

data are less than ideal. For example, whenever data are

not available the FAO provides its own estimates based on

past crop performance, other crops and other country's

data. Jamaican officials claim that production data for

some crops reported on the FAO data base are not monitored

in Jamaica.

The crop output and price data are annual series, and

are collected at the farmgate through periodic production







73

surveys and agricultural censuses. Farmers in the Caribbean

rarely keep records of production, cost expenditures and so

on. Hence, for these farmers it is difficult to recall data

for production surveys that are conducted between long

periods. Consequently, this casts some amount of suspicion

on the reliability of the data. Nonetheless, these are the

data reported as official statistics, and are used by the

government as a basis for policy analysis/discussion.

Several important economic aggregates were not

adequately covered in the data sources. In particular, data

on agricultural wages are not available on Jamaica. As a

result, a proxy for this variable was constructed from data

on compensation to agricultural employees recorded in the

national accounts. Data on consumer price indexes--the

basis for deflating crop and input prices in this study--

were not available as continuous series over the 1962-1999

period. Consequently, a combination of splicing indexes and

converting the final 1962-1999 series to a 1980 base year

was undertaken.

Although the data are less than ideal in several

areas, and are bound to contain some noise, the advantages

of using them are:







74

(1) they are from a common source and are characterized

by a common accounting/estimating procedure in their

derivation; and

(2) the data are the most comprehensive time series

available on Jamaican crop supply in a single

source.

3.6 Summary

There are competing approaches in the literature on

how to evaluate the impacts of economic reforms associated

with Fund/Bank-type policy packages. Each approach utilizes

the idea of a counterfactual against which actual economic

outcomes in the program period are compared. Estimating

crop supply responses over a period in which significant

policy changes have occurred, requires a modeling framework

that is capable of capturing both the long-run and short-

run changes. The Nerlovian-type supply response models have

been used in the literature for this purpose. However, in

the context of data series that are non-stationary, this

approach can produce spurious regressions. A more

appropriate analytical framework is that provided by error

correction models based on cointegration analysis.















CHAPTER 4
EMPIRICAL ESTIMATION OF CROP SUPPLY RESPONSE


The aim of this chapter is four-fold. First, the

central motivations for using error correction modeling

based on cointegration theory in this study are presented.

Second, the time-series data on prices and quantities to be

used in the ECMs to estimate crop supply responses in

Jamaica are tested for stationarity. Third, long-run supply

responses are estimated for eight agricultural crops.

Finally, the short-run dynamics of the crop supply

responses are analyzed.

4.1 Motivations for Using Cointegration Analysis

Chapter 3 provides a fairly elaborate treatment of the

theoretical and statistical aspects of error correction

modeling based on cointegration theory. Against that

background, it is useful to recall in a cryptic, condensed

and fairly non-technical way, the central motivations for

using this approach to address the issues with which this

study is concerned.

The empirical purpose of this study is to investigate

the impact of economic reforms on crop supply responses in







76

Jamaica. Generally, supply response models in agriculture

postulate a long-run relationship between output and

agricultural incentives (Askari and Cummings, 1976).

Deviations from this long-run equilibrium occur in the

short-run and may involve considerable adjustment costs to

the economic agent. This is especially the case when

significant policy changes are implemented. Jamaica

provides a good case for evaluating the connections between

economic reforms and supply responses. The country has

traditionally been heavily dependent upon agriculture for

food, employment and export earnings. A combination of

excesses in state intervention and adverse world economic

conditions prompted major economic reforms in the late

1970s in the form of serious macroeconomic stabilization

policies under the direction of the IMF.

During the early 1980s more far-ranging structural

reforms were instituted in an effort to reverse the current

account and fiscal deficits, reduce inflation and monetary

growth to achieve financial stability, restore economic

growth, and so on. Implicitly in these early reforms, but

more explicitly in the late 1980s and throughout the 1990s,

the aim has been to re-orient the economy towards a more

liberal economic system. These reforms took expression in

progressive devaluation of the Jamaican dollar, elimination







77

of state marketing boards, liberalization of agricultural

input and output prices and privatization of state held

monopolies and public enterprises. Intuitively, therefore,

the analysis of supply responses in these situations would

require a modeling framework that is capable of incor-

porating both the long-run and short-run changes.

A preliminary analysis of the Jamaican crop output and

price data series over the 1962-1999 period reveals sub-

stantial fluctuations, especially since the early 1980s.

The strongly time trended data imply a statistical problem

that has not been addressed in applied analysis on Jamaican

crop supply responses. This statistical problem is referred

to, in the literature, as non-stationarity of the time

series.

The analytical approach chosen in this study to deal

with the above mentioned issues, emphases the importance of

considering the interactions between the variables in the

system in a simultaneous equation model, and to distinguish

between the short-run and long-run effects. The modeling

approach differs from those previously used to model

Jamaican crop supply responses in two very important ways.

First, the data are analyzed as a full system of equations.

This allows for possible interactions in determining the

precise relationships among the variables in the system.







78

Second, the multivariate cointegration modeling which

is used in this study, is designed for this type of

empirical work by explicitly classifying the non-stationary

and stationary components and facilitating an inter-

pretation in terms of the dynamics of short-run and long-

run effects. There are two general motivations for using

error correction models based on cointegration theory in

applied economic analyses. First, when time-series data are

non-stationary, i.e., their linear properties, such as mean

and variance, are time dependent, then conventional

econometric analysis on such data may produce "spurious"

results.

Second, even though a set of time series may

individually be non-stationary, there may be a linear

combination among them that is stationary. Such series are

said to be cointegrated, that is, they have a tendency to

move together in the long-run, even though in the short-run

they may diverge from each other. This co-movement of

related series suggests the existence of long-run

relationship between them. So when data are non-stationary

there is the additional possibility that the data

generating process contains information about the

equilibrium process that makes the process adjust toward

the long-run steady state or the equilibrium path.







79

There is, therefore, additional economic insights

gained from the recognition that the data are non-

stationary. The component FkXt-1, (=- ap'Xt-i, since II = ap'),

in equation (3.20), is directly related to the non-

stationarity in the data. Since this component contains

information about the speed of adjustment, a, to some long-

run relations, P'Xt, and if the data are cointegrated, then

the economic process can generally be understood within a

theoretical model that assumes some adjustment behavior.

This means that theoretical models such as the partial

adjustment model (PAM), which assume static equilibrium,

cannot be used when data are non-stationary.

4.2 Definition of Variables

In this chapter, error correction models based on

cointegration theory are used to test the hypothesis that

long-run relationships exist between crop output and price

incentives in Jamaica. The estimation procedure is based on

the work of Johansen (1988) and Johansen and Juselius

(1990). There are basically three steps in this estimation

procedure

(1) test the order of integration of the variables and

specify the lag length of the variables using a

standard vector auto-regressive (VAR) specification.







80

(2) estimate the ECM and the number of cointegrating

relationship(s) among the variables included in the

ECM.

(3) perform short-run analysis by conducting innovation

accounting on the ECM.

For each crop a p-variable vector autoregression (VAR)

of lag k is specified. The model is re-parameterized into

an error correction model (ECM) as specified in equation

(3.20) and estimated according to the methodology of

Johansen (1988) and Johansen and Juselius (1990). Esti-

mation is done using the Regression Analysis for Time

Series (RATS), version 4.3 (Doan, 1996), and Cointegration

Analysis for Time Series (CATS) in RATS, (Hansen and

Juselius, 1995), computer programs. The variables in each

ECM include the output (quantity) of the crop of interest,

the price of the crop, the price of a substitute (or

alternative) crop, and two input prices, average agri-

cultural wage rate and average fertilizer price.

The choice of a crop price variable is critical for

the estimation of crop supply responses. In this regard,

Askari and Cummings (1976) suggest using any one of the

following prices:

(1) Nominal farmgate price;

(2) Farmgate price deflated by any one of the following:









(a) a price index of farmer's inputs;

(b) a consumer price index; and

(c) some index of the prices of competitive crops (or

the price of the most competitive crop).

An additional issue that is related to the choice of

an appropriate price variable for the crop supply functions

especially within the context of the IMF/World Bank

programs in Jamaica, is that agricultural price incentives

are influenced by various macroeconomic policies. Of

particular importance, in this regard, is the real exchange

rate (RER), defined as the ratio of prices of tradable (PT)

to non-tradable (PN) goods (Tsakok, 1990). Krueger (1992)

and Schiff and Vald6s (1992a, 1992b), have shown

extensively how macroeconomic policies in developing

countries generally have a RER effect, which ultimately

affect output price and agricultural supply. In the

empirical literature the RER is usually approximated as

RER=e*WPI/CPI, where WPI is the foreign wholesale price

index, CPI is the domestic consumer price index, and e is

the official exchange rate. The World Bank's approach to

showing the link between macroeconomic policies (as

represented by the RER) and real crop prices, decomposes

real producer price as follows:

RPP = PF/CPI = PF/PBe = e(WPI/CPI) = NPC*RER*pB







82

where RPP is real producer price, PF is the farmgate

producer price, PB is the border price, NPC is the nominal

protection coefficient, and PB is the real border price of

the country's exports (World Bank, 1994). This definition

of RPP shows that it contains information on the RER and

also reveals that it is really hazardous to include both

the RER and RPP in the same equation (Mamingi, 1997).

In addition to the issues raised above, two additional

considerations were taken into account in the choice of the

crop price variable. First, since farm producers sell most

of their marketable surplus immediately after reaping,

farmgate prices seem to be a good approximation for prices

received. Second, since farmers purchase most of their

requirements from retail markets, the consumer price index

seems to be an appropriate deflator for producer prices.

For these reasons, therefore, the farmgate price deflated

by the CPI was used as the real producer price (RPP)

variable in the supply functions. In effect, this price

variable reflects not only price incentives to the

producers, but also macroeconomic (reform) policies (as

represented by the RER). In particular, this link between

the price variable and macroeconomic policies is important

in order to pursue an investigation of the hypotheses

advanced on page 11.







83

The question regarding which alternative (or

substitute) crop to include in a particular crop's ECM

proved to be difficult. In the absence of recorded data on

this issue, the final choice of the substitute crop

resulted from a consideration of a number of factors. Among

these were, the nature of the crops (traditional export

versus domestic crops); the terrain where the crops are

grown (mountainous versus flat lands); tree-crops versus

annuals; and, root-crops versus vine-crops. The difficulty

of identifying an appropriate substitute crop (or crops)

can be demonstrated in the case of banana. Possible

substitute crops for banana are coffee and sugar, since

these are all traditional export crops. Banana is grown

extensively on flat lands, but is also cultivated on hilly

terrains as is coffee, whereas, sugar growing has been

confined to the relatively flat plains in Jamaica. Hence,

both sugar and coffee are plausible substitutes for banana,

and the choice of either one or both becomes an empirical

issue.

A further complication in the choice of a substitute

crop arises. In the early 1990s, this researcher observed

two cases where lands which were previously used to

cultivate sugar and banana were converted into papaya

groves. Telephone interviews conducted with officials of







84
the Ministry of Agriculture, Government of Jamaica,

indicate that these are not isolated cases, and that

similar land conversions have been observed into cut-

flowers, decorative foliage, aloe-vera, ochro, and other

non-traditional export crops.

Several substitute crops were initially included in

each ECM but in all cases it was found that the inclusion

of only one substitute crop price improved the statistical

properties of the model. The alternative crops in each

crop's ECM are reported in Table 4.1.



Table 4.1: Alternative (Substitute) Crops in Each ECM.
Crop Alternative Crops Considered
Banana Sugar coffee, papaya
Sugar Banana papaya
Coffee Banana', pimento, sugar, orange
Pimento Banana', coffee, cocoa bean
Yam Cassava potato, sweet potato
Orange Grapefruit', tangerine, coffee
Cocoa Bean Banana, pimento, coffee
Potato Cassava', yam, sweet potato
Crop chosen as substitute crop.



Two input prices are included in each crop's ECM,

namely, fertilizer price and average wage in the

agriculture sector. The fertilizer price variable is a

weighted price index of all types of fertilizers imported

into Jamaica, with quantities as weights. With respect to

the wage variable, there are no data on agriculture wages




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID ET80BDWF6_QDDYE0 INGEST_TIME 2014-07-17T21:53:06Z PACKAGE AA00022172_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


ECONOMIC REFORMS AND AGRICULTURAL SUPPLY RESPONSE
IN JAMAICA
By
BALLAYRAM
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
2001

ACKNOWLEDGEMENTS
First, I would like to express my deepest gratitude to
my wife for all her support throughout my studies here at
the University of Florida. Her support in this process was
overwhelming and unselfish. My daughters were also very
understanding, supportive, and encouraging, and I thank
them sincerely for the sacrifices they had to make so that
I could complete my graduate studies. My parents and in¬
laws (all now deceased) gave me encouragement and support,
both financial and otherwise, in my quest to achieve a
higher education. Without their love, confidence and moral
support, none of this would have been possible. I owe them
an eternal debt of gratitude, because their support came at
a time when the cost of higher education was far beyond the
reach which their economic resources could accommodate.
I would also like to express my deepest gratitude to
Dr. Carlton G. Davis, chairman of my supervisory committee.
He opened the door for me at the University of Florida, and
provided me with countless hours of individual attention
during my course of studies. On numerous occasions he went
beyond what would normally be the call of duty of an
ii

academic supervisor, to provide me with the support that
enabled me to continue my studies.
I would also like to acknowledge the contributions
from the other members of my supervisory committee: Dr.
Robert Emerson, Dr. Clyde Kiker, Dr. Richard Kilmer and Dr.
David Denslow, who provided me with valuable suggestions
during the preparation of this dissertation.
I also benefited immensely, in my course of studies at
U.F., from discussions with Dr. Max Langham, Dr. Charles
Moss, Dr. Ronald Ward and Dr. James Seale, Jr.
Financial support from the Food and Resource Economics
Department is greatly acknowledged. Dr. Carlton Davis'
efforts in securing this funding for me are greatly
appreciated.
Finally, I wish to acknowledge the excellent computer
guidance and support services which Alex Heyman, Roger and
Laura Clemons, Ed Howard and Juan Carlos, of the Food and
Resource Economics Support Center, provided throughout my
studies here at U.F.
iii

TABLE OF CONTENTS
page
ACKNOWLEDGEMENTS Ü
LIST OF TABLES vi
LIST OF FIGURES ix
ABSTRACT xv
CHAPTERS
1. INTRODUCTION 1
1.1 Background to Jamaica's Economic
Reforms 2
1.2 Problem Statement and Study
Justification 8
1.3 Objectives of the Study 10
1.4 Hypotheses 11
1.5 Conceptual Issues 11
1.6 Procedural and Methodological
Considerations 20
1.7 Summary 22
2. ECONOMIC STRUCTURE, GROWTH, AND POLICY
REFORMS IN THE JAMAICAN ECONOMY, 1962-1999 ... 24
2.1 Structure of the Economy and
Economic Growth: An Overview 26
2.2 Output Trends and Policies in
Jamaican Agriculture 30
2.3 The Economic Reforms 38
2.4 Summary 4 6
3. METHODOLOGICAL AND EMPIRICAL ISSUES IN
ECONOMIC REFORMS AND SUPPLY RESPONSE 47
3.1 Review of the Economic Reform
Literature 47
iv

3.1.1 Before-After Approach 51
3.1.2 With-Without Approach 52
3.1.3 Comparison of Simulations
Approach 54
3.2 Preliminary Issues in Modeling Supply
Response in Jamaica 55
3.3 Error Correction Model 60
3.4 An Error Correction Model for Crop
Supply Response in Jamaica 63
3.5 The Data 71
3.6 Summary 74
4. EMPIRICAL ESTIMATION OF CROP SUPPLY RESPONSE ... 75
4.1 Motivations for Using Cointegration
Analysis 75
4.2 Definition of Variables 79
4.3 Testing for Stationarity 87
4.4 Estimating Long-run Supply Response 90
4.5 Analysis of Short-run Dynamics 106
4.6 Summary 136
5. SUPPLY RESPONSE AND COUNTERFACTUAL ANALYSIS ... 137
5.1 Before-After Analysis 137
5.2 Counterfactual Analysis 145
5.3 Summary 164
6. CONCLUSIONS AND POLICY IMPLICATIONS 165
APPENDICES
A. PRICE OF SELECTED CROPS PRODUCED IN JAMAICA,
1962-1999 (J$) 174
B. QUANTITY OF SELECTED CROPS PRODUCED IN
JAMAICA, 1962-1999 (METRIC TONES) 175
C. SELECTED ECONOMIC STATISTICS ON JAMAICA 176
D. DIAGNOSTIC STATISTICS FOR MODEL SPECIFICATION
AND TEST STATISTICS FOR COINTEGRATION 177
REFERENCES 183
BIOGRAPHICAL SKETCH 192
v

LIST OF TABLES
Table page
2.1 Sectoral Contribution to Real Gross
Domestic Product (Period Averages,
Percentages) 27
2.2 GDP and Sectoral Growth (1995=100)
(Percentage) 28
2.3 Growth Rates of Total Agricultural and
Broad Agricultural Aggregate Indexes 30
2.4 Composition of Agricultural Output
(J$M, 1995=100), Selected Years 32
2.5 Agriculture Sub-sectors as a Percentage
of Total Agriculture (Period Averages) 32
2.6 Growth Rates of Agricultural Components
(Percentages) 34
2.7 Average Rates of Growth of Farmgate
and F.O.B. Prices, Output and the
Nominal Protection Coefficient (NPC)
for the Period 1970-78, Jamaica 36
2.8 Net Barter Terms of Trade and Agricultural
Income Terms of Trade (1980=100)
(Period Averages) 38
2.9 Outcomes of Economic Reform Policies
in Jamaica, 1986 vs. 1995 45
3.1 Empirical Studies on Agriculture
Supply Responses in Jamaica 56
4.1 Alternative (Substitute) Crops in Each ECM .... 84
4.2 Unit Root Tests—Prices and Quantities
in Levels 88
vi

4.3 Unit Root Tests—Prices and Quantities
in First Difference 89
4.4 Diagnostic Statistics for Residual
Tests for the Banana ECM 91
4.5 Tests of Cointegration Rank for Banana ECM .... 95
4.6 Estimated Long-run and Adjustment
Coefficients, P's, a's—Banana ECM 96
4.7 Estimated Long-run and Adjustment
Coefficients, p's, a's for all Crops 100
4.8 Endogenous and Exogenous Variables in Crops'
Impulse Response Functions 109
4.9 Responses of Banana Quantity to Shocks
in the Banana VAR Ill
4.10 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance—Banana VAR 114
4.11 Responses of Quantity to Shocks in Exogenous
Variables 117
4.12 Summary of Forecast Error Variance in
Crop Quantities Explained by Variables
in the VAR (Percentage) 119
4.13 Responses of Sugar Quantity to Shocks
in the Sugar VAR 121
4.14 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance—Sugar VAR 121
4.15 Responses of Coffee Quantity to Shocks
in the Coffee VAR 122
4.16 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance—Coffee VAR 122
4.17 Responses of Pimento Quantity to Shocks
in the Pimento VAR 123
Vll

4.18 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance—Pimento VAR 123
4.19 Responses of Yam Quantity to Shocks
in the Yam VAR 124
4.20 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance—Yam VAR 124
4.21 Responses of Orange Quantity to Shocks
in the Orange VAR 125
4.22 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance—Orange VAR 125
4.23 Responses of Cocoa-bean Quantity to Shocks
in the Cocoa-bean VAR 126
4.24 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance—Cocoa-bean VAR 126
4.25 Responses of Potato Quantity to Shocks
in the Potato VAR 127
4.26 Variance Decomposition Percentage of One-
period and Three-period Forecast Error
Variance—Potato VAR 127
5.1 Descriptive Price and Quantity Statistics
-1962-1979 and 1980-1999 140
5.2 Descriptive Price and Quantity Statistics
-1975-1979 and 1980-1984 141
5.3 Diagnostic Statistics for Forecasts of
Fitted Values 155
5.4 Estimated Long-run and Adjustment
Coefficients, P's, a's for all Crops,
1962-1979 157
5.5 Estimated Long-run and Adjustment
Coefficients, P's, a's for all Crops,
1980-1999 159
viii

LIST OF FIGURES
Figure page
1.1 Conceptualization of the Jamaican
Agricultural Sector 14
2.1 GDP and Sectoral Contribution to GDP
in Constant (1995) Dollars 27
2.2 GDP Growth (Percentage Change Over
Previous Year, 1995=100) 29
2.3 Growth of Total Agriculture, Industry
and Services (Percentage Change Over
Previous Year, 1995=100) 29
2.4 Total Agricultural Output Index (1995=100) .... 31
2.5 Output Indexes for Broad Agricultural
Aggregates (1995=100) 31
2.6 Sub-sectors as a Proportion of Agriculture .... 33
2.7 Selected Economic Policy Reforms
in Jamaica, 1977-98 43
4.1 Residuals for Banana Price in Banana ECM 92
4.2 Residuals for Banana Quantity in Banana ECM ... 92
4.3 Residuals for Sugar Price in Banana ECM 93
4.4 Residuals for Wage in Banana ECM 93
4.5 Cross- and Autocorrelograms in Banana ECM .... 94
4.6 Plot of Eigenvalues for the Banana ECM 94
4.7 Impulse Responses to an Innovation in Sugar
Price—Banana VAR 112
IX

4.8 Impulse Responses to an Innovation in
Fertilizer Price—Banana VAR 112
4.9 Impulse Responses to an Innovation in Wage
—Banana VAR 112
4.10 Impulse Responses to an Innovation in
Banana Price—Banana VAR 112
4.11 Impulse Responses to an Innovation in Banana
Quantity—Banana VAR 113
4.12 Impulse Responses to an Innovation in Sugar
Quantity—Sugar VAR 113
4.13 Impulse Responses to an Innovation in Coffee
Quantity—Coffee VAR 113
4.14 Impulse Responses to an Innovation in Pimento
Quantity—Pimento VAR 113
4.15 Impulse Responses to an Innovation in Banana
Price—Sugar VAR 128
4.16 Impulse Responses to an Innovation in
Fertilizer Price—Sugar VAR 128
4.17 Impulse Responses to an Innovation in Wage
—Sugar VAR 128
4.18 Impulse Responses to an Innovation in
Sugar Price—Sugar VAR 128
4.19 Impulse Responses to an Innovation in Banana
Price—Coffee VAR 129
4.20 Impulse Responses to an Innovation in
Fertilizer Price—Coffee VAR 129
4.21 Impulse Responses to an Innovation in Wage
—Coffee VAR 129
4.22 Impulse Responses to an Innovation in
Coffee Price—Coffee VAR 129
4.23 Impulse Responses to an Innovation in Banana
Price—Pimento VAR 130
x

4.24 Impulse Responses to an Innovation in
Fertilizer Price--Pimento VAR 130
4.25 Impulse Responses to an Innovation in Wage
— Pimento VAR 130
4.26 Impulse Responses to an Innovation in
Pimento Price—Pimento VAR 130
4.27 Impulse Responses to an Innovation in Cassava
Price—Yam VAR 131
4.28 Impulse Responses to an Innovation in
Fertilizer Price—Yam VAR 131
4.29 Impulse Responses to an Innovation in Wage
--Yam VAR 131
4.30 Impulse Responses to an Innovation in
Yam Price—Yam VAR 131
4.31 Impulse Responses to an Innovation in Yam
Quantity—Yam VAR 132
4.32 Impulse Responses to an Innovation in Orange
Quantity—Orange VAR 132
4.33 Impulse Responses to an Innovation in Cocoa-
bean Quantity—Cocoa-bean VAR 132
4.34 Impulse Responses to an Innovation in Potato
Quantity—Potato VAR 132
4.35 Impulse Responses to an Innovation in
Grapefruit Price—Orange VAR 133
4.36 Impulse Responses to an Innovation in
Fertilizer Price—Orange VAR 133
4.37 Impulse Responses to an Innovation in Wage
—Orange VAR 133
4.38 Impulse Responses to an Innovation in
Orange Price—Orange VAR 133
4.39 Impulse Responses to an Innovation in Banana
Price—Cocoa-bean VAR 134
xi

4.40 Impulse Responses to an Innovation in
Fertilizer Price—Cocoa-bean VAR 134
4.41 Impulse Responses to an Innovation in Wage
—Cocoa-bean VAR 134
4.42 Impulse Responses to an Innovation in
Cocoa-bean Price—Cocoa-bean VAR 134
4.43 Impulse Responses to an Innovation in Cassava
Price—Potato VAR 135
4.44 Impulse Responses to an Innovation in
Fertilizer Price—Potato VAR 135
4.45 Impulse Responses to an Innovation in Wage
— Potato VAR 135
a
4.46 Impulse Responses to an Innovation in
Potato Price—Potato VAR 135
5.1 Nominal Price Changes—Banana (BANP)
and Sugar (SUGP) 143
5.2 Nominal Price Changes—Coffee (COFP)
and Pimento (PIMP) 143
5.3 Nominal Price Changes—Yam (YAMP)
and Orange (ORP) 143
5.4 Nominal Price Changes—Cocoa-bean (COBP)
and Potato (POTP) 143
5.5 Real Price Changes—Banana and Sugar
(1980=100) 144
5.6 Real Price Changes—Coffee and Pimento
(1980=100) 144
5.7 Real Price Changes—Yam and Orange
(1980=100) 144
5.8 Real Price Changes—Cocoa-bean and Potato
(1980=100) 144
5.9 Banana Quantity—Actual (LBANQ) and
Forecasted (FOLBANQ) 148
Xll

5.10 Banana Price—Actual (LBANPR) and
Forecasted (FOLBANPR) 148
5.11 Sugar Price—Actual (LSUGPR) and
Forecasted (FOLSUGPR) 148
5.12 Wage—Actual (LWAGE) and Forecasted
(FOLWAGE) 148
5.13 Fertilizer Price—Actual (LFERP) and
Forecasted (FOLFERP) 149
5.14 Sugar Quantity—Actual (LSUGQ) and
Forecasted (FOLSUGQ) 149
5.15 Coffee Quantity—Actual (LCOFQ) and
Forecasted (FOLCOFQ) 149
5.16 Coffee Price—Actual (LCOFPR) and
Forecasted (FOLCOFPR) 149
5.17 Pimento Quantity—Actual (LPIMQ) and
Forecasted (FOLPIMQ) 150
5.18 Pimento Price—Actual (LPIMPR) and
Forecasted (FOLPIMPR) 150
5.19 Yam Quantity—Actual (LYAMQ) and
Forecasted (FOLYAMQ) 150
5.20 Yam Price—Actual (LYAMPR) and
Forecasted (FOLYAMPR) 150
5.21 Cassava Price—Actual (LCASPR) and
Forecasted (FOLCASPR) 151
5.22 Orange Quantity—Actual (LORQ) and
Forecasted (FOLORQ) 151
5.23 Orange Price—Actual (LORPR) and
Forecasted (FOLORPR) 151
5.24 Grapefruit Price—Actual (LGRPR) and
Forecasted (FOLGRPR) 151
5.25 Cocoa-bean Quantity—Actual (LCOBQ) and
Forecasted (FOLCOBQ) 152
xiii

5.26 Cocoa-bean Price—Actual (LCOBPR) and
Forecasted (FOLCOBPR) 152
5.27 Potato Quantity—Actual (LPOTQ) and
Forecasted (FOLPOTQ) 152
5.28 Potato Price—Actual (LPOTPR) and
Forecasted (FOLPOTPR) 152
5.29 Banana—Fitted Output for Reform Period
(BANFITA) and Counterfactual (BANFITC) 161
5.30 Sugar—Fitted Output for Reform Period
(SUGFITA) and Counterfactual (SUGFITC) 161
5.31 Coffee—Fitted Output for Reform Period
(COFFITA) and Counterfactual (COFFITC) 161
5.32 Pimento—Fitted Output for Reform Period
(PIMFITA) and Counterfactual (PIMFITC) 161
5.33 Yam—Fitted Output for Reform Period
(YAMFITA) and Counterfactual (YAMFITC) 162
5.34 Orange—Fitted Output for Reform Period
(ORFITA) and Counterfactual (ORFITC) 162
5.35 Cocoa-bean—Fitted Output for Reform Period
(COBFITA) and Counterfactual (COBFITC) 162
5.36 Potato—Fitted Output for Reform Period
(POTFITA) and Counterfactual (POTFITC) 162
xiv

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
ECONOMIC REFORMS AND AGRICULTURAL SUPPLY RESPONSE
IN JAMAICA
By
Ballayram
May 2001
Chairman: Dr. Carlton G. Davis
Major Department: Food and Resource Economics
A number of economic reform programs have been
undertaken in Jamaica, over the past two decades. Designed
largely by the International Monetary Fund (IMF) and the
World Bank, these reforms focused on correcting internal
policy weaknesses and creating an environment conducive to
sustained growth. The reforms emphasized liberal trade and
exchange rate regimes, a less intrusive and smaller public
sector, and reliance on market forces to determine
agricultural prices and quantities.
Against this background, this study investigates the
impact of these recent economic reforms on agricultural
crop supply responses in Jamaica. The estimation technique
used an error correction modeling framework based on
xv

cointegration theory, within an estimation framework
developed by Johansen. The results of the crop supply
response estimation confirm that there is a long-run
relationship between agricultural crop output and price
incentives. Most of the estimated crop price elasticities
are low, statistically significant, and fall within the
range estimated by other studies on Jamaica. The adjustment
process of the short-run to the long-run was found to be
slow for some crops and higher for others.
Using a counterfactual, which assumed no change in
policy regime, fitted series of supply response functions
from the pre-reform period were forecasted within a
univariate ARMA(p,q) framework and compared to fitted
series of supply responses from the reform period. The
results are mixed. It was found that the impacts of the
economic reforms in Jamaica are crop and time specific.
Mean output was higher in the reform period for four of the
eight crops analyzed. Higher real price shifts were
observed in the reform period for some crops but these
price shifts were also accompanied by higher price-
variability. This suggests that the pro-competitive effects
that were expected to accompany the reforms may have
outweighed the stability impulses of administered prices in
the pre-reform era.
xvi

CHAPTER 1
INTRODUCTION
The past two decades have seen a shift in economic
policies throughout Latin American and Caribbean countries.
Generally, these policy changes have been in response to
debt crises, persistent payments-imbalances, and years of
negative and slow growth. By the 1990s a new orthodoxy on
development thinking could be discerned. This is manifested
in reductions in public sector activities and greater
reliance on private sector initiatives and market forces, a
departure from the policy regimes of the 1960s and 1970s.
In Jamaica, as elsewhere, the new orthodoxy has
influenced economic policies in agriculture. In particular,
policies have shifted away from far-reaching government
interventions and control, to increased reliance on market
signals to allocate resources and form prices. In addition,
economy-wide macro-economic policy changes have been made
which can potentially influence agricultural incentives,
income, and output. Designed largely by the International
Monetary Fund (IMF) and the World Bank in consultation with
1

2
the Government of Jamaica, these reforms1 are expected to
favor Jamaican agriculture under the presumption by the
Bretton Woods institutions that previous policies severely
discriminated against the sector. (This point is elaborated
in Chapter 2,
section
2.4). This study pertains
to
the
impacts that
these
economic policy
reforms
have
on
agricultural
supply
responses in
Jamaica,
as
the
agricultural sector emerges from a heavily regulated sector
to one that is more open2 and liberal. This chapter
discusses the problems that this study addresses.
1.1 Background to Jamaica's Economic Reforms
For purposes of this study, the reform period is
defined as the years 1980-1999, while the pre-reform period
is 1962-1979. The year 1962 is sufficiently far back to
facilitate a meaningful analysis of economic policies and
conditions that led up to the dramatic changes in policies
in the reform period. The reform period marks a re¬
orientation of economic policy. This change arose from the
stabilization programs of the IMF in Jamaica in the late
1970s. It gathered momentum in the 1980s and 1990s in the
In this study, "policy reforms," "economic policy reforms," "economic
reforms," and "structural reforms," are used interchangeably.
2
Openness is defined here in terms of trade/GDP ratio. A high
trade/GDP ratio can coexist with high import tariffs, export subsidies,
and non-tariff barriers, signaling a non-liberal trade regime.

3
form of structural adjustment programs cum reforms for
economic liberalization, under the direction of the IMF,
the World Bank and other leading external funding agencies,
notably the Inter-American Development Bank (IDB) and the
United States Agency for International Development (USAID).
In effect, therefore, for the purposes of this study,
economic reforms refer to the policies stipulated in the
stabilization and structural adjustment programs which
Jamaica has been pursuing since the late 1970s.
When the People's National Party (PNP) took office in
1972 under Prime Minister Michael Manley, it was under
pressure to keep its election campaign promises to increase
real wages and government spending, and to reduce social
inequalities. The responses of the new administration to
this situation, coupled with subsequent external and
domestic events, were to plunge the economy into a
sustained recession and major disequilibrium. During the
previous administration under the Jamaica Labor Party
(JLP), the performance of the economy was impressive. GDP
growth rates were positive and relatively high (5.1 percent
per annum over the 1960-1970 period), inflation was less
than five percent per annum and the balance of payments
showed modest surpluses (Jefferson, 1972; IMF, 1998). Over
the 1970-1980 period, however, real GDP declined by -0.9

4
percent. In fact, with the exception of a negligible 0.7
percent GDP growth in 1978, negative growth rates were
recorded for all other six years between 1974 and 1980.
Similarly, in the productive sectors, agricultural value-
added growth declined from 6.7 percent in 1970-1972 to -0.4
percent in 1973-1980. For these same periods, declines in
growth rates were recorded for services from 9.2 percent to
-0.08 percent, manufacturing from 6.7 percent to -4.2
percent, construction from 6.0 to -10.6 percent, and mining
from 13.2 percent to 1.3 percent, respectively (Singh,
1995; IMF, 1998).
The external account reflected the production crisis
described above. Between 1972 and 1980, the accumulated
deficit on the balance of payments was US$679.2 million,
compared to an accumulated surplus of US$95 million between
1960 and 1971 (Jamaica, Bank of Jamaica, 1985) . Further,
the deficit on the current account increased almost three¬
fold from US$570.2 million in 1962-1970 to US$1,620.0
million in 1972-1980, and the capital account became
negative in 1977, signaling capital flight. At the same
time, the dwindling net capital inflows could not match the
current account deficit. This depleted international
reserves, which plummeted sharply from US$103.1 million in
1972 to minus US$461 million in 1980 (Sharpley, 1984). The

5
foreign exchange shortage harmed the economy given its
relative openness and dependence on foreign intermediate
imports.
Finally, the inflation rate, which was 4.2 percent per
annum over the 1960-1970 period, recorded double-digit
rates throughout the following decade. Domestic prices, as
measured by the consumer price index, rose sharply,
recording an annual increase of 25 percent and 47 percent
over the 1974-1975 and 1978-1979 periods respectively
(Sharpley, 1984). By the late 1970s economic conditions had
deteriorated to such an extent that, when combined with
manifest signs of social distress (such as a dramatic
increase in violent crimes and emigration of skilled
workers), they generated a sense of instability and crises
(Stephens and Stephens, 1986; Kaufman, 1985).
The crises indicated a need for policy changes, and it
appears that the government was thinking along those lines
as reflected in a Stand-by Agreement it negotiated with the
IMF just before the December 1976 elections. The Agreement
included, inter alia, a wage freeze, fiscal restraint and a
currency devaluation of 20 to 40 percent (Boyd, 1988).
However, in the December 1976 election the PNP was returned
to office and the new administration promptly rejected the
IMF Stand-by Agreement as being inconsistent with the

6
mandate obtained in the recent election. Instead, and as a
way of avoiding agreements with the IMF (Boyd, 1988), and
after significant slippages in the external and domestic
accounts (Sharpley, 1984), the government implemented its
own adjustment policy measures in January 1977. These
included (Boyd, 1988)
(1) The reversal of the previous wage indexation policy;
(2) Implementation of import and exchange controls, a
dual exchange rate system and foreign exchange
rationing;
(3) Suspension of foreign debts for 18 months; and,
(4) A search for funding from sympathetic governments.
However, it became increasingly clear to the
government that in order to attract foreign capital some
formal agreement with the IMF was necessary. In this
regard, the government signed three agreements with the IMF
between August 1977 and June 1979. The first was suspended
after four months for failure to meet the stipulated
quarterly tests. The second, signed in May 1978, was
pursued in all its aspects by the government for a year.
Although economic performance under this agreement was poor
(Sharpley, 1984; Boyd, 1988), it was re-negotiated in June
1979. However, the agreement collapsed in December 1979

7
after the government failed to meet the program's
performance criteria.
In 1980, in light of increasing social unrest,
continued slippages in the fiscal deficits and external
accounts, severe foreign exchange shortages, and rising
inflation (Sharpley, 1984; Boyd, 1988; Thomas, 1999), Prime
Minister Manley called a general election so that the
nation could decide on a path of economic development and
"what part the IMF should play; or whether it should play
any part at all" (Jamaica, API, Manley, 1980) . Dubbed "the
IMF election", the October 1980 election was won by the
strongly pro-IMF JLP, led by Edward Seaga. Expectedly,
therefore, both the IMF and the government began
negotiations on funding and policies aimed at ameliorating
the country's economic and social crises.
The new administration relied heavily on funding from
external sources. Several loan agreements were negotiated
with the leading collaborating lending institutions, the
IMF, World Bank, IDB and USAID which have developed a
system of cross-conditionalities on their loans, given
their basic set of shared objectives on economic policies
and on Jamaica (Anderson and Witter, 1994).
The prescription by the IMF and the World Bank
(Fund/Bank) on how best to deal with the economic crisis in

8
Jamaica was a series of economic policy reforms. These
reforms went beyond the expenditure-switching and
expenditure-reducing policies which macroeconomic policy
would normally prescribe under these circumstances, to
include a built-in bias in favor of a more liberal economic
system. The latter aimed at, inter alia, fiscal discipline,
liberalizing the domestic market and the external trade
sector, privatizing of State Owned Enterprises (SOEs) and
other social services, and generally, greater reliance on
market signals to allocate resources. Over all, between
1981 and 1992, IMF financial support totaled US$1,036.9
million (under 10 agreements) , and that of the World Bank
US$360.4 million (under 8 agreements).
1.2 Problem Statement and Study Justification
The cumulative effect of the reforms mentioned above
is that the Jamaican agricultural sector is now expected to
operate within an economic framework that differs vastly
from that of the 1960s and 1970s. Historically, the sector
has played an important role in the economy. Agricultural
output to GDP ratio is about eight percent, agricultural
food export in total export is approximately 20 percent,
and agriculture's share of the total labor force exceeds 26
percent (Davis et al., 1999). Given the importance of
agriculture in the economy, how the sector performs in this

9
new policy environment demands urgent attention. In this
study, agricultural sector performance is evaluated in
terms of crop supply responses. Quantitative analyses of
the impact of these reforms on Jamaican agricultural supply
responses are both scarce and generally not particularly
robust. In addition, the literature on similar economic
reforms in other countries suggests that the outcomes of
these reforms are ambiguous, which justifies more empirical
work (Khan, 1990) .
Additionally, this study is justified on the following
grounds. First, the economic reforms have generated major
changes in the long-term prospects for the national economy
within which agriculture will operate. After two decades a
good set of data on these policy reforms is now available
which should facilitate empirical research on the impact of
these reforms on agricultural crop supply responses. This
evaluation can throw light on courses of action which
policy makers in Jamaica and elsewhere can consider as they
seek to reactivate and sustain long-term growth in the
agricultural sector.
Second, while a few studies have analyzed specific
aspects of the impact of the economic policy reforms on
Jamaican agriculture (Singh, 1995; Anderson and Witter,
1994; Newman and Le Franc, 1994; and Brown, 1994), what is

10
lacking is an over-all analytically rigorous assessment of
agricultural supply responses within the context of the new
policy framework as defined by the recent economic reforms.
Finally, analysis of the impact of these reforms on
agricultural supply responses is both timely and urgent in
light of an on-going debate. A number of writers in the
Caribbean and elsewhere have been advancing the view that
the kinds of economic reforms implemented in Jamaica and in
other countries have harmed agriculture (Anderson and
Witter, 1994; Green, 1989; Bates, 1989). In contrast a
recent series of studies sponsored by the World Bank
suggests that policy reforms similar to Jamaica's should
boost agriculture in countries which have traditionally
discriminated against the sector (Schiff and Valdés, 1992a;
Krueger, 1992).
1.3 Objectives of the study
The general objective of this research is to evaluate
the impact of economic reforms on selected agricultural
crop supply responses in Jamaica over the 1980-1999 period.
The specific objectives are to provide
(1) A statistical evaluation of the performance of the
agricultural sector in the pre-reform (1962-1979)
and reform (1980-1999) periods.

11
(2) A descriptive, diagnostic and statistical analysis
of the economic reforms that have taken place over
the past two decades within the agricultural sector.
(3) An econometric (modeling) evaluation of selected
agricultural crop supply responses over the 1962-
1999 period, with the number and types of crops
chosen for study to be determined by data
availability.
(4) A comparative analysis of agricultural crop supply
responses in the reform period with a counter-
factual, the latter defined as crop supply responses
that would have resulted in the absence of the
reforms.
1.4 Hypotheses
The following hypotheses are advanced in the study:
(1) Structural reforms significantly improve agri¬
cultural supply responses.
(2) Sustained improvements in agricultural supply
responses require further broadening and deepening
of the reform process.
1.5 Conceptual Issues
Agriculture features prominently in policy discussions
when (1) it is believed that agricultural policies and
institutions help precipitate an economic problem which

12
policy makers are addressing; and (2) when agriculture is
an important contributor to employment, GDP, export
earnings, domestic food supply, and revenue for the
government (Binswanger and Deininger, 1997) . Various
publications by the government of Jamaica suggest that both
of these factors have featured prominently in the recent
reforms, while (2) was a major consideration in the pre¬
reform (1962-1979) period. Consequently, the conceptual¬
ization is based on two perspectives that are believed to
have influenced the design of the policy reforms:
(1) The Jamaican government perspective. From numerous
government publications, it would appear that this
perspective is clustered around three main goals:
(a) increasing food supply; (b) developing rural
areas; and (c) increasing agriculture's contribution
to the over-all economy, through, inter alia,
employment, contribution to GDP, and export earn¬
ings; and,
(2) The IMF/World Bank perspective, whose focus on the
agricultural sector clusters around (a) getting
prices right; (b) improving efficiency; and (c)
rationalizing public expenditure in the agricultural
sector.

13
These two perspectives are not mutually exclusive. On
the contrary, the choice of policies in a particular
program package results from extensive discussions between
officials of the IMF/World Bank and the Jamaican
government. The program policy mix therefore reflects the
particular economic situation in the country as well as the
preferences of the government (Khan, 1990). Despite this, a
review of the literature on Jamaican agricultural policy
formation suggests that since the late 1970s the influence
of the IMF/World Bank perspective has exceeded the Jamaican
government's.
Given the two perspectives mentioned above, and a
review of the content of the reforms since the late 1970s,
the evidence suggests that the reforms have focused on
three inter-related issues of relevance to the agricultural
sector: (i) agricultural (inter-sectoral) terms of trade;
(ii) agricultural growth, its adjustment, and supply
response; and (iii) efficiency in the agricultural sector
(Nallari, 1992; Singh, 1995; World Bank, 1996).
Figure 1.1 conceptualizes the Jamaican agricultural
sector, dividing the factors impacting agriculture into
external (e.g., world prices) and internal. The latter can
be separated into those that are exogenous (weather,
terrain) and those that are policy induced. Another useful

14
Policy Instrumenta:
PI. Agriculture Projects:
infrastructure, land
development, mechanization,
crop improvement
P2. Rural industries
P3. Monetary Policies: interest
rates, exchange rates.
P4. Tax Policies: income,
indirect, customs duties.
P5. Public consumption, rural
invest.
P6. Rural health, education,
welfare.
P7. Inport/Export controls,
tariffs, subsidies.
P8. Price controls, input
subsidies.
P9. Agriculture credit
P10.Marketing: infrastructure,
information.
t
i
Agriculture Sector
Performance Indicators:
1. Economic Outcomes
(a) quantity flows
- supply response
- agri. in GDP
- agri. growth rate
(b) agri. multi-factor
productivity
(c) relative prices
(TOT, RER) .
2. Accumulation
Indicators
- coefficient of
investment
- capital formation
- agri. employment
- agri. tax revenues
- rural nutrition
- farm incomes
J
Figure 1.1: Conceptualization of the Jamaican Agricultural
Sector.

15
distinction is between prices and non-price factors
(including exogenous shocks) that influence agricultural
output and income. At the micro level, the price variables
include prices of the particular crops, prices of
alternative crops, input prices, and the general price
level. Government policies exercise both direct and
indirect influence on agricultural prices and thus on
agricultural performance. Direct (sectoral) policies such
as price ceilings, guaranteed prices and trade
taxes/subsidies provide incentives to shift resources among
crops and sectors of the economy (Binswanger, 1989).
Government policies also can influence output indirectly
through macroeconomic (or economy-wide) policies. The most
critical components in this regard are fiscal policies and
exchange rate policies (Mamingi, 1997).
The terms of trade as a policy variable turns on the
hypothesis that if all prices were determined in markets,
and effective rates of taxation were equalized across
commodities, then agriculture's supply responses, growth,
income, and contribution to GDP would be higher compared to
situations of negative effective protection of the sector
(Schiff and Valdés, 1992b). A negative effective protection
of agriculture arises when:

(1.1)
16
I [fe - Pi') + scj < 0
where P is a general price deflator, Pi denotes the producer
price of crop i, P/ is the free-market price for crop i,
and 5 is a subsidy to producers as a proportion of cost, Ci.
For many developing countries, agricultural pricing
policies have consistently kept Pi below p/, and subsidies
have not sufficiently counterbalanced this disparity
(Schiff and Valdés, 1992b). Given severe budgetary
constraints, to achieve at least zero effective protection,
the policy implication of (1.1) is to raise the real
P.
producer price, —, or to liberalize all markets thereby
P
eliminating the disparity between Pi and P£. In effect,
improving agriculture's terms of trade.
The exchange rate sets an upper limit on agricultural
export earnings and, when combined with input taxes and
subsidies, affects input prices and competing agricultural
imports. Changes in the real exchange rate can affect
agricultural output and growth by altering the terms of
trade between agriculture and non-agricultural sectors.
Further, policies that lead to over-valuation of the
exchange rate can adversely affect export crops, encourage

17
rent-seeking activities and generate unproductive uses of
resources (Jaeger, 1991).
In addition to the price policies, non-price factors
constitute an important influence on agricultural output.
These factors reflect the material conditions of production
in the agricultural sector. Some of these factors include,
inter alia, government expenditure and investment in the
agricultural sector, construction of rural infrastructure
(roads, drainage and irrigation works), extension services,
rural credit institutions, and dissemination of relevant
scientific information to farmers. The aim of agricultural
sector-specific programs is to find a mix of policies for
increasing efficiency and productivity in the sector. It is
assumed that these two outcomes are necessary (if not
sufficient) conditions for increasing agricultural output
and growth by reducing costs of production and/or
increasing product prices. These policies focus on using
public investment in agriculture efficiently, reducing
marketing and transport bottlenecks, improving agricultural
extension services, health and education (capital formation
in agriculture), rationalizing input prices, and enhancing
the efficiency of parastatals in the agriculture sector.
Behrman (1990) has advanced an important perspective
on the channels through which policies affect performance

18
indicators. He argues that analysis of the impact of policy
reforms on performance indicators must explicitly consider
the conduits through which the effects of the policies are
transmitted to the observed (or desired) outcome. These
"meso-level" variables, identified as markets (product,
input and financial) and infrastructure (economic and
social), are the interface between the policies (sector-
specific and economy-wide) and targets. In Figure 1.1, for
example, various policies, Pi, i = 1,2,3,...10, are
identified. Some are macroeconomic (e.g., monetary policies
and exchange rate adjustments), whereas others are sector
specific (e.g., agriculture credit, tariff reductions in
the trade sector and elimination of price controls in
agriculture). These policies are then combined to achieve
specific targets, either at the sectoral or macroeconomic
levels.
Behrman (1990) emphasizes, however, that the meso-
level variables condition the effectiveness of policies on
the target variables and that analyses of policy impacts
must also evaluate how policies have influenced these meso-
level variables. For example, if poor transportation, lack
of effective irrigation, or in-efficient research and
extension services restrain farmers' responses to higher

19
prices, then improving these meso-level variables may do
more for the farmer than price increases (Chhibber, 1989).
The influence of the factors mentioned above on
agricultural supply responses has been well documented in
the empirical literature. However, the recent economic
reforms have had an enduring effect on these factors. This
points to an important issue raised in the literature,
viz., how to measure the magnitude of reforms (IDB, 1997).
Economic statistics such as exchange rate differentials,
inflation rates, tax changes and so on deal with outcomes
rather than the policies that gave rise to them. In order
to address this issue, structural policy indexes have been
constructed by Lora (1997) and others.
The index by Lora (1997) was constructed for twenty
countries in Latin America and the Caribbean. This index
seeks to measure the extent of market freedom accorded to
economic policies in areas of trade, tax, finance,
privatization and labor. In each of these areas indices of
market freedom are identified. For example, in trade policy
the indices are average tariffs and tariff spreads; in tax
policy the indices include, inter alia, tax rates on
companies and on individuals; and on financial policy,
indices include freedom of interest rates on deposits,
loans, reserves on bank deposits, etc. (IDB, 1997). The

20
structural policy index is a simple average of the indices
in the five areas. The index can range from 0 to 1, based
on the worst and best observations respectively, on market
freedom in the country. Further, the index is an important
indicator of the extent to which countries are departing
from past ways of operating their economies. Jamaica's
structural policy index has shown continuous movement
towards market freedom, increasing from 0.426 in 1985 to
0.684 in 1995. These index values are higher than the
average reported for the twenty Latin American and
Caribbean countries in the sample (Lora, 1997) .
1.6 Procedural and Methodological Considerations
In approaching the general and specific objectives of
the study, critical analysis of the Jamaican agricultural
sector's performance is first undertaken. To achieve the
general objectives and, more specifically, objectives (2) ,
(3) and (4), requires four specific tasks. The first task
is to identify and evaluate the structural reforms
undertaken since the late 1970s. Both the stabilization and
the structural reform policies are evaluated according to
their (1) theoretical bases; (2) breath of vision; and (3)
logical consistency. Second, supply response models are
developed and estimated. Finally, specific objective (4) is

21
addressed within the context of simulating an alternative
path to that of the reforms.
Any appraisal of the reforms must compare their
impacts not to the pre-reform indicators but rather to some
specified, hypothetical alternative. This requires some
simulation exercises. The idea here is to generate
simulated time series over the reform period for the
performance indicators. The simulated series then
constitute the counterfactual to which the actual outcomes
in the reform period are compared. For the counterfactuals,
at least three scenarios appear to be logical extensions.
Scenario 1 assumes that the policies pursued over the pre¬
reform period continued into the reform period, i.e., that
the reforms over the 1980-1999 period were not instituted.
Scenario 2 assumes policies based on the Jamaican
government's critique of the IMF/World Bank programs were
implemented. Finally, Scenario 3 assumes that IMF/World
Bank assistance came with no conditionalities. These
components form the basis of the data series for the
simulation exercise.
The three scenarios appear logical and plausible
options given (1) the initial defiance by the government of
the mandated reforms in the late 1970s; (2) the frequent
failures to meet conditionality tests because the reforms

22
were not implemented; and (3) government's own protest¬
ations over the years of what are considered "acceptable"
reforms. While numerous instances of disagreements exist
between the Jamaican government and the IMF/World bank with
respect to appropriate reform policies, the clearest
statement to this effect is captured in the following
quotation:
Even though lender and borrower therefore shared the
common concern regarding the need for economic reform,
there is still much debate about appropriate social
policies and about the specific effects of
manipulating major economic variables such as exchange
rates and interest rates. This was illustrated by the
lengthy and tortuous negotiations between the
government of Jamaica and the IDB and World Bank on
the 1989-90 Agricultural Sector Adjustment Loan.
(Anderson and Witter, 1994, p.14)
1.7 Summary
It is generally believed that developing countries
have historically discriminated against their agricultural
sectors in favor of industrial development. It is further
felt that policies that reverse that discrimination should
boost agricultural output and income. Over the past two
decades far-reaching economic reforms have been undertaken
in Jamaica leading to a new economic framework within which
the agricultural sector must now operate. This study seeks
to assess the impact of these reforms on agriculture crop
supply responses, an undertaking that is both timely and

23
urgent given the importance of the sector in the economy
and the absence of rigorous analytical studies on the
Jamaican experience with these reforms.

CHAPTER 2
ECONOMIC STRUCTURE, GROWTH, AND POLICY REFORMS IN THE
JAMAICAN ECONOMY, 1962-1999
It is useful, at the outset, to situate the discussions
in this chapter against the backdrop of three distinct
phases of economic policy that characterize the economic
history of Jamaica over the period 1962-1999. In the first
phase, 1962-1972, the Jamaica Labor Party (JLP) emphasized
free markets. However, Bonnick (1984) argues that despite
the rhetoric of economic liberalism, in reality the
government pursued an import-substitution (IS) strategy
based on protectionism, trade restrictions, and price
controls.
In the second phase, 1972-1980, the People's National
Party (PNP) under Prime Minister Michael Manley pursued
economic populism and state directed control (dirigisme) in
an effort to build democratic socialism. This phase was
marked by extensive government intervention, which included
nationalization of major industries, price controls, and
subsidization of basic foods and some agricultural imported
inputs. These policies made significant demands on national
resources and created a large bureaucratic economic
24

25
structure. Consequently, the government was unable to
respond to the crisis in the world economy in the mid-
1970s, and was forced to approach the IMF for stabilization
funding in 1977.
The third phase, 1980-present, is directly linked to
the policies of the previous phases, particularly the 1977
IMF stabilization funding. Phase three can be viewed as
having two sub-phases. The first is the return of the JLP
to government in 1980-1989 and the implementation of
various Fund/Bank stabilization and structural adjustment
programs. The second, 1989-present, marks the return to
government by the PNP, the continuation of the market led
policies of the preceding JLP administration, but more
importantly, the intensification of the country's commit¬
ment to liberalization of the economy. The rest of this
chapter is organized as follows: Section 2.1 provides an
overview of the structure of the economy and its growth
performance. Section 2.2 highlights trends and policies in
the agriculture sector. Sector 2.3 describes the economic
reforms, and Section 2.4 provides a summary of issues
raised in this chapter.

26
2.1 Structure of the Economy and Economic Growth:
An Overview
The relative contribution of economic sectors to GDP in
Jamaica has not changed much over the past two decades. The
data presented in Table 2.1 and Figure 2.1 show that
agriculture's share has been fairly stable, ranging from an
average of 8.4 percent over the period 1969-197 9 to 8.0
percent over 1990-1998. Although agricultural contribution
to GDP is lower than that of the other sectors, these
comparatively small percentage contributions are deceptive.
Davis et al. (1999) have shown that the sector plays an
important part in the country's employment, food
production, and foreign exchange earnings.
The industrial sector, which includes mining,
manufacturing and construction, contributed on average 39.1
percent to GDP in the period 1969-1979 but declined to 37.3
percent in 1990-1998. The manufacturing sub-sector also
shows a decline from 18.4 percent in 1969-1979 to 16.3
percent in 1990-1998. Finally, services constitute the
largest sector of the Jamaican economy, accounting for over
50 percent of GDP over the period 1969-1998. Its
contribution to real GDP has increased from 48.5 percent in
1969-1979 to 54.7 percent in 1990-1998.

27
Table 2.1: Sectoral Contribution to Real Gross Domestic
Product (Period Averages, Percentage).
1969-79
1980-89
1990-98
1969-98
Agriculture
8.4
6.4
8.0
7.6
Industry
39.1
34.5
37.3
37.0
Manufacturing
18.4
20.9
16.3
17.1
Services
48.5
59.1
54.7
53.9
Source: Computed using data from Jamaica, Planning
Institute of Jamaica (various issues).
Note: The 1998 data used were preliminary estimates.
1969 1973 1977 1981 1985 1989 1993 1997
Years 1969-1998
—•—Real GDP —♦—Agri. —«—Indus. a Serv.
Figure 2.1: GDP and Sectoral Contribution to GDP in
Constant 1995 Dollars.
Real sectoral and GDP growth rates are estimated and
shown in Table 2.2. The estimation uses a log-linear model,
Yt = oto + P*Time + et , and corrected for autocorrelation
whenever it exists. Over the 1969-1998 period, growth rates
were positive but consistently low, while sub-period growth
reveal mixed results. GDP growth rates were just over two

28
Table 2.2: GDP and Sectoral Growth (1995=100)
(Percentage) .
1969-79
1980-89
1990-98
1969-98
GDP
2.3a
2.4C
-0.2
1.1*
Agriculture
-0.3
1.1
2.3d
0.6b
Industry
0.9
4.3C
-3.8a
00
o
Services
4.6a
0.1
Q
CM
1.6a
Source: Computed using data from Jamaica, Planning
Institute of Jamaica (various issues).
Note: The 1998 data used were preliminary estimates.
a, b, indicate statistical significance at the five
and 10 percent levels, respectively.
percent in the 1969-1979 and 1980-1989 periods. Services
growth was reduced from 4.6 percent over 1969-1979 to 2.4
percent over 1990-1998, and was insignificant over the
1980-1989 period. Industrial growth, which was 4.3 percent
in the 1980-1989 period, declined significantly to -3.8
percent over the 1990-1998 period. Agriculture growth over
the 1990-1998 period was 2.3 percent, compared to its
insignificant growth in the two previous sub-periods. Over
the entire 1969-1998 period agriculture and industry
recorded less than one percent growth, while GDP and
services grew just over one percent, respectively.
These periodic growth rates mask the highly volatile
annual growth rates in these economic series as shown in
Figures 2.2 and 2.3. Over the period 1969-1998, annual
growth rates for GDP ranged between -9.0 and 13.0 percent,
and were negative in 12 of the 30 years. For the

29
Figure 2.2: GDP Growth (Percentage Change Over Previous
Year, 1995-100).
1970 1974 1978 1982 1986 1990 1994 1998
Years 1969-98
—♦—Agri. Growth —■—Ind. Growth -a Ser. Growth
Figure 2.3: Growth of Total Agriculture, Industry and
Services (Percentage Change over Previous Year, 1995=100).
1969-1998 period, agriculture growth rates ranged between
-12.6 and 29.3 percent, and were negative in 15 years.

30
Similar annual volatilities characterize the industrial and
services growth rate series.
2.2 Output Trends and Policies in
Jamaican Agriculture
The growth rates of volume indexes for total
agriculture and broad aggregates (food, crops, livestock
and cereals) and graphs for these series are shown in Table
2.3 and Figures 2.4 and 2.5. With the exception of the
cereals index, which declined over 1971-1998 at an annual
rate of -3.2 percent, the other indexes grew but at less
than two percent over the 1971-1998 period.
Table 2.3: Growth Rates of Total Agricultural and
Broad Agricultural Aggregate Indexes.
1971-79
1980-89
1990-98
1971-98
Total Agriculture
1.6a
1.7
1.9a
1.7a
Food
1.6a
1.6
1.9a
1.7a
Crops
2.6b
1.5a
-3.6
1.7a
Livestock
1.6a
-16.5
0.7
1.3a
Cereals
16.9a
-9.0
-4.5b
-3.2b
Source: Computed using data from Food and Agriculture
Organization, FAO Production Yearbook (various issues).
a, b, indicate statistical significance at five and ten
percent levels, respectively.
The composition of agricultural output is shown in
Table 2.4. Export agriculture, domestic agriculture, and
livestock, forestry and fishing constitute the aggregative
components of agriculture in GDP. The data in Table 2.5 and
the graph in Figure 2.6 show that domestic agriculture

31
Years 1971-98
—♦—Total Agriculture
Figure 2.4: Total Agricultural Output Index (1995=100).
o
o
H
H
m
o>
o>
X
•o
c
H
+J
3
A
•U
3
O
1971 1976 1981 1986 1991 1996
Years 1971-1998
Crops ♦ Livestock —«—Cereal —•—Food
Figure 2.5: Output Indexes for Broad Agricultural
Aggregates (1995=100) .

32
Table 2.4: Composition of Agricultural Output (J$M,
1995=100), Selected Years.
Agriculture
and Sub¬
sectors
1969
1975
1980
1985
1990
1995
1998
Total
Agriculture
95.9
105.9
101.3
72.2
104.4
154.4
120.4
Export
Agriculture
34.3
25.9
14.8
13.9
13.6
15.4
12.3
Sugar
Cane
22.6
17.7
9.9
7.8
9.2
9.4
7.9
Other
Main
Exports
11.6
8.2
4.8
6.1
4.4
5.9
4.4
Domestic
Agriculture
33.0
47.1
56.9
38.0
66.2
114.5
82.8
Root
Crops
14.4
26.9
25.3
14.6
35.1
62.8
42.7
Other
Primary
Products
18.6
20.2
31.7
23.4
31.1
51.7
40.1
Livestock,
Forestry &
Fishing
28.6
32.9
29.5
20.3
24.6
24.5
25.3
Source: Jamaica, Planning Institute of Jamaica (various
issues).
Table 2.5: Agriculture Sub-sectors as a Percentage of Total
Agriculture (Period Averages) .
Agriculture Sub-sectors
1969-79
1980-89
1990-98
1969-98
Export Agriculture
23.47
(5.50)
15.95
(2.24)
11.21
(1.46)
7.29
(6.26)
Domestic Agriculture
44.78
(5.32)
57.21
(6.35)
68.79
(3.62)
56.13
(11.17)
Livestock, Forestry &
Fishing
31.73
(1.35)
26.85
(4.49)
20.03
(2.50)
26.59
(5.65)
Source: Computed using data from Jamaica, Planning
Institute of Jamaica (various issues).
Note: Figures in parentheses are standard deviation.

33
Figure 2.6: Sub-sectors as a Proportion of Agriculture.
constitutes the largest proportion of total agriculture in
the 1969-1998 period and other sub-periods. In 1969-1979
the domestic agriculture sub-sector averaged 44.8 percent
in total agriculture and increased to 68.8 percent in 1990-
1998. Both of the other two sub-sectors, export agriculture
and livestock, forestry and fishing, have declined over the
period 1990-1998 compared to the two previous decades. In
terms of growth rates, Table 2.6 shows that domestic
agriculture and its sub-groups are the only sub-sectors
within the agri-sub-sector that have positive growth rates
over the 1969-1998 period. This has compensated somewhat
for the negative and low growth in the other agriculture
sub-sectors.

34
Table 2.6: Growth Rates of Agricultural Components
(Percentage).
Agriculture Sub-sectors
1969-79
1980-89
1990-98
1969-98
Export Agriculture
-4.4a
-1.3d
-4 . 1D
-2.1a
Sugar
-4.2a
-2.6a
-5.4a
-2.0a
Other main exports
-4.6a
1.3
1.0
-2.7a
Domestic Agriculture
1.8
11.2a
25.6a
2.6a
Root crops
-2.7
19.7a
-1.2b
2.6d
Other domestic crops
4.0a
3.4b
Q
rH
1
2.7a
Livestock, forestry, &
fishing
-1.5
-3.3a
0.1
-1.7a
Source: Computed using
data
from Jamaica,
Planning
Institute of Jamaica (various issues).
a, b, indicate statistical significance at the five and ten
percent levels, respectively.
The stagnation of agriculture in Jamaica, which can be
inferred from the growth rates in Table 2.6 reflects, to a
large degree, the state of the agricultural sectors in many
less developing countries (LDCs). A plausible explanation
for this, is that the development literature in the 1950s
and 1960s viewed agriculture as a static sector, from which
resources could be shifted to promote industry, considered
as the dynamic sector. To a large extent, this view has
influenced the kinds of agricultural policies that have
been pursued in the past by LDCs. In this regard, Schiff
and Valdés (1992a, p.59) state:
In many developing countries, the high rate of
agricultural taxation has been part of an explicit or
implicit policy of industrialization-led growth,
justified in part by the belief that industry was the
dynamic sector while agriculture was static and not
very responsive to incentives. That means that
economic growth could be accelerated by shifting

35
resources from agriculture to industry, by taxing
agriculture directly, and by protecting the industrial
sector.
Many governments in the developing countries have
intervened in their agricultural sectors, both directly
through agricultural sector policies, and indirectly,
through economy-wide policies such as industrial
protection. Direct interventions take numerous forms. Some
of these include procurement measures (e.g., government
marketing boards as sole buyers of agricultural output, and
suppliers of major agricultural input); quotas and direct
taxation on various agricultural export crops; subsidies on
farm credit and farm inputs; and quantitative restrictions
and tariffs on imported agricultural imports. While some of
the direct interventions have benefited agricultural
producers, some are tantamount to an implicit tax on
agriculture, depressing farmgate prices and farm incomes
below levels that would otherwise prevail (Schiff and
Valdés, 1992a). Indirect forms of interventions affect
agricultural production incentives via macroeconomic
policies (e.g., overvaluation of the exchange rate) and
industrial protection policies (Krueger, 1992) .
Various measures have been employed by the Jamaican
government to extract surpluses from the agricultural
sector. These include taxing agricultural exports,

36
controlling farmgate prices by the state marketing boards,
over valuating exchange rates and reducing internal
agricultural terms of trade relative to manufacturing.
Studies by Gafar (1980, 1997) and Pollard and Graham (1985)
strongly support this proposition. Table 2.7 shows that
Table 2.7: Average Rates of Growth of Farmgate and F.O.B.
Prices, Output and the Nominal Protection Coefficient (NPC)
for the Period 1970-1978, Jamaica.
Commodities
Growth
^ates (%)1970-78 a
Farmgate
Prices
F.O.B.
Prices
Output
Nominal Protection
Coefficient b
Sugar Cane
-1.25
4.13
-2.90
0.82
Banana
5.06
7.77
-7.43
0.77
Cocoa
1.64
14.06
-0.54
0.72
Coffee
9.80
8.34
2.02
0.61
Coconut
7.06
4.67
-21.00
n. a.
a Adapted from Pollard and Graham (1985), Table 2.
b Based on the data in Pollard and Graham (1985), Table 4
for 1970-1979. The NPC is defined as the ratio of farmgate
prices (Pf) to F.O.B. prices (Pw) received minus marketing
and processing costs (C): NPC = PF / (Pw - C) .
over the 1970-1978 period, the nominal protection
coefficient (NPC), for Jamaican agriculture was less than
one, indicating the extent to which the sector was taxed.
For example, a NPC of 0.82 means that the commodity is
taxed at a rate of 18 percent. Subtracting one from the NPC
gives the nominal protection rate, NPR, which, according to
Table 2.7, is negative, suggesting that producers of the
crops reported were not supported, but were taxed instead.

37
It should be emphasized also, that developing countries
have a tendency to overvalue their domestic currencies, so
that the official exchange rates used to calculate the NPC
overstate internal prices. Hence government taxation of
agriculture is even greater than is usually captured by the
NPC (Gardner, 1987).
More recent data compiled for this study also support
the proposition that Jamaican agriculture has been taxed.
Table 2.8 reports on the net barter and income terms of
trade between agriculture and manufacturing for Jamaica.
The net barter terms of trade, N, is defined here as
P
N = —, where PA and PM are the price indexes of agriculture
PM
and manufacturing, respectively. Values of N greater than
one indicate that prices of agricultural commodities have
risen relative to those in manufacturing (Tsakok, 1990).
The estimates in Table 2.8 reveal that over the 1966-1978
and 1989-1999 periods, the net barter terms of trade were
unfavorable to Jamaican agriculture, but were favorable
over the decade 1979-1988.
Although the net barter terms of trade may move
against agriculture, the sector can still increase its
purchasing power if agricultural output increases pro¬
portionately more than the decrease in agricultural prices.

38
The purchasing power of agriculture is captured in the net
P
income terms of trade, I, estimated as: I = — * QA, where Qa
is the agricultural output index. The index I is therefore
the net barter terms of trade adjusted for marketed
amounts. Increases in I indicate a rising purchasing power
of agriculture for buying manufactured goods (Tsakok,
1990). The data in Table 2.8 suggest that agriculture's
ability to purchase manufactured goods have peaked in the
1979-1988 period.
Table 2.8: Net Barter Terms of Trade and Agricultural
Income Terms of Trade (1980=100; Period Averages).
1966-78
1979-88
1989-98
Net Barter Terms of Trade
0.60
1.39
0.86
Income Terms of Trade
58.20
159.18
134.03
Source: Compiled by author.
2.3 The Economic Reforms
After the late 1970s, Jamaica undertook major changes
in economic policies, both sector-specific and economy¬
wide. These policy changes have been an integral part of
the conditionalities usually attached to program packages
by external funding agencies. The reforms vary in
intensity, both inter-temporally and in terms of the areas
in which policy reforms were enacted, such as foreign
trade, taxation, and liberalization measures. Underlying

39
the reforms has been a conviction by funding agencies that
freer markets allow for a more productive and efficient use
of scarce resources, a condition not necessarily accepted
uncritically by the Jamaican government. Nevertheless, a
number of market restrictions were eliminated, regulations
deemed necessary were simplified and made more transparent,
and private enterprises were encouraged. The most
conspicuous reforms were the liberalization efforts to
facilitate foreign trade and financing activities through
exchange rate and tariff adjustments.
The reform period, 1980-1999, has been punctuated by
successive IMF/World Bank stabilization and structural
adjustment programs, as well as programs financed by other
international funding agencies. The stabilization and
structural adjustment programs of the IMF and the World
Bank for Jamaica were developed as programs of policy
reforms, aimed at reversing, within a few years, the
economic crisis in the country. Three facets of this crisis
emphasized in the IMF/World Bank literature are structural
factors, external factors and poor domestic policies,
including an anti-agricultural bias in pricing policies.
However, it is a matter of public record that these
IMF/World Bank reforms have been resisted by the Jamaican
government at various times, and have been hotly debated in

40
the popular media regarding the social impacts of
externally imposed structural adjustment programs.
Despite this apparent dissatisfaction, the government
has utilized operational policies along the lines of those
prescribed by the Bretton Woods institutions. While in
principle the early reforms were aimed at stabilizing the
economy, from the outset there was a built-in bias in the
policies for a more liberalized economic system. For
example, the three year Extended Fund Facility (EFF), which
the Jamaican government signed with the IMF in May 1978,
contained specific conditions for exchange rate devaluation
as well as the liberalization of domestic prices (Boyd,
1988). More extensive economic reforms were advanced by the
new administration under Prime Minister Edward Seaga who
held office from 1980-1989.
Under the 1982 and 1983 structural adjustment
programs, quantitative restrictions (QR), on trade were
replaced by tariff equivalents, and by 1984 all QRs were
eliminated. Exchange rate devaluation was pursued in 1984-
1985, followed by the unification of the official and
parallel rates. At the same time, the government began to
reduce the public sector via the divestment of state owned
enterprises (SOEs).

41
The 1990s witnessed the most protracted reforms aimed
at deregulation and liberalization of the Jamaican economy.
The PNP, which took office in 1989 under Prime Minister
Michael Manley, had campaigned in the general elections on
a platform similar to the JLP, vowing to continue the JLP's
market oriented policies. This was an opportunity for a
fresh departure for the PNP, which was noted for its
previous support for democratic socialism and extensive
government intervention. Garrity (1996, p.52) argues that
...the Manley-led government made the decision in 1990
to truly embrace the market and proceed with economic
liberalization reforms. For Manley...[t] o continue the
reforms of the previous JLP government would not be
sufficient to bring about the needed structural
changes, nor were there sufficient resources to
continue the previous reforms....
Based on the limited available options, Manley's
commitment to the liberalization process represented a
turning point in state-society relations in Jamaica. In
particular, the administration sought an accommodative
pattern of governance between state and social actors by
energizing the private sector. In this sense, the
liberalization process can be viewed as "state-sponsored
disengagement from a command role in the economy ... and the
creation of an enabling environment for private sector-led
growth and development." (Garrity, 1996, p.52)

42
Against this background, the sectoral adjustment loan,
which the government signed with the World Bank in 1989,
focused on liberalizing the agriculture sector along the
following lines:
(1) Deregulation of the coffee and cocoa boards by 1990;
(2) Shifts in government focus from production and
markets to support services and infrastructure;
(3) Continuation of the divestment of agro-enterprises
(e.g., sugar) ;
(4) Elimination of subsidies on agricultural credit;
(5) Reduction of tariffs; and
(6) Elimination of the Generalized Food Subsidies;
As a result, by 1991 the generalized food subsidies
were eliminated, and complete liberalization of the
exchange market and credit interest rate were achieved.
Consequently, the overall liberalization process, since the
first agreement with the IMF in 1977, opened markets to
more competition and reduced the role of the public sector.
As would be expected, the conditions that accompanied
the loans from the Bretton Woods Institutions involved
reforms at both the macroeconomic and the sectoral levels.
Figure 2.7 is a schematic presentation of the kinds of
economic reforms and a partial list of specific tasks
undertaken since the late 1970s. For convenience, the

43
Overall
â–º
Economic Stability, Growth,
Objectives
Improved Living Standard
Specific
Changes
* Elimination
* Wage
* Devaluation
* Elimination
* Social
of govt.
restraint
* Exchange rate
of
safety net
marketing
* Reduced
liberalization
quantitative
program
board
subsidies
* Open capital
restrictions
* Land
* Removal of
* Cost
account
* Tariff
capability
price
recovery
* Banking
reductions
classificati
controls
* Tax changes
reforms
* Elimination
on
* Improve
* Expansion of
* Interest rate
of exports
* Soil
land tenure
agriculture
and credit
restraints
conservation
* Sell/lease
non-price
liberalization
* Rationaliz-
* Irrigation
state land
incentives
* Capital market
ing of
rehab.
* Divestment
* Elimination
Reforms
import-
* Planned
of govt.
of subsidies
* Creation of
licensing
rehab, of
assets
on interest,
Agriculture
system
the sugar
* New pricing
loans and
Credit Bank
* Elimination
Industry
system for
food
and National
of subsidies
* Increase
coffee.
* Introduction
Development
on imported
availability
cocoa,
of
Bank
foods.
of foreign
citrus and
Generalized
inputs
pimento and
Consumption
* Agro-21
more
Tax
program
appropriate
* Hillside
sharing of
project
export
earnings
Figure 2.7: Selected Economic Policy Reforms in Jamaica,
1977-98.

44
policy reforms are organized into five blocks: (1) Domestic
Deregulation; (2) Fiscal Reforms; (3) Monetary/Financial
Reforms; (4) Trade Liberalization and (5) Other Reforms.
Within each block, specific tasks are listed, reflecting
implicitly and explicitly policy changes, which seek to
enhance the role of market forces, encourage private sector
initiatives and reduce government regulation and
intervention in the economy. It is interesting to note that
in spite of intense criticisms regarding the reforms' lack
of success in producing the economic recovery predicted by
its proponents, the trend of economic reforms has not been
reversed in periods of economic stress. Such periods have
instead been met with the broadening and deepening of
reform efforts.
Table 2.9 shows selected economic statistics in the
pre-reform and reform periods in Jamaica. In the area of
trade, average tariffs in 1986 were 56 percent but fell to
11 percent in 1995. Exchange rate systems have been an
important policy instrument in Jamaica to establish
restrictions on capital outflows and restrictions for
repatriating export revenues and foreign exchange.
Following the exchange rate liberalization, many of these
restrictions have been dismantled. As evidence of the
process of exchange rate unification and deregulation,

45
Table 2.9 shows that the exchange rate differential (i.e.,
the difference between the average market price for foreign
exchange—inclusive of transaction costs and exchange rate
taxes—and the official rate) was 25 percent in 1986 but
fell to 5 percent in 1995.
Table 2.9: Outcomes of Economic Reform Policies in Jamaica,
1986 vs. 1995.
Maximum Tax Rate
Structural
Policy
Tariff
Reduction
Exchange
Differential
Companies
Individuals
Index, I
Net
larii
iver.
:fs %
Percent
Percent
Percent
0<]
[<1
1986
1995
1986
1995
1986
1995
1986
1995
1985
1995
56
11
25
5
45
32
50
25
0.402
0.684
Source: Adapted from IDB (1997
In the area of tax reforms, Jamaica adopted various
changes aimed at administrative simplicity and ease of tax
collection. Taxes for companies were reduced from 45
percent in 1986 to 32 percent in 1995. The maximum taxes
for individuals were reduced from 50 percent to 25 percent
over the same period. Finally, the structural policy index,
which is a summary statistic of the extent to which market
forces are allowed to operate in the economy, increased
from 0.402 in 1985 to 0.684 in 1995.

46
2.4 Summary
The evidence presented in this chapter suggests that
the direction of change of economic structure of the
Jamaican economy that began around 1980 continued over the
past two decades. Over the period 1966-1998, growth rates
of GDP and economic sectors were low but positive. With few
exceptions, sectoral growth rates in the reform period,
1980-1999, were higher than in the pre-reform (1962-1979)
period.
Some structural change within the agricultural sector
was discerned from the data. In particular, export
agriculture declined relative to other agri-sub-sectors.
Domestic agriculture recorded impressive growth, and it is
this agri-sub-sector's overall growth performance that
partially compensated for the negative and low growth in
the other agri-sub-sectors. Significant economic reforms
were implemented, and analysis of their impact on
agriculture will be undertaken in a later chapter.

CHAPTER 3
METHODOLOGICAL AND EMPIRICAL ISSUES IN ECONOMIC
REFORMS AND SUPPLY RESPONSE
The purpose of this chapter is four-fold. First, it
reviews the methods that have been used in the literature
to analyze the impact of economic reforms in developing
countries. Second, it reviews the literature on supply
response analysis. Third, it develops a crop supply
response model for Jamaica; and finally, it presents the
data sources for the analysis.
3.1 Review of the Economic Reform Literature
In this review of the economic reform literature two
sets of issues are addressed. First, attention is directed
at the way the economic reforms are conceptualized. Second,
the analytical methods used to evaluate the impact of the
reforms are examined.
One of the most challenging problems in capturing the
impact of economic reforms on target variables is how to
isolate the effect of each of the reforms undertaken. Since
the reforms were undertaken in various areas (trade
reforms, fiscal/monetary reforms, etc), at different points
in time, and with varying levels of intensities, their
47

48
effects on particular variables in the economy become
compounding and difficult to isolate. Khan (1990) argues
that it is theoretically and empirically difficult to link
all the policy reform measures to the ultimate targets of
the policies. Hence most studies have attempted to assess
the effects of the overall policy package on particular
target outcomes. This is the thrust of the studies by
Chadha et al. (1998), and some of the studies in Tshibaka
(1998), Campbell and Stein (1991), La Guerre (1994), Valdés
and Muir-Leresche (1993), Commander (1989) and Weeks
(1995). In these studies, neither the precise nature of the
underlying economic relationships, nor the specific
policies adopted are made explicit. Instead, the attention
is on whether or not the program package, which in effect
gives rise to a particular policy environment, has been
"effective" in the sense of achieving broad macroeconomic
objectives (Khan, 1990).
Most of
the
studies
reviewed
proceeded
from
the
premise that
the
economic
reforms
originated
from
the
conditionalities that accompanied Fund/Bank-supported
packages. In all country experiences the problem of
economic instability provided the initial imperative to
seek IMF assistance, which, when it came took the form of
stabilization policies. However, in most of the studies

49
reviewed the stabilization policies are treated en passant,
or not at all, and instead attention is focused exclusively
on economic outcomes associated with structural adjustment3.
For most developing countries that have been pursuing
the Fund/Bank economic programs, the last decade has been
one in which the reforms have intensified the process of
liberalizing the economic system. A frequently raised
question with respect to Fund/Bank-supported structural
adjustment programs cum structural reforms for economic
liberalization, has been whether these programs are
effective in achieving stated economic objectives. A second
and related question is how to measure the effects of these
reform policies on the target variables identified in the
programs. With respect to this latter question, Guitián
(1981) has argued that economic performance under a policy
package should be compared to a counterfactual. The latter
is defined as economic outcomes that would have taken place
in the absence of the program package. The concept of a
counterfactual is intuitively appealing and is a standard
yardstick widely used in economics to measure the impact of
policy interventions. Since the counterfactual is
Stabilization policies are designed to put the economy back on its
equilibrium path, whereas, structural adjustment policies aim at
putting the economy on a new (higher) equilibrium.

50
unobservable, it has to be estimated, hence alternative
methodologies used to evaluate the program effects are
judged in terms of their estimates of the counterfactual.
Although the preceding two questions have generated a
large body of literature over the past two decades, there
is little consensus in the economics profession either
about the impact of past programs on target variables, or
about how to estimate the effects of program packages
(Khan, 1990). With respect to measuring the impact of
reforms, there are three main approaches that have been
applied in the empirical literature. These are
(1) The before-after approach, which compares the
behavior of key macro-economic variables before and
during, or after, any particular reform period, or
policy package.
(2) The with-without approach compares the performance
of macro-economic variables of non-program countries
(the control group) with those from program
countries. A modified version of this approach is a
reduced form regression estimate that controls for
initial conditions in program and non-program
countries.
(3) The comparison-of-simulations approach. This
approach simulates performance of Fund/Bank-type

51
programs then compares them with simulated outcomes
from another set of policies.
Approaches (2) and (3) have been used extensively in cross¬
country studies to assess the impact of Fund/Bank-supported
programs. The before-after approach has been the most
prevalent in country case studies using time series data.
3.1.1 Before-After Approach
The before-after approach has been the most popular in
the literature on Fund/Bank-support programs. Reichmann and
Stillson (1978) were the first to use this approach in an
examination of 79 Fund/Bank-supported programs over the
period 1963-1972. They compared growth, balance of payments
and inflation in the two years prior to and after the
implementation of the program. Connors (1979) also used the
before-after approach to evaluate 31 programs in 23
countries over the period 1973-1977. More recent studies
include Singh (1995) and the papers in Le Franc (1994), on
Jamaica. Similar studies on other developing countries
include most of the papers in Tshibaka (1998), Campbell and
Stein (1991), La Guerre (1994), Valdés and Muir-Leresche
(1993), Commander (1989) and Weeks (1995).
The before-after approach is viewed by some analysts
as providing a relatively poor estimate of the counter-
factual to program effects by assuming that non-program

52
determinants of economic variables remain constant between
non-program and program periods. This assumption has been
questioned in the literature in light of the fact that non¬
program determinants of economic outcomes (e.g., terms of
trade variations, changes in international interest rates,
weather, etc.) typically change year after year. Conse¬
quently, Goldstein and Montiel (1986) have demonstrated
that the before-after estimates of program effects are
biased, since all economic changes in program period are
incorrectly attributed to program factors. These authors
suggested further that the before-after estimates are
unsystematic over time, since in any particular year
program effects will often be dominated by non-program
factors. For example, if a hurricane damages infra¬
structural works and agricultural crops, then agricultural
growth may decline, causing all programs in that year to
appear to have performed poorly.
3.1.2 With-Without Approach
The with-without approach seeks to overcome the
shortcomings of the before-after approach by comparing
economic outcomes between program and non-program
countries. Since both groups of countries are subjected to
the similar external environments, it is argued that this
comparison cancels out the non-program determinants, so

53
that any observable differences in outcomes in the two
groups of countries are attributable to the Fund/Bank-
supported programs. In other words, the economic outcomes
observed in non-program countries are taken as the
counterfactual of what would have occurred in program
countries in the absence of Fund/Bank-supported programs.
Donovan (1981, 1982) was the first to use this
approach on a sample of Fund/Bank-supported programs
implemented between 1970-1980. Loxley (1984) also applied
this approach to 38 less developed economies (income of
$690 or less) with Fund/Bank program during 1971-1982.
Other similar studies include Gylfason (1987) and Pastor
(1987).
A major problem with the with-without approach is that
countries in the sample of non-program and program
countries are not randomly selected. They are program
countries because of poor economic conditions prior to
entering into a Fund/Bank-support program. Consequently
there is a systematic difference between these two groups
of countries and the non-random selection of program
countries produces a biased estimate of program effects.
This is the result of attributing observable differences in
economic outcomes between program and non-program countries
to program status, when in fact the initial economic

54
position of the two groups of countries is an important
determinant of economic performance.
To overcome this problem requires identifying the
specific differences between program and non-program
countries in the pre-program period and controlling for
these differences prior to comparing economic outcomes.
This is the idea underlying a modified version of the with-
without approach. In this regard/ Goldstein and Montiel
(1986) proposed a generalized evaluation estimator to
control for pre-program differences between program and
non-program countries. This estimator is a reduced-form
relationship that links changes in macroeconomic outcome
(program target variables) to lagged values of the target
variables, lagged values of policy variables and variables
that represent exogenous effects on the target variables.
These authors applied this approach to a sample of 58
developing countries in which 68 programs were implemented
over the period 1974-1981. The approach was extended by
Serieux (1999) to include the effect of democratization in
Fund/Bank-supported program countries.
3.1.3 Comparison of Simulations Approach
Finally, the simulation approach relies on simulations
of economic models to make inferences on hypothetical
outcomes of Fund/Bank-type policy packages. Khan and Knight

55
(1981), created a simulation using panel data for 29
developing countries in a dynamic econometric model. The
objective was to investigate the hypothetical effects of
pursuing a stabilization program with similar policy
characteristics as a Fund/Bank-supported one. The authors
later (Khan and Knight, 1985) extended their simulation
exercise to include comparisons of alternative packages.
More recent studies include Robinson and Gehlhar (1996),
and Chadha et al. (1998), who use computable general
equilibrium (CGE) models for the Egyptian and Indian
experiences with recent economic reforms, respectively.
3.2 Preliminary Issues in Modeling Supply
Response in Jamaica
A number of empirical studies have been conducted in
the past to estimate crop supply response in Jamaica. A
partial list of these studies is shown in Table 3.1. With
the exception of Gafar (1997), all of the studies were
conducted prior to the period 1980-1999. In addition, none
of the studies addressed the issue of the impact of
economic reforms on agricultural supply response.
One approach to capturing the long-run and short-run
changes in agriculture supply response is to use the
Nerlovian-type partial adjustment models. Both quantity and
prices can be modeled to adjust to their long-run or
equilibrium path, and the model is capable of estimating

56
Table 3.1: Empirical Studies on Agriculture Supply
Responses in Jamaica.
Crop/
Category
Period
Author & Year
Funct.
Form
Price Elasticity
Short-run
Long-run
Banana
1954-1972
1961-1979
Gafar (1980)
Pollard & Graham
(1985)
Linear
log
0.16
0.49
0.57
-2.72
Cocoa
1954-1972
1961-1979
Gafar (1980)
Pollard & Graham
(1985)
Linear
Log
0.41
0.74
2.56
0.76
Coffee
1953-1968
1954-1972
1961-1979
Williams (1972)
Gafar (1980)
Pollard & Graham
(1985)
Log
Linear
Log
0.70
0.92
0.10
-0.80
1.15
0.07
Citrus
1961-1979
Pollard & Graham
(1985)
Log
0.24
-1.33
Sugar
1954-1972
1961-1979
Gafar (1980)
Pollard & Graham
(1985)
Linear
log
0.17-0.29
0.24
0.31-0.7
1.41
Broad Agri.
Aggregates
Export
Domestic
Livestock
Forestry &
Fishing
Total Agri.
1964-1990
Gafar (1997)
log
0.20
0.15
0.15
0.02
0.12
0.35
1.08
0.21
0.21
0.23
Source: Adapted from Gafar (1997, p.213).
both long-run and short-run parameters as well as the speed
of adjustment towards the long run equilibrium. A potential
complication in such an analysis of long-run and short-run
changes is that most economic time series data are non¬
stationary and usually characterized by a unit root.
This means that the linear properties of the series
such as its mean and variance, are not constant over the
sample but change over time (Greene, 1993; Gujarati, 1995).
Nelson and Plosser (1982) have shown that the economic
implications of an economic time series that is

57
characterized by a unit root are different from those of a
stationary process. In particular, an economic time series
with a unit root will have a permanent response consequent
upon any shock in the system. In contrast, a stationary
series will reflect only a transitory response. Therefore,
the response to economic reform policies is not simply a
"policy on", "policy off" or one-shot choice. Even if the
reforms were temporary, so long as the economic series
possesses a unit root, there will be a permanent response.
Non-stationarity of variables poses problems of
estimation of functional relationships using conventional
econometric methods. In the first place, Fuller (1976) has
shown that under non-stationarity the limiting distribution
of the asymptotic variance of the parameter estimates is
not finitely defined, hence the conventional t and F tests
are inappropriate. Secondly, non-stationarity gives rise to
spurious correlation among variables (Greene, 1993) . In
macroeconomic time series it is not unusual to find that a
variable is non-stationary in its level4 but stationary in
first differences. In technical terms, if Yt is non-
stationary, but AYt (the first difference) is stationary,
then Yt is integrated of order one, i.e., Yt~I(l). If two
■g ——
Data that have not been transformed in any way (such as logarithmic
transformation, first differences, etc.), are said to be in 'level'
form.

58
series Yt and Xt are 1(1) then a linear combination of them,
e.g., Zt = Yt - aXt may also be 1(1). However, there may also
exist, a value for a that ensures that Zt is stationary. In
such an event, the two series are said to be cointegrated,
and the cointegrating vector is denoted (1, a).
It is tempting to conclude that in order to estimate
meaningful relationships among non-stationary variables,
all that is necessary is to difference the variables until
they achieve stationarity and then estimate the
relationship. However, Johansen and Juselius (1990) argue
that unless the difference operator is also explicitly
applied to the error process, such differencing results in
loss of information. In this event, resorting only to
estimating the relationship in difference form captures
only the short-term effects, while the long-term
relationship among the variables is left undetected
(Nickell, 1985). Finally, differencing the economic series
may not be appropriate, such as when economic theory
postulates a relationship among variables in levels, not in
difference forms.
To overcome these problems, econometricians have
developed an approach known as error correction models
based on cointegration. Hylleberg and Mizon (1989, p.124)
claim that

59
...when estimating structural models it is our
experience from practical applications that the error
correction formulation provides an excellent framework
within which, it is possible to apply both the data
information and the information obtainable from
economic theory.
Error correction modeling requires two conditions
(1) All variables must be integrated to the same order,
i.e., Yi~I(d) , where d is the number of times Y has
to be differenced to achieve stationarity; and,
(2) All variables must be cointegrated of order (d - b),
where b>0.
The idea underlying cointegration is that one or more
linear combinations of non-stationary variables are
stationary. That is, cointegration approaches the station¬
arity issue as linear combinations of economic series,
rather than by differencing the series. The implication of
this is that if a set of variables is cointegrated then,
following the Granger Representation Theorem (Fuller, 1985)
a valid error correction representation of the data exists.
In effect, then, cointegration is a test of existence of a
long-run relationship of variables that are integrated of
the same order (Greene, 1993; Gujarati, 1995). However, an
important feature of error correction models based on
cointegration is that the data in both levels and
differences are included, thereby facilitating investi¬
gation of both short-run and long-run effects in the data.

60
3.3 Error Correction Model
The error correction model (ECM) can be derived from a
re-parameterization of an Autoregressive Distributed Lagged
(ADL) model (Hendry et al., 1984). Alternatively, the ECM
can be derived from the dynamic optimizing behavior of
economic agents. This latter approach is presented here.
Following Nickell (1985), suppose economic agents
optimize their behavior with respect to an inter-temporal
quadratic loss function:
(3.1) L = ¿ as^(Yt+s - Yt\J + (Yt+S - Yt+S.,)2 - 292
s=0
(Yt+S -Y^.jY^ - Y*+s_1)]
where Y* is the desired or long-run equilibrium value of Y
that the economic
agent can
control
to
minimize
L,
conditional on
information at
time t,
and
subject
to
movements in Y*.
The
discounting
and weighting
factors
are
a,(00), respectively.
Minimizing (3.1) with respect to Yt+S gives a second
order difference equation whose solution at time t is:
AYt = 92AYt* + (l - Ul)
e2Yt-i + (l - «HiX1 “ 02)£
3*0
- Y
t-i
(3.2)

61
where yi is the stable root (i.e., the root that is < 1)
from solving the characteristic equation of the general
Euler equation which was derived from minimizing (3.1).
Equation (3.2) is an optimal rule and has an error
correction term in brackets [.]; the coefficient (l-pi) is
the speed of adjustment, i.e., the speed of closure to any
discrepancy between desired and actual values of Y. In the
error correction term in (3.2) the long-run target is a
convex combination of all target values from Y¿_x onwards.
Nickell (1985) has shown that when 02 = 0 in the loss
function (3.1) then equation (3.2) nests the forward-
looking partial adjustment model (PAM):
(3.3)
It is important to note that the dynamic equation
(3.3) does not necessarily have to be of a standard PAM
type. To estimate (3.3) empirically requires the para¬
meterization of Yt*+S. One way is to model this sequence of
expected target values as a stochastic process such that
Yt*+S are expressed in terms of current and lagged values and
substituted into (3.3). The resulting equation may not
necessarily result in a PAM. Indeed, Nickell demonstrates
that when

62
...we allow the target [y*+s] to follow anything more
complex than a first order autoregression, the
structural equation [3.3], which is fundamentally a
partial adjustment model, will reduce to an error
correction mechanism in terms of observable variables
(Nickell, 1985, p.124).
Suppose that the actual values of the series follow a
second order autoregressive scheme with a unit root and a
drift:
(3.4) Yt+S = g + pY^.i + (1 - PK+s-2 + et+s
where s>0, 8t+s is white noise, and g is the drift. Denoting
the expected value with an (*), the expectation of (3.4)
is:
(3.5) Yt*+S = g + PYt+8_i + (1 - PK+s_2
Following the derivations by Nickell (1985) and
Alogoskoufis and Smith (1991) the solution to the second-
order difference equation (3.5) is:
(3*6) Yt+* = frr gs + (i - (p - i)s+iK*
Substituting (3.6) into (3.2) yields the decision rule in
the form:
(3.7) AYt = c + 9-ft ^ ~ ’ft^'Ivy9^ AY‘ + - Vi)
(1 + apj(l - P))
wh_r_ r = g(l - PX2 - PaUaXl - UiXl - Qz) _ g(aUiXl - Pi)
(2 - p)2[l + aPl(l - p)] (2 - pXl - aPl)
Equation (3.7) is written in the form of an ECM. The
parameterization of Y* in terms of exogenous variables would

63
reflect the long-run cointegrating relationship. According
to Nickell (1985, p.124),
since it is almost a stylized fact that aggregate
quantity variables in economics follow second order
autoregression with a root close to unity, we may find
the error correction mechanism appearing in many
different contexts.
3.4 An Error Correction Model for Crop Supply
Response in Jamaica
The dynamically unrestricted version of the ECM in
(3.7) can be expressed as:
(3.8) AQt = i|r0 + TjTiAQ* + - Qt.j)
where Qt is output in logarithm and ijro, fi, ijf2 are straight¬
forward analogues of the intercept and coefficients in
(3.7). Generally, Q* is parameterized in terms of other
(weakly) exogenous variables. This parameterization shows
the long-run cointegrating relationship between the
exogenous variables and the dependent variable, Qt. Given
the unrestricted nature of (3.8) a wide range of possible
processes that describe the law of motion of Qt can be
accommodated. Following previous agricultural supply
response models, quantity supplied is postulated as a
function of the expected values of a set of variables that
are believed to capture agricultural incentives. These
exogenous variables are, the price of the crop, Pc, the
price of substitute crops, Psi, i=l,2... and the prices of

64
inputs. In this study, two inputs are considered,
fertilizer and labor, whose prices are denoted as F and W,
respectively. With these specifications of the exogenous
variables, the parameterization of the long-run equilibrium
supply function of the cth crop is:
0.9) Qo,t = e0 + Pi^c.t + ^iPsei,t + Mt + m;6 + ct
c = l...n; i=l,2,...; t=l...T; et~iid(0,ct) .
where all variables are measured in logarithms (to
facilitate interpretation of estimated coefficients), and
have been previously defined. The superscript e denotes
expected value of the variables. Given the parameterization
in 3.9, the general error correction model, with all
variables as previously defined, can be written as:
(3.10) AQt = A0 + A1AQt.1 + A2AP®t_1 + A31APseit + A4AW®_1
+ A5AFte.1 + A6(q,._1 - Q^) + v
To give empirical content to these models, the
specification of expected values must be addressed. Clearly
if economic agents had full information about the current
variables at the time they are set, then actual values of
the variables would be substituted for their expected
values. When this is not the case, some process of
expectations formation has to be assumed. The applied
literature on ECM has largely ignored issues of

65
expectations. Because of its simplicity in practice, actual
values are usually substituted for expected values. An
alternative to this procedure is to assume that crop and
input prices are determined by policy changes for one year
in advance. With this assumption, expectations about
current economic variables have to be based on information
that is available up to the end of the previous period,
(t - l) . This can be expressed as:
(3.11) Xf = E(xt|<5tJ =
where Xt denotes the economic variable of interest, and fl> is
the information set available to the economic agent.
Previous studies of crop supply response in Jamaica
assumed naive expectation, based on past (simple) lagged or
polynomial lagged prices. Under rational expectations,
expected current and future prices of crops, wages and
inputs will reflect the generation process for these
explanatory-forcing variables. As noted previously, a
second order auto-regressive process with a unit root with
drift seems adequate to describe the process followed by
many economic series. The data on quantities, prices and
inputs for this study have been analyzed and found to
follow this process. Consequently, the expected prices of
crops and inputs are characterized as follows:
Pc,,-. = ao + a.P,,-, + (l_ ai )Pc,t-2
(3.12)

66
(3.13) P^, = b0 + bjP^j.j + (l - b, )Psi t_2
(3.14) Ft = c0+CjFt_, + (l — c,)Ft_2
(3.15) W;*d0+dIWM+(l-d,)WM
Using (3.8)-(3.15), the general error correction model for
the cth crop is:
2 2
(3.16) AQ ct = a0c + oi1CjtAQC(t._1 + ^ a2jAPc,t-j a3i.k^Psi.t-k
j=l k=l
2 2
+ £ a^AWt.i + X Oi5mAFt.m + a6(Qc - 5XPC - 521Psl
1=1 m=l
-83Wt-84Ft)t_,;
where, i=l,2, all variables as previously defined, and the
random error term is suppressed.
The last term in (3.16) is the error correction term. In
the empirical estimation of the ECM (3.16), the error
correction term is usually specified as the residual from
the cointegrating relationship.
The Engle-Granger (1987) two-step method has been used
extensively in the applied literature to estimate the ECM
(3.16). However this method assumes that the cointegration
vector is unique. Except in the bivariate model, this
assumption may be violated in multivariate models. To test
for, and estimate multiple cointegrating vectors, Johansen
(1988) and Johansen and Juselius (1990) have devised an
appropriate method within the following framework. Define a

67
standard vector autoregressive (VAR) model with lag length
k as:
(3.17) xt = n:xt_x + n2xt_2 + ... + nkxt_k + et
t=l...,T
where X is an Nxl vector of N endogenous variables, et~iid
(0,A) with dimension NxN. The long-run, or cointegrating
matrix is:
(3.18) i — nx - n2 — ... — nk = n
The number of distinct cointegrating vectors, r, which
exists between the variables of X, is given by Rank (II) .
Most economic time series appear to be integrated to
the order of one, in which case, r number of variables in the vector X. In the case of a
bivariate model, N=2, and therefore if the variables are
cointegrated, then there is a unique cointegrating vector.
The matrix n is then decomposed as:
(3.19) n = a (3'
where P represents the matrix containing the r
cointegrating vector, and a is the matrix of weights with
which each cointegrating vector enters each of the
differenced X equations. A large a value implies that the
system will respond to any deviation from the long-run
equilibrium path with a rapid adjustment. If a's are zero

68
in some equations, this is a sign of a weak exogeneity,
implying that the variable does not respond to the
disequilibrium in the system. The parameters a and P form
an over-parameterization of the model. However, the space
spanned by P, sp(p), can be estimated, and shown to be the
empirical canonical variates of Xt-k with respect to AXt.
This is in effect the following theorem advanced by
Johansen:
The maximum likelihood estimator of the space spanned
by P is the space spanned by the r canonical variates
corresponding to the r largest squared canonical
correlations between the residuals of Xt-k and AXt
corrected for the effect of the lagged differences of
the X process. (Johansen, 1988, p.233)
Implementation of this theorem begins by re-para-
meterizing (3.17) (detailed derivations are in Johansen,
(1988) and Enders, (1995)), into the following error
correction model:
(3.20) Axt = I^AX,..* + . . . + r^AX^+j + rkXt_k + et
where Ti = —I + III + II2 + ... + Eh, i=l ...k
Without any loss of information, the ECM in (3.20) is
therefore a transformation of the VAR(k) model in equation
(3.17), and is expressed in first differences and augmented
by the error correction term, rkXt_k . The long-run equi¬
librium or impact matrix is the matrix Tk and is equivalent
to n
oí' P in (3.19). The rank of n is the basis of

69
determining the number of cointegrating relationship
between the variables in the ECM (3.20). Johansen (1988)
identifies three possibilities with regards to Rank (II),
(1) rank(II) = 0. This means that the variables are not
cointegrated and the model is basically a VAR in
first differences.
(2) 0 < Rank(II) < p. In this case, the variables are
cointegrated and the number of cointegrating
relationship (s) is less than the number of
variables, p, in the model.
(3) rank(II) = p. This means that all variables are
stationary and the model is in effect a VAR in
levels.
The loglikelihood representation of (3.20) is:
(3.21) L(a, p, Q) = |0|_T/2
exp
2
2 SÍ
Johansen's procedure begins by regressing AXt on the lagged
differences of AXt and generating fitted residuals Rot, then
regressing Xt-k on the lagged differences and generating
fitted residuals, Rkt- These fitted residuals are then used
to construct the following product moment matrices:
(3.22) SLi
1
Rjt' i' 3 - 0' ^
T

70
The product moment matrices (3.22) are then used to find
the cointegrating vectors by solving the determinant:
(3.23) |XSkk — Sk0S00S0k| = 0
This will yield the estimated eigenvalues (Xi, ..., Xn) and
eigenvectors (vi,...,vn)/ which are normalized such that:
(3.24) V'SkkV = I
where V is the matrix of eigenvectors. The most significant
eigenvectors then constitute the r cointegrating vectors,
i . e.,
(3.25) (3 = (vj, . . .vr)
Using (3.25), a is then estimated from (3.19).
The critical issue in all of this is to determine
which, and how many, of the eigenvectors in (3.24)
represent significant cointegrating relationships. First,
the P vectors that have the largest partial correlation
with AXt, conditional on the lags of AXt, are identified.
Second, the eigen vectors that correspond to the r largest
eigen values are chosen. Finally, to determine the value of
r the following test statistics suggested by Johansen
(1988) are employed:
(3.26) Q1(q, n) = -T ^ ln(l - Xt)
i=q+l
(3.27)
Q2(q, q + l) = -Tln(l - Xq+1)

71
The null hypothesis HO: r HO: r = q is tested against HI: r = q + 1 with (3.27) . The
critical values for these tests are taken from Osterwald-
Lenum (1992) . The critical values from this source
recalculates and extends those critical values from
Johansen (1988) and Johansen and Juselius (1990), to handle
a full test
sequence
from full rank
(r = p
, i.e., Xt is
stationary)
to zero
rank
(r = 0,
i. e.,
all linear
combinations
of X are
Id) ) ,
for at
most
11-dimensional
systems.
The cointegration technique is used to determine the
long-run relationships among variables. This co-movement of
variables in a long-run supply function has not been
explored for Jamaica. Cointegration analysis is appropriate
in this regard. It will also suggest which variables in the
supply function are in the long-run equilibrium.
3.5 The Data
The principal sources of annual time series data,
which are used in this study, are the Food and Agriculture
Organization (FAO) agricultural database (FAOSTAT), avail¬
able on the internet, and annual publications of various
government agencies in Jamaica. These include, Economic and
Social Survey of Jamaica (Jamaica, Planning Institute of
Jamaica); Production Statistics (Jamaica, Statistical

72
Institute of Jamaica (STATIN)); Statistical Digest
(Jamaica, Bank of Jamaica) , Statistical Year Book of
Jamaica (Jamaica, Statistical Institute of Jamaica
(STATIN)); Census of Agriculture, 1968, 1978, 1996
(Jamaica, Statistical Institute of Jamaica (STATIN));
International Financial Statistics (IMF); and data from
published and unpublished documents from the Ministry of
Agriculture in Jamaica.
It should be noted that the data on Jamaica contained
in the FAO database, FAOSTAT, IMF and World Bank sources
are based on publications and data supplied by STATIN,
Ministry of Agriculture, and other agencies of the
Government of Jamaica. The data used to estimate the supply
functions in this study are taken from the FAO FAOSTAT
database. This is the most comprehensive data set on crop
output and prices in Jamaica. However, in several areas the
data are less than ideal. For example, whenever data are
not available the FAO provides its own estimates based on
past crop performance, other crops and other country's
data. Jamaican officials claim that production data for
some crops reported on the FAO data base are not monitored
in Jamaica.
The crop output and price data are annual series, and
are collected at the farmgate through periodic production

73
surveys and agricultural censuses. Farmers in the Caribbean
rarely keep records of production, cost expenditures and so
on. Hence, for these farmers it is difficult to recall data
for production surveys that are conducted between long
periods. Consequently, this casts some amount of suspicion
on the reliability of the data. Nonetheless, these are the
data reported as official statistics, and are used by the
government as a basis for policy analysis/discussion.
Several important economic aggregates were not
adequately covered in the data sources. In particular, data
on agricultural wages are not available on Jamaica. As a
result, a proxy for this variable was constructed from data
on compensation to agricultural employees recorded in the
national accounts. Data on ' consumer price indexes—the
basis for deflating crop and input prices in this study—
were not available as continuous series over the 1962-1999
period. Consequently, a combination of splicing indexes and
converting the final 1962-1999 series to a 1980 base year
was undertaken.
Although the data are less than ideal in several
areas, and are bound to contain some noise, the advantages
of using them are:

74
(1) they are from a common source and are characterized
by a common accounting/estimating procedure in their
derivation; and
(2) the data are the most comprehensive time series
available on Jamaican crop supply in a single
source.
3.6 Summary
There are competing approaches in the literature on
how to evaluate the impacts of economic reforms associated
with Fund/Bank-type policy packages. Each approach utilizes
the idea of a counterfactual against which actual economic
outcomes in the program period are compared. Estimating
crop supply responses over a period in which significant
policy changes have occurred, requires a modeling framework
that is capable of capturing both the long-run and short-
run changes. The Nerlovian-type supply response models have
been used in the literature for this purpose. However, in
the context of data series that are non-stationary, this
approach can produce spurious regressions. A more
appropriate analytical framework is that provided by error
correction models based on cointegration analysis.

CHAPTER 4
EMPIRICAL ESTIMATION OF CROP SUPPLY RESPONSE
The aim of this chapter is four-fold. First, the
central motivations for using error correction modeling
based on cointegration theory in this study are presented.
Second, the time-series data on prices and quantities to be
used in the ECMs to estimate crop supply responses in
Jamaica are tested for stationarity. Third, long-run supply
responses are estimated for eight agricultural crops.
Finally, the short-run dynamics of the crop supply
responses are analyzed.
4.1 Motivations for Using Cointeqration Analysis
Chapter 3 provides a fairly elaborate treatment of the
theoretical and statistical aspects of error correction
modeling based on cointegration theory. Against that
background, it is useful to recall in a cryptic, condensed
and fairly non-technical way, the central motivations for
using this approach to address the issues with which this
study is concerned.
The empirical purpose of this study is to investigate
the impact of economic reforms on crop supply responses in
75

76
Jamaica. Generally, supply response models in agriculture
postulate a long-run relationship between output and
agricultural incentives (Askari and Cummings, 1976).
Deviations from this long-run equilibrium occur in the
short-run and may involve considerable adjustment costs to
the economic agent. This is especially the case when
significant policy changes are implemented. Jamaica
provides a good case for evaluating the connections between
economic reforms and supply responses. The country has
traditionally been heavily dependent upon agriculture for
food, employment and export earnings. A combination of
excesses in state intervention and adverse world economic
conditions prompted major economic reforms in the late
1970s in the form of serious macroeconomic stabilization
policies under the direction of the IMF.
During the early 1980s more far-ranging structural
reforms were instituted in an effort to reverse the current
account and fiscal deficits, reduce inflation and monetary
growth to achieve financial stability, restore economic
growth, and so on. Implicitly in these early reforms, but
more explicitly in the late 1980s and throughout the 1990s,
the aim has been to re-orient the economy towards a more
liberal economic system. These reforms took expression in
progressive devaluation of the Jamaican dollar, elimination

77
of state marketing boards, liberalization of agricultural
input and output prices and privatization of state held
monopolies and public enterprises. Intuitively, therefore,
the analysis of supply responses in these situations would
require a modeling framework that is capable of incor¬
porating both the long-run and short-run changes.
A preliminary analysis of the Jamaican crop output and
price data series over the 1962-1999 period reveals sub¬
stantial fluctuations, especially since the early 1980s.
The strongly time trended data imply a statistical problem
that has not been addressed in applied analysis on Jamaican
crop supply responses. This statistical problem is referred
to, in the literature, as non-stationarity of the time
series.
The analytical approach chosen in this study to deal
with the above mentioned issues, emphases the importance of
considering the interactions between the variables in the
system in a simultaneous equation model, and to distinguish
between the short-run and long-run effects. The modeling
approach differs from those previously used to model
Jamaican crop supply responses in two very important ways.
First, the data are analyzed as a full system of equations.
This allows for possible interactions in determining the
precise relationships among the variables in the system.

78
Second, the multivariate cointegration modeling which
is used in this study, is designed for this type of
empirical work by explicitly classifying the non-stationary
and stationary components and facilitating an inter¬
pretation in terms of the dynamics of short-run and long-
run effects. There are two general motivations for using
error correction models based on cointegration theory in
applied economic analyses. First, when time-series data are
non-stationary, i.e., their linear properties, such as mean
and variance, are time dependent, then conventional
econometric analysis on such data may produce "spurious"
results.
Second, even though a set of time series may
individually be non-stationary, there may be a linear
combination among them that is stationary. Such series are
said to be cointegrated, that is, they have a tendency to
move together in the long-run, even though in the short-run
they may diverge from each other. This co-movement of
related series suggests the existence of long-run
relationship between them. So when data are non-stationary
there is the additional possibility that the data
generating process contains information about the
equilibrium process that makes the process adjust toward
the long-run steady state or the equilibrium path.

79
There is, therefore, additional economic insights
gained from the recognition that the data are non¬
stationary. The component ficXt-i, (= aP'Xt-i, since n = a(3' ) ,
in equation (3.20), is directly related to the non-
stationarity in the data. Since this component contains
information about the speed of adjustment, a, to some long-
run relations, 3'Xt, and if the data are cointegrated, then
the economic process can generally be understood within a
theoretical model that assumes some adjustment behavior.
This means that theoretical models such as the partial
adjustment model (PAM), which assume static equilibrium,
cannot be used when data are non-stationary.
4.2 Definition of Variables
In this chapter, error correction models based on
cointegration theory are used to test the hypothesis that
long-run relationships exist between crop output and price
incentives in Jamaica. The estimation procedure is based on
the work of Johansen (1988) and Johansen and Juselius
(1990). There are basically three steps in this estimation
procedure
(1) test the order of integration of the variables and
specify the lag length of the variables using a
standard vector auto-regressive (VAR) specification.

80
(2) estimate the ECM and the number of cointegrating
relationship(s) among the variables included in the
ECM.
(3) perform short-run analysis by conducting innovation
accounting on the ECM.
For each crop a p-variable vector autoregression (VAR)
of lag k is specified. The model is re-parameterized into
an error correction model (ECM) as specified in equation
(3.20) and estimated according to the methodology of
Johansen (1988) and Johansen and Juselius (1990). Esti¬
mation is done using the Regression Analysis for Time
Series (RATS), version 4.3 (Doan, 1996), and Cointegration
Analysis for Time Series (CATS) in RATS, (Hansen and
Juselius, 1995), computer programs. The variables in each
ECM include the output (quantity) of the crop of interest,
the price of the crop, the price of a substitute (or
alternative) crop, and two input prices, average agri¬
cultural wage rate and average fertilizer price.
The choice of a crop price variable is critical for
the estimation of crop supply responses. In this regard,
Askari and Cummings (1976) suggest using any one of the
following prices:
(1) Nominal farmgate price;
(2) Farmgate price deflated by any one of the following:

81
(a) a price index of farmer's inputs;
(b) a consumer price index; and
(c) some index of the prices of competitive crops (or
the price of the most competitive crop).
An additional issue that is related to the choice of
an appropriate price variable for the crop supply functions
especially within the context of the IMF/World Bank
programs in Jamaica, is that agricultural price incentives
are influenced by various macroeconomic policies. Of
particular importance, in this regard, is the real exchange
rate (RER), defined as the ratio of prices of tradable (PT)
to non-tradable (Pn) goods (Tsakok, 1990). Krueger (1992)
and Schiff and Valdés (1992a, 1992b), have shown
extensively how macroeconomic policies in developing
countries generally have a RER effect, which ultimately
affect output price and agricultural supply. In the
empirical literature the RER is usually approximated as
RER=e*WPI/CPI, where WPI is the foreign wholesale price
index, CPI is the domestic consumer price index, and e is
the official exchange rate. The World Bank's approach to
showing the link between macroeconomic policies (as
represented by the RER) and real crop prices, decomposes
real producer price as follows:
RPP = PF/CPI = PF/PBe = e (WPI/CPI) = NPC*RER*pB

82
where RPP is real producer price, Pf is the farmgate
producer price, PB is the border price, NPC is the nominal
protection coefficient, and pB is the real border price of
the country's exports (World Bank, 1994). This definition
of RPP shows that it contains information on the RER and
also reveals that it is really hazardous to include both
the RER and RPP in the same equation (Mamingi, 1997).
In addition to the issues raised above, two additional
considerations were taken into account in the choice of the
crop price variable. First, since farm producers sell most
of their marketable surplus immediately after reaping,
farmgate prices seem to be a good approximation for prices
received. Second, since farmers purchase most of their
requirements from retail markets, the consumer price index
seems to be an appropriate deflator for producer prices.
For these reasons, therefore, the farmgate price deflated
by the CPI was used as the real producer price (RPP)
variable in the supply functions. In effect, this price
variable reflects not only price incentives to the
producers, but also macroeconomic (reform) policies (as
represented by the RER). In particular, this link between
the price variable and macroeconomic policies is important
in order to pursue an investigation of the hypotheses
advanced on page 11.

83
The question regarding which alternative (or
substitute) crop to include in a particular crop's ECM
proved to be difficult. In the absence of recorded data on
this issue, the final choice of the substitute crop
resulted from a consideration of a number of factors. Among
these were, the nature of the crops (traditional export
versus domestic crops); the terrain where the crops are
grown (mountainous versus flat lands); tree-crops versus
annuals; and, root-crops versus vine-crops. The difficulty
of identifying an appropriate substitute crop (or crops)
can be demonstrated in the case of banana. Possible
substitute crops for banana are coffee and sugar, since
these are all traditional export crops. Banana is grown
extensively on flat lands, but is also cultivated on hilly
terrains as is coffee, whereas, sugar growing has been
confined to the relatively flat plains in Jamaica. Hence,
both sugar and coffee are plausible substitutes for banana,
and the choice of either one or both becomes an empirical
issue.
A further complication in the choice of a substitute
crop arises. In the early 1990s, this researcher observed
two cases where lands which were previously used to
cultivate sugar and banana were converted into papaya
groves. Telephone interviews conducted with officials of

84
the Ministry of Agriculture, Government of Jamaica,
indicate
that
these
are not isolated
cases,
and
that
similar
land
conversions have been observed
into
cut-
flowers,
decorative
foliage, aloe-vera,
ochro,
and
other
non-traditional export crops.
Several substitute crops were initially included in
each ECM but in all cases it was found that the inclusion
of only one substitute crop price improved the statistical
properties of the model. The alternative crops in each
crop's ECM are reported in Table 4.1.
Table 4.1: Alternative (Substitute) Crops in Each ECM.
Crop
Alternative Crops Considered
Banana
Sugar’, coffee, papaya
Sugar
Banana’, papaya
Coffee
Banana’, pimento, sugar, orange
Pimento
Banana’, coffee, cocoa bean
Yam
Cassava*, potato, sweet potato
Orange
Grapefruit*, tangerine, coffee
Cocoa Bean
Banana’, pimento, coffee
Potato
Cassava’, yam, sweet potato
Crop chosen as substitute crop.
Two input prices are included in each crop's ECM,
namely, fertilizer price and average wage in the
agriculture sector. The fertilizer price variable is a
weighted price index of all types of fertilizers imported
into Jamaica, with quantities as weights. With respect to
the wage variable, there are no data on agriculture wages

85
in Jamaica. A proxy for this variable was constructed from
data on compensation to agricultural employees recorded in
the national income accounts. Compensation of agricultural
employees is
defined to include all payments by ... producers of
wages and salaries to their employees in kind and in
cash, and of contributions paid or imputed, in respect
of their employees to social security schemes and to
private pensions, family allowance, casualty
insurance, life insurance and similar schemes.
(Jamaica, National Income and Product, 1970, p.50)
Total compensation to agricultural workers is divided by
total agricultural labor force to yield a proxy for the
average nominal wage rate. The latter is then divided by
the consumer price index and used as the proxy for average
real wage rate in the agricultural sector.
All variables are measured in logarithms and indexed
at 1980=100. The logarithmic transformation of the data was
done because of ease of interpretation of the estimated
coefficients and the frequency of its use in the
econometric literature. In addition, Doan (1996) has argued
that preliminary data transformation helps to straighten
out trends and eliminate systematic tendency of the
variance of the series to depend on the magnitude of the
data. To save on degrees of freedom, the smallest lag-
length was sought while at the same time seeking to pass
the diagnostic tests on the residuals. After some

86
experimentation and testing, a lag-length of k=2 was
settled upon. This is the minimal lag length required for
an ECM, and is sufficiently long to accommodate the issues
of expectations discussed in Chapter 3.
The residual analysis for adequacy of model
specification combines statistical testing and inter¬
pretation of graphs. Tests for autocorrelation utilize the
Ljung-Box (L-B), LM(1) and LM(4) tests. The (L-B) test is a
test of the null hypothesis that the residuals from the
first T/4 observations are not autocorrelated. The LM(1)
and LM(4) are tests of the null hypotheses of no first- and
fourth-order autocorrelation. A multivariate test using the
procedure of Doornik and Hansen (1994) is used to test the
null hypothesis that the residuals from the system of
equations are normally distributed. Finally, the auto¬
regressive conditional heterosckedasticity (ARCH) test is
used to test the null hypothesis that the residuals of each
equation in the system are normally distributed.
All of the tests mentioned above are distributed as x2
with the appropriate degrees of freedom and corrected for
small sample bias when necessary. A sample of graphs that
aid in the residual analysis is presented in Section 4.4.
These residual graphs and residual test statistics for the

87
tests mentioned above, are normal output from the RATS/CATS
program when estimating ECMs, using the Johansen approach.
4.3 Testing for Stationarity
A preliminary step in the estimation of the ECM models
is to test the data for stationarity. Three tests are used
in this study, the Dickey-Fuller (D-F) and Augmented-
Dickey-Fuller (ADF) , the Weighted Symmetric (Vl-S) , and the
Phillips-Perron (P-P) tests. The Time Series Processor
(TSP), (version 4.4), was used to conduct these three tests
on all quantity and price data. First, tests were conducted
on the variables in levels, then on their first difference.
The results are shown in Tables 4.2 and 4.3.
For the Dickey-Fuller test, the following models were
used:
(4.
• 1)
AXt
= a0 +
+ m
(4.
.2)
AXt
= a0 +
+ a2T + fi
where T is time, Xt is a time series, oci are coefficients to
be estimated, |it is white noise, and AXt the first
difference. The ADF tests were conducted on the following
equations:
m
(4.3) AXt = a0 + + P^X^ + nt
i = l
m
AXt = a0 + a1Xt_1 + a2T + P^ AXt_, + *it
i=l
(4.4)

88
Table 4.2: Unit Root Tests--Prices and Quantities in
Levels.
Variable
Tests for Price
Tests for Quantity
W-Sa
DT
P-Pc
W-Sa
DF5
P-Pc
Banana
-2.85
(0.13)
-2.49
(0.34)
-10.22
(0.42)
-1.74
(0.80)
-3.70
(0.02)
-5.39
(0.79)
Sugar
-1.66
(0.83)
-1.24
(0.90)
-4.24
(0.87)
-1.39
(0.92)
-1.14
(0.92)
-5.41
(0.79)
Coffee
-2.91
(0.11)
-2.13
(0.53)
-9.08
(0.50)
-1.08
(0.97)
-3.08
(0.11)
-25.21
(0.02)
Pimento
-2.96
(0.09)
-2.57
(0.29)
-15.83
(0.16)
-2.21
(0.49)
-2.40
(0.38)
-12.44
(0.29)
Yam
-1.96
(0.67)
-1.59
(0.80)
-7.81
(0.60)
-2.06
(0.61)
-4.10
(0.01)
-9.46
(0.47)
Orange
-1.55
(0.88)
-1.34
(0.88)
-4.95
(0.82)
-1.33
(0.93)
-2.43
(0.36)
-10.63
(0.40)
Cocoa Bean
-2.36
(0.38)
-2.07
(0.56)
-7.94
(0.59)
-2.89
(0.12)
-2.53
(0.32)
-20.69
(0.06)
Potato
-2.36
(0.39)
-2.02
(0.59)
10.74
(0.39)
-3.12
(0.06)
-0.58
(0.98)
-20.94
(0.06)
Fertilizer
-1.972
(0.667)
-1.142
(0.922)
-13.460
(0.141)
Wage
-2.092
(0.135)
0.743
(1.000)
-16.639
(0.120)
'lote: Numbers in parentheses are P-values.
a Weighted Symmetric Test.
b Dickey-Fuller Test.
c Phillips-Perron Test.
The order of the ADF test (i.e., the value of m) was
chosen on the basis of residual whiteness, since the ADF
test is not valid when serial correlation among the errors
exists. The test statistics from equations (4.1)-(4.4) are
not distributed as a t-distribution because under Ho: non-
stationarity, the variance is not finite. Fuller (1976) has
simulated the critical values of the distribution for
equations (4.1)-(4.4).
The (W-S) test is sometimes recommended over the (D-F)
tests due to its higher power (Pantula et al., 1994). The

89
Table 4.3: Unit Root Tests—Prices and Quantities in First
Difference.
Variable
Tests for Price
Tests for Quantity
W-Sa
DF/ADF*5
P-Pc
W-Sa
DF/ADF*
P-Pc
Banana
-3.80
(0.01)
-2.12
(0.54)
-18.21
(0.10)
-4.81
(0.00)
-3.23
(0.08)
-18.32
(0.10)
Sugar
-3.67
(0.01)
-2.11
(0.54)
-17.33
(0.12)
-3.65
(0.01)
-3.36
(0.06)
-42.73
(0.00)
Coffee
-3.92
(0.01)
-3.61
(0.03)
-29.02
(0.01)
-3.88
(0.01)
-3.65
(0.03)
-43.70
(0.00)
Pimento
-3.25
(0.04)
-3.00
(0.13)
-44.74
(0.00)
-3.66
(0.01)
-3.47
(0.04)
-39.44
(0.00)
Yam
-3.39
(0.03)
-2.97
(0.14)
-24.80
(0.03)
-3.32
(0.03)
-4.05
(0.01)
-24.77
(0.03)
Orange
-2.90
(0.11)
-3.56
(0.03)
-24.54
(0.03)
-4.17
(0.00)
-3.90
(0.01)
-42.52
(0.00)
Cocoa Bean
-3.24
(0.04)
-2.94
(0.15)
-31.13
(0.01)
-3.54
(0.02)
-3.29
(0.07)
-39.41
(0.00)
Potato
-3.99
(0.00)
-3.70
(0.02)
-28.38
(0.01)
-2.47
(0.31)
-3.12
(0.10)
-28.49
(0.01)
Fertilizer
-6.112
(0.063)
-5.180
(0.052)
-42.821
(0.000)
Wage
-6.472
(0.024)
-6.639
(0.017)
-43.081
(0.000)
Note: Numbers in parentheses are P-values.
a Weighted Symmetric Test.
b Dickey-Fuller/Augmented Test.
c Phillips-Perron Test.
(W-S) test uses a weighted double-length regression. The
first half of the regression regresses Yt on Yt-i and lags of
AYt, with weights (t-l)/T, where T is the full sample size,
and t the size of the sub-sample used for the first half of
the regression. In the second half of the regression Yt is
regressed on Yt+i and leads of Yt-Yt-i with weights [l-(t-
1)]/T (Hall and Cummins, 1998). Phillips and Perron (1988)
have suggested a variant to the Dickey-Fuller test by using
a non-parametric General Methods of Moments (GMM) type

90
method to compute a residual variance which is robust to
serial correlation.
The unit root tests indicate that for each variable
tested, at least one test shows that the variable is non¬
stationary in levels, but stationary in first differences.
The conflicting test results reported in Table 4.3 derive
from the different power of these tests, which in turn
depend upon the estimators used (Pantula et al., 1994).
This should not pose any serious estimation problem, since
Hansen and Juselius (1995, p.l), argue that
...not all individual variables included in [a
multivariate cointegration model] need be 1(1), as is
often incorrectly assumed. To find cointegration
between non-stationary variables, only two of the
variables have to be 1(1).
Consequently, the estimation of crop supply response which
follows is based on the conclusion that the variables
included in the models are all integrated of order one,
i.e., Yi~I(1), which implies that, AYi~I(0).
4.4 Estimating Long-run Supply Response
For the banana ECM, the vector of stochastic variables
is:
Xt = (lbanq, lbanpr, lsugpr, lwage, lferp)'
where lbanq = quantity of banana; lbanpr = price of banana;
sugpr = sugar price (the alternative crop to banana) ; lwage
= average wage in the agricultural sector; lferp = average

91
price of fertilizer. The '1' that begins each variable name
indicates that the variables are measured in logarithms.
The diagnostic statistics for residual analysis for
the unrestricted model for banana are shown in Table 4.4
and suggest that the assumed model for the banana ECM is
supported by the data. This is also supported by the graphs
in Figures 4.1-4.5, where in particular, the standardized
residuals all show a zero-reverting pattern. The next step
Table 4.4: Diagnostic Statistics for Residual
Tests for the Banana ECM.
Tests
Df a
x2
p-value
For normality
of residuals
10
10.322
0.21
For autocorrelation
L-B(9)
124
102.050
0.06
LM (1)
25
23.629
0.54
LM (4)
25
20.661
0.71
For ARCH(2)
Lbanq
:T
1.961
Lbanpr
2
1.492
Lsugpr
2
0.575
Lferp
2
0.638
Lwage
2
0.913
Notes: The critical value for y? [2) = 5.99 at
the 95 percent significance level.
a Df = degrees of freedom.
estimates the number of cointegrating relationships, r, in
the system.
In Table 4.5 the likelihood ratio test statistics for
the rank of II (from equation 3.19) are presented along with

92
Figure 4.1: Residuals for Banana Price in Banana ECM.
Figure 4.2: Residuals for Banana Quantity in Banana ECM

93
Actual and Fitted for DLSUGPR
Standardized Residuals
Lag
Figure 4.3: Residuals for Sugar Price in Banana ECM.
Figure 4.4
Residuals for Wage in Banana ECM

Cross- and autocorrelograms of the residuals
DLBANQ DLBANPR DLSUGPR DLWAGE
DLBANQ
DLBANPR
DLSUGPR
DLWAGE
1
U
L
.... .
- - â– â– 
m mm m
— - — ■ —
â–  â– 
:â– 
” ■ ■
â–  â– 
ni
t
m~ -■ —
â– 
.....
L
M
* *
" 1
â– r
r
""
Lags 1 to 9
Figure 4.5: Cross- and Autocorrelograms in Banana ECM.
Figure 4.6: Plot of Eigenvalues for the Banana ECM.

95
Table 4.5: Tests of Cointegration Rank for
Banana ECM.
Eigen
Values
Amax
Trace
Ho: r
P-r
Amax 90
Trace 90
0.649
37.66
80.99
0
5
21.74
71.66
0.369
16.56
43.33
1
4
18.03
49.92
0.332
14.54
26.27
2
3
14.09
31.88
0.022
8.96
12.23
3
~2~~
10.29
17.79
0.087
3.27
3.27
4
1
7.50
7.50
the 90 percent quantiles of the appropriate limiting
distributions. The estimated maximum eigenvalue and trace
statistics are shown in the columns A,max and Trace,
respectively. The corresponding critical values at the 90
percent level are in the columns Amax 90 and Trace 90.
Testing is done sequentially (i.e., row by row), beginning
with the first row in Table 4.5, until non-rejection of the
null hypothesis is found. This sequential testing begins by
testing the null hypothesis Ho:r = 0, (i.e., there is no
cointegrating relationship in the system) in row one of
Table 4.5. Testing is done by comparing the estimated A.max
and Trace statistics with their theoretical (tabular)
counterparts, A.max 90 and Trace 90, respectively. The null
hypothesis Ho:r = 0 is rejected since the estimated test
statistics, Umax = 37.66 and Trace = 80.99), are greater
than the corresponding critical values for these statistics
Umax 90 = 21.74 and Trace 90 = 71.66), respectively. Row

96
two is then examined. In Table 4.5, Ho:r = 1 cannot be
rejected, since the estimated test statistics (Xmax = 16.56
and Trace = 43.33) are less than their critical values
(Xmax 90 = 18.03 and Trace 90 = 49.92), respectively. This
suggests that there is one long-run or cointegrating
relationship in the banana ECM. This is also supported by
the graph in Figure 4.6 which shows that the eigenvalues
are all within the unit circle.
The model is then re-estimated under the restriction
that only one cointegrating relationship exists in the
system. The resulting cointegrating relationship is:
(4.5) -6.1581banq - 3.9431banpr + 0.9571sugpr
+ 0.6551wage + 3.4641ferp - 23.365 = 0
After normalizing on the banana variable and transferring
the banana quantity variable to the left hand side of the
equal sign the estimates for the cointegrating vector, P,
and the vector of weights, a, are shown in Table 4.6.
Table 4.6: Estimated Long-run and Adjustment Coefficients,
S's, a'3—Banana ECM.
lbanq
lbanpr
lsugpr
lwage
lferp
constant
P's
1
0.640
-0.156
-0.567
-0.106
-3.795
a's
-0.141
(-1.909)°
-0.352
(-1.949)°
0.511
(2.452)b
0.169
(3.329)a
0.097
(0.641)
Note:
Figures in parent!
teses are t-values.
a, b, c, indicate statistical significance at one, five and
ten percent levels, respectively.

97
The cointegrating parameters can be interpreted as
long-run elasticities. Thus, a ten percent increase in the
price of banana will lead to a 6.40 percent increase in
banana supply in the long run, other things remaining
constant. Similarly, a ten percent increase in the average
price of fertilizer will lead to a 1.06 percent decrease in
banana supply. With respect to the substitute crop, a ten
percent increase in sugar price will lead to a 1.60 percent
decrease in banana quantity. Finally, a ten percent
increase in agricultural wage rate will lead to a 5.70
percent decrease in banana output.
The coefficients in the a-vector are the weights with
which the cointegrating vector(s) enter(s) each of the
individual equations in the system. This can be seen from
the estimated n matrix (4.6):
(4.6)
n = aP'
- 0.141
- 0.352
0.511
0.097
0.169
0.640 - 0.156 - 0.567
0.106
lbanq
- 0.141
- 0.091
0.022
0.080
0.015
0.536"
lbanpr
- 0.352
- 0.225
0.055
0.198
0.037
1.334
lsugpr
=
0.511
0.327
- 0.079
- 0.287
- 0.054
- 1.938
lferpr
0.097
0.062
- 0.015
- 0.055
- 0.010
- 0.369
lwage
0.169
0.108
- 0.026
- 0.095
- 0.018
- 0.642
In terms of an economic interpretation, these a
coefficients can be considered as the average speed of
adjustment towards the estimated equilibrium. Thus, a small

98
coefficient indicates a slow adjustment whereas a large
coefficient indicates rapid adjustment. At the indicated
levels of statistical significance, all variables in the
banana ECM (except fertilizer price) adjust to a deviation
from the long-run equilibrium relationship.
In the equation that estimates the long-run supply
response of banana (lbanq in equation 4.6), the average
speed of adjustment is aibanq = -0.141, implying that about
15 percent of any deviation from the long-run equilibrium
in quantity is corrected within the first period. When the
speed of adjustment is low, this might be caused by
government regulations, high cost of adjustment, lags
between planting and harvesting, inadequate supportive
infrastructure in the agricultural sector, and other
factors that tend to slow down the process of adjustment to
the equilibrium growth path.
The estimation and diagnostic testing of the long-run
cointegrating relationship for the other crops that are
included in this study, follows the same format and
procedure as for banana. First, the vector of stochastic
variables for each crop's ECM was specified as follows:
Sugar ECM: Xt = (lsugq, lsugpr, lbanpr, lwage, lferp)';
Coffee ECM: Xt = (lcofq, lcofpr, lbanpr, lwage, lferp)';
Pimento ECM: Xt = (lpimq, lpimpr, lbanpr, lwage, lferp)';

99
Yam ECM: Xt = (lyamq, lyampr, lcaspr, lwage, lferp)';
Orange ECM: Xt = (lorq, lorgpr, lbanpr, lwage, lferp)';
Cocoa-bean ECM: Xt = (lcobq, lcobpr, lbanpr, lwage,
lferp)';
Potato ECM: Xt = (lpotq, lpotpr, lcaspr, lwage, lferp)';
where lsugq = sugar quantity; lsugpr = price of sugar;
lcofq = coffee quantity; lcofpr = price of coffee; lpimq =
pimento quantity; lpimpr = price of pimento; lyamq = yam
quantity; lyampr = price of yam; lorq = orange quantity;
lorpr = price of orange; lcobq = cocoa-bean quantity;
lcobpr = price of cocoa bean; lpotq = potato quantity;
lpotpr = price of potato. The '1' that begins each variable
name indicates logarithms, and all other variables are as
previously defined.
Second, residual analysis was conducted to ensure that
the models were appropriate for the data. Finally, the
hypothesis of reduced rank is then tested on the matrix
IT = ap', which defines the cointegrating vectors P and
adjustment coefficients a. A normalization on the crop
variable is taken, and the normalized P and a vector (s) is
(are) reported.
The estimates of the P and a vectors for each crop are
presented in Table 4.7. The first column in Table 4.7 lists

100
Table 4.7: Estimated Long-run and Adjustment Coefficients,
P's, a's for all Crops.
Crops
Q
Po
Pa
W
F
C
Banana
P's
1
0.640
-0.156
-0.567
-0.106
-3.795
a's
-0.141
(-1.909)b
-0.352
(-1.949)b
0.511
(2.452)a
0.169
(3.469)a
0.097
(0.641)
Sugar
P's
1
0.155
-0.640
-0.106
-0.563
-3.290
a's
-0.141
(-2.899)a
0.511
(2.452)a
0.352
(1.949)b
0.169
(3.329)a
0.097
(2.641)a
Coffee
CE #1
P's
1
0.969
0.291
-0.213
-0.724
-5.348
a's
-0.069
(-2.420)
-0.202
(-1.316)
-0.664
(-4.823)a
0.088
(1.617)
0.107
(1.326)
CE #2
P's
1
0.474
-0.005
-0.448
-2.938
6.503
a's
0.049
(2.717)b
-0.091
(-1.911)°
-0.100
(-1.930)°
-0.002
(-2.091)b
-0.193
(-5.697)a
Pimento
CE #1
P's
1
0.576
-0.849
-0.534
-1.526
1.683
a's
-0.047
(-2.763)a
-0.318
(-6.173)a
0.031
(0.400)
0.004
(2.156)b
0.070
(2.455)a
CE #2
P's
1
-0.225
-0.026
0.542
0.468
-10.579
a's
-0.820
(-4.588)
0.140
(0.932)
0.200
(0.889)
-0.106
(-1.575)
-0.036
(-0.252)
Yam
CE #1
P's
1
-3.755
5.138
0.484
1.225
-16.819
a's
-0.060
(-2.184)b
0.128
(2.813)b
-0.124
(-4.377)a
-0.012
(-0.539)
-0.098
(-2.125)b
CE #2
P's
1
0.418
-0.967
1.860
1.941
-23.344
a's
-0.186
(-4.718)a
-0.273
(-4.147)a
0.156
(3.808)a
-0.045
(-1.730)°
-0.028
(-0.427)
CE #3
P's
1
1.706
-2.958
-5.880
-1.700
23.210
a's
-0.033
(-2.851)a
-0.062
(-3.240)a
0.026
(2.184)b
0.021
(2.257)b
-0.042
(-2.189)b
Orange
P's
1
2.347
-1.620
-1.979
-1.492
11.270
a's
-0.139
(-1.472)
-0.366
(-1.858)
0.021
(0.094)
0.062
(2.134)
0.174
(3.047)
Cocoa
Bean
P's
1
0.359
-0.081
-1.405
-0.983
-12.350
a's
-0.797
(-3.971)
0.676
(2.103)
-0.841
(-3.449)
-0.284
(-1.621)
-0.062
(-0.694)

101
Table 4.7: Cont'd.
Crops
Q
Po
Pa
W
F
C
Potato
CE #1
p's
1
1.952
-1.030
-5.197
0.704
-38.815
a's
-0.053
(-1.108)
-0.043
(-1.273)
0.077
(4.035)®
0.031
(3.052)*
-0.068
(-2.793)*
CE #2
P's
1
-0.251
0.220
0.127
0.323
-5.860
a's
-1.008
(-3.569)*
0.071
(2.364)3
-0.056
(-2.504)®
-0.018
(-1.999)b
-0.117
(-2.815)®
Note: Figures in parentheses are t-values.
CE indicates 'cointegration equation'.
a, b, c, indicate statistical significance at one, five and
ten percent levels, respectively.
Q = quantity; P0 = own-price; Pa = price of the alternative
crop; W = average wage rate in agriculture sector; F =
average price of fertilizer; C = constant (intercept).
the crops, which are included in this study, followed by
the 3 and a vectors. When there are more than one long-run
cointegrating relationships, these are numbered as "CE
#(.)" with the associated 3 and a vectors. Columns 2-6
indicate the variables and coefficients in the ECM for each
crop, and column 7 records the intercepts. Thus, in the
case of sugar, there is one long-run or cointegrating
relationship, which, from Table 4.7, can be expressed as:
Q = -3.290 + 0.957Po - 0.640Pa - 0.106W - 0.563F
where Q = quantity of sugar; P0 = own-price (i.e., price of
sugar); Pa = price of alternative crop; W = average wage in
agriculture; F = average price of fertilizer.
A number of inferences can be drawn from the
information presented in Table 4.7. First, the long-run

102
relationship between supply and own-price is inelastic for
five of the eight crops. Crops that are price elastic are
yam, orange and potato, with own-price elasticities of 1.7,
2.3, and 2.0, respectively. The own-price elasticity
coefficients reported in Table 4.7 are generally within the
range estimated by other studies for Jamaica (see Table
3.1) .
A second point to note with respect to the information
in Table 4.7 is the relatively low elasticities with
respect to the substitute crop price. For the banana ECM
it is -0.16, for CE #2 in the coffee ECM, -0.01, and for
sugar and cocoa-bean ECMs, -0.64 and -0.08 respectively.
Relatively higher elasticities are recorded for the pimento
-0.85, yam -3.0 , orange -1.62 and potato -1.03.
Third, the relationship between quantity and input-
prices is generally inelastic. The exception are for wages
in the yam and potato ECMs, and for fertilizer price in
coffee, pimento, orange, and yam. Thus, in the coin¬
tegration equation #3 for yam, in the long-run a one
percent increase in average agricultural wage will lead to
a 6 percent fall in yam supply, other things remaining
constant. A similar conclusion is arrived at for the first
cointegrating relationship for potato. Similarly, the long-
run relationship between supply and fertilizer price is

103
elastic for coffee (CE #2) , pimento (CE #1) orange and yam
(CE #3).
A fourth point to note with respect to Table 4.7 is
about the adjustment coefficients a. Generally, the adjust¬
ment coefficients are significant. The exceptions include
fertilizer price in the banana ECM, own-price, and wages in
CE #1 for the coffee ECM, price of the alternative crop in
CE #1 for the pimento ECM, and a few others. An
insignificant adjustment coefficient, ai, is a sign of weak
exogeneity of the variable Xi in the vector of stochastic
variables, Xt, ih the ECM. Weak exogeneity means that
although a long-run relationship exists between Xi and the
other variables in Xt, Xi does not adjust to deviations from
the long-run equilibrium. As such, any deviation from the
long-run equilibrium in the system after a shock, is
restored by adjustments made by variables other than Xi in
the system (Johansen and Juselius, 1990, 1992) . According
to Johansen (1995) , this does not preclude Xi from the
cointegrating relationship in Xt.
An issue of some importance is raised here regarding
the results in Table 4.7. This relates to the negative
relationship observed between quantity and own-price for
some of the crops, and what interpretation is possible in
these cases. An examination of the data on quantity and

104
own-price reveals strong negative correlations for banana
(-0.737), and sugar (-0.760), and modest negative
quantity-own-price correlations for coffee (-.420) and
potato (-0.390). The quantity-own-price correlations for
the other crops are positive.
In Table 4.7 negative relations between quantity and
own-price show up for pimento (CE #2) , yam (CE #1) and
potato (CE #2) . Given the strong negative quantity-own-
price correlation coefficients for banana and sugar, it is
surprising that negative relations between quantity and
price do not show up in the estimated cointegrating
relationships for these two crops.
A number of possible explanations can be offered for
these results. First, graphical examination of the
quantity-own-price relationship reveals that the negative
correlations are less pronounced than what the magnitude of
the estimates would imply. Second, and as an extension of
the previous point, the negative relationship is observed
for specific and short period(s), rather than for the
entire 1962-1999 period. In fact, in chapter 5 where supply
functions are estimated for the eight crops over two sub¬
sample periods, 1962-1979 and 1980-1999, the negative
quantity-own-price relations are more prevalent (see Tables
5.4, and 5.5, pp.157-160).

105
Third, error correction models based on cointegration
account for the effects of not one or two, but of all the
cointegrated variables in the system. In particular, the
ECM technique provides a way of separating the long-run
relationship between the economic variables (yt = Pxt ) from
the short-run responses (i.e., the Ayt, AXt terms). In this
regard, Engle and Granger (1987) have demonstrated that,
asymptotically, these effects are dominated by the long-run
relationships, and as the sample size becomes larger, the
estimates of the cointegrating vector converges to the true
value at a rate equal to T"1 , where T is the sample size.
This rate of convergence is faster than the case where the
variables are stationary, in which case the rate of
convergence is t"1/2 . This is referred to as the super¬
consistency property of the estimates produced by ECMs. In
effect, therefore, the estimates of the cointegrating
vectors produced by ECMs based on cointegration, are not
confined to, but rather, go beyond the simple correlations
which may be observed between variables.
Finally, cases where the cointegrating relationships
reveal a negative relationship between quantity and own-
price, need to be commented on. It is tempting to interpret
these cases as reflecting relationships on a demand
function, i.e., a negative relationship between quantity

106
and own-price. However, this might not be a plausible
explanation, since variables which would usually be
included in a demand function such as consumer income, an
index for consumer tastes, and other demand forcing
variables, are not included in the crops' ECM. For those
cases observed in Table 4.7, the strongest statement which
could be offered for the negative relationships between
quantity and own-price is that there is a long-run
relationship observed for some crops, with possible
properties of an incompletely specified real demand
relationship.
The diagnostic statistics for model specification and
tests for cointegration rank for each estimated ECM are
shown in Appendix D. Graphs for residual analysis (not
reported), similar to those for banana, were examined in
conjunction with the diagnostic statistics. The residual
analysis suggests that for all crops, the ECMs are
appropriate for the data.
4.5 Analysis of Short-run Dynamics
The previous section suggests that the series in the
crop ECMs are cointegrated. Since the cointegrated
variables are in equilibrium, exogenous shocks to one of
the variables in time to, will result in time paths of the
system which will eventually stabilize to a new equilibrium

107
unless further shocks occur in the system. Lütkepohl and
Reimers (1992), contend that these time paths provide
insights into the short-run and long-run relationship
between/among the variables in the system, especially in
cases where there are more than one cointegrating
relationships in an ECM. In this regard, innovation
accounting (analysis of impulse response functions and
variance decompositions) has become a useful tool in the
analysis of cointegrated systems (Enders, 1995, 1996;
Lütkepohl and Reimers, 1992) .
The impulse response functions, specified as VARs,
generate impact or dynamic multipliers. These coefficients
capture the effects of exogenous shocks of each variable on
its own time path, and on those of the other variables in
the system, by imposing a recursive structure on the moving
average representation of the VAR model (Enders, 1995). The
moving average representation facilitates analysis of the
interaction between sequences in the series of a VAR.
Impulse response functions can be plotted to provide a
visual representation of the behavior of each series in
response to shocks in the system.
Identification of the parameters in the VAR
necessitates imposing some structure (via parameters
restrictions) on the system. One method of imposing

108
restrictions uses the Choleski decomposition (Doan, 1996).
In this method, the ordering of the variables in the VAR is
important and is based on statistical properties of the
data as well as theoretical insights (Enders, 1995).
However, the method can be tedious since, for example, in a
five variable system there are 5! (=120) possible orderings
of the variables.
Sims (1980) and Bernanke (1986) have proposed an
alternative method of imposing restrictions for the
identification of the VAR. In this scheme, exact
identification of the structural model for an n-variable
VAR necessitates (n2 - n)/2 restrictions but additional (over
identifying) restrictions can be imposed and tested
statistically. The restrictions are based on economic
theory.
The VAR in standard form (Enders, 1996), which is used
for the impulse response analysis is specified as:
(4.7) Xt = Ao + AiXt-i + A2Xt-2 + ... + AXt-p + et
where:
Xt = (AQ, APo, APa, AF, AW) ' ;
Xt-i = (AQt-i, AP0,t-if APa,t-i/ AF,t-i, AW,t-i) ' , i = 1,2...
Ai = parameters to be estimated; et = vector of error terms.
All other variables as previously defined.

109
Restrictions on (4.7) are based on the relationships
that are specified in Table 4.8. In effect, the
specification suggests that P0, Pa, F and W are exogenous
variables, and further, that innovations in any one of
these variables will induce contemporaneous changes in
itself and on Q, but on no other variable in the system.
However, this specification, when combined with the Sims-
Bernanke error variance decomposition, does not rule out
induced changes in the other variables in future periods.
This is to be expected, given that the variables are
cointegrated.
Table 4.8: Endogenous and Exogenous Variables in
Crops' Impulse Response Functions.
The Contemporaneous
Value of:
Is Affected by the
Contemporaneous value of:
Q
All variables in the system
Po
No other variable
Pa
No other variable
F
No other variable
W
No other variable
Notes: Q = crop output; Po = (own-price), i.e., price
of output; Pa = price of substitute crop; F = fertilizer
price; W = average agriculture wage rate.
Related to impulse response functions are forecast
error variance decompositions, which provide information
about the proportion of the movements in a variable due to
its own shocks and to those from other variables in the

110
system (Enders, 1996). A twelve-period forecasting horizon
is specified, and the Sims-Bernanke forecast error variance
decompositions are used based on the restrictions implied
in Table 4.8.
Impulses for 12 periods were examined for each crop's
VAR and the first and second period results are reported in
table format, while the plots for the entire forecast
period are reported graphically. For all variables, the
effects of the shocks after the fourth and fifth periods
converge to zero or the eguilibrium path, which is to be
expected from a cointegrated system.
The results for the banana VAR are reported in Table
4.9. Thus, a one-standard-deviation shock in banana price
(egual to 0.220 units) induces contemporaneous decrease of
0.165 units in banana guantity. By model specification,
there is no contemporaneous change in the other variables.
After one period, banana price is still 0.112 units above
its mean, while banana quantity has increased by 0.076
units. In the second period, the shock in banana price has
induced an increase in sugar price of 0.081 units, a
decrease in fertilizer price of 0.007 units, and an
increase in wage of 0.041 units. These effects are also
shown graphically. As seen in Figures 4.7-4.14, equilibrium
is achieved between the fourth and sixth periods.

Ill
Table 4.9: Responses of Banana Quantity to Shocks
in the Banana VAR.
Responses to Shock in sugar price
Period
dlsugpr
dlferp
dlwage
dlbanpr
Dlbanq
1
0.210
0
0
0
-0.162
2
0.071
0.020
-0.051
0.171
0.092
Res
ponses to Shock in fertilizer price
Period
dlsugpr
dlferp
dlwage
dlbanpr
dlbanq
1
0
0.150
0
0
-0.039
2
-0.007
0.023
-0.047
0.014
0.018
Responses to Shock in wage rate
Period
dlsugpr
dlferp
dlwage
dlbanpr
dlbanq
1
0
~o~
0.055
0
0.002
2
-0.014
0.020
0.031
-0.018
0
Responses to Shock in banana price
Period
dlsugpr
dlferp
dlwage
dlbanpr
dlbanq
1
0
0
0
0.220
-0.165
2
0.081
-0.007
0.041
0.112
0.076
Notes: The prefix 'd' to the variable names
denotes change in the variable. All variables as
previously defined.
To complement the results from the impulse response
functions, the forecast error decompositions of the banana
VAR are reported in Table 4.10. Enders (1996) argues that
in applied research, it is typical for a variable to
explain almost all of its forecast error variance at short
horizons and smaller proportions at longer horizons. The
Sims-Bernanke forecast error variance decompositions (based
on restrictions specified in Table 4.8) reported in Table
4.10 show that each variable explains 100 percent of its
forecast error variance in the first period. However, after
the first period, the forecast error variance of a
variable, which is explained by its own shock, falls. In

112

Innovation in Coffee Quantity—Coffee VAR.
dlbanpr
dlferp
dlwage
Isugpr
dlsugq
1 ' 1 ' 1 1
6 8 10
Figure 4.12: Impulse Responses to an
Innovation in Sugar Quantity—Sugar VAR.
VAR.
113

114
the case of the banana VAR, in the third period, sugar
price explains 75.7 percent of its own forecast error
variance, fertilizer price 71.2 percent, wage 33.0 percent,
banana price 42.9 percent, and banana quantity 57.3
percent. These percentage distribution of the forecast
error variance among the variables remain constant after
the second and third forecast periods.
Table 4.10: Variance Decomposition Percentage of One-period
and Three-period Forecast Error Variance—Banana VAR.
Forecast error
variance in:
Percentage forecast error
explained by shocks in:
Sugar price
dlsugpr
dlferpr
dlwage
dlbanpr
dlbanq
1 Yr
100
0
0
0
0
3 yrs
75.712
0.105
0.613
10.395
13.176
Fertilizer price
1 Yr
0
100
0
0
0
3 yrs
11.999
71.220
1.999
1.119
13.663
Wage
1 Yr
0
0
100
0
0
3 yrs
24.430
24.341
32.961
18.196
0.072
Banana price
1 Yr
0
0
0
100
0
3 yrs
21.262
0.149
0.340
42.880
35.368
Banana quantity
1 Yr
19.407
1.105
0.004
19.940
59.543
3 yrs
21.416
1.147
0.006
20.094
57.337
The cross-variable effects are zero in the first
period (by model specification), but change with the time
horizon. In Table 4.10, the cross-variable effects in the
third period show that the shock in banana price can

115
explain 43 percent of its own forecast error variance,
shocks in banana quantity and sugar price account for 35
percent and 21 percent, respectively. Shocks in fertilizer
price and the wage rate account for only 0.149 and 0.340
percent, respectively, of the forecast error variance in
banana price. Similarly, in period three, the cross-price
effect of a shock in fertilizer price can explain 71
percent of its own forecast error variance, while 14
percent and 12 percent are accounted for by shocks in
banana quantity and sugar price, respectively. Finally, a
shock in wage rate explains 33 percent of its own forecast
error variance, while shocks in banana, fertilizer and
sugar prices account for 18.2, 24.3 and 24.4 percent,
respectively.
The last two rows in Table 4.10 show the forecast
error variance decompositions when the exogenous variables
(Po, Pa, W, F) are modeled to impact banana quantity in all
time periods. Thus, the banana quantity variable explains
60 percent of its own 1-step ahead forecast error variance,
and 57 percent in the 3-step ahead forecast error variance.
Similarly, in the first period, banana and sugar prices
account for 20 percent and 19 percent of the forecast error
variance in banana quantity, respectively. These percent¬
ages increase slightly in the 3-year ahead forecast.

116
Fertilizer price and wages account for negligible
percentage variations in banana forecast error variance in
both of the periods, suggesting thereby, the relative
importance of own-price and price of the substitute crop on
the short-run variations of banana quantity.
The estimates from the impulse response functions and
forecast error decompositions for the other crops are
reported in Tables 4.13-4.26. The associated plots of the
impulse response functions are shown in Figures 4.15-4.46.
In order to present the information in a manner that is
succinct and easy to read, a major portion of the estimates
that are reported in Tables 4.13-4.26 are summarized in
Tables 4.11 and 4.12. Table 4.11 shows the responses of
quantity to shocks in the exogenous variables in the system
in periods one and two. Table 4.12 reproduces the last two
rows of the tables that record the forecast error variance
decompositions for the crops' ECM. That table shows the
percentage forecast error variance in quantity, which is
explained by the variables in the system.
The response of quantity to positive shocks in own-
price (P0) is expected to be positive. However, Table 4.11
shows that in period one, five of the eight crops (banana,
sugar, yam, orange, and potato) show a decrease in output,
which is induced by an own-price shock. Thus, in period one

117
Table 4.11: Responses of Quantity to Shocks in Exogenous
Variables.
Variable
Period
Responses of Quantity to Shocks in:
Po
Pa
W
F
Banana
1
-0.165
-0.162
0.002
-0.039
Quantity
2
0.076
0.092
0
0.018
Sugar
1
-0.162
-0.165
0.002
-0.039
Quantity
2
0.092
0.076
0
0.018
Coffee
1
0.021
-0.006
-0.016
0.004
Quantity
2
-0.005
0.021
0.009
0.004
Pimento
1
0.022
-0.006
-0.023
-0.021
Quantity
2
0.033
0.078
0.019
-0.090
Yam
1
-0.087
-0.008
-0.046
-0.019
Quantity
2
0.100
0.018
0.014
0.026
Orange
1
-0.038
0.061
0.097
0.025
Quantity
2
0.210
-0.148
-0.026
0.023
Cocoa Bean
1
0.007
0.028
0.031
0
Quantity
2
-0.062
-0.005
-0.046
0.006
Potato
1
-0.194
-0.026
-0.050
0.049
Quantity
2
0.177
-0.028
0.021
-0.012
Source: Tables 4.13
, 4.15, 4
.17, 4.19
4.21, 4
.23 and 4.
a one-standard-deviation-shock in banana price (equal to
0.220 units—see Table 4.13), induces a decrease in banana
supply of 0.165 units. Other crops whose supply decrease in
period one consequent upon a one-standard-deviation-shock
in own-price are sugar (-0.162 units) and potato (-0.194
units). However, with the exception of coffee and cocoa-
bean, all crops increase supply in the second period
following a shock in own-price.
The response of quantity to positive shocks in the
substitute crop price, Pa, wage, W, and fertilizer price, F,
is expected to be negative. In period one this expectation
is confirmed for shocks in Pa on all crop supply except

118
orange and cocoa-bean. Thus, in the case of coffee supply
response, a one-standard-deviation-shock in banana price
(equal to 0.223 units—see Table 4.15), induces a
contemporaneous decrease in coffee supply of 0.006 units.
Interestingly, in the second period, with the exception of
orange, cocoa-bean and potato supply, shocks in the
substitute crops' price induce positive response in crop
supply response for the other crops.
Wage rate shocks appear to have negligible
contemporaneous effect and no effect in period two on
banana and sugar responses. Wage shocks do have negative
contemporaneous effect on coffee, pimento, yam, and potato
supply responses. However, in period two, it appears that
wage shocks generally have a positive impact on short-run
supply response. The exceptions are orange and cocoa-bean.
Finally, shocks in fertilizer price have negative
contemporaneous effect on banana, sugar, pimento and yam
supply response. However, with the exception of potato, and
pimento, supply response of crops to a shock in fertilizer
price are positive in period two.
Table 4.12 shows the proportion of forecast error
variances in quantity, which are attributable to the
variables that are specified as exogenous in the VARs for
each crop. As previously mentioned, it is common for a

119
variable to explain a larger proportion of its
contemporaneous forecast error variance and smaller
proportions at later periods. This is generally the case
for the crops shown in Table 4.12. Thus, 92 percent of
coffee's contemporaneous forecast error variance is
explained by its own shock, five percent by own-price (P0),
three percent by wage shocks and by negligible percentages
in fertilizer and the substitute crop prices. However, in
the third period, the percentage forecast error variance
Table 4.12: Summary of Forecast Error Variance in Crop
Quantities Explained by Variables in the VAR (Percentage).
Forecast
Error
Variance
in:
Period
Percentage of Forecast Error Variance
in Quantity Explained by Shocks in:
Pa
F
W
Po
Q
Banana
1
19.407
1.105
0.004
19.940
59.543
3
21.416
1.147
0.006
20.094
57.337
Sugar
1
19.940
1.105
0.004
19.407
59.543
3
20.094
1.147
0.006
21.416
57.337
Coffee
1
0.394
0.150
2.727
4.564
92.165
3
4.732
0.805
2.857
3.950
87.655
Pimento
1
0.070
0.839
0.948
0.917
97.226
3
7.285
10.403
2.914
1.959
77.439
Yam
1
0.127
0.653
3.691
13.330
82.199
3
1.869
2.100
3.244
25.605
67.183
Orange
1
6.461
1.110
16.195
2.465
73.769
3
21.606
1.548
5.842
43.470
27.534
Cocoa Bean
1
2.026
0
2.451
0.141
95.383
3
1.588
0.133
6.502
11.047
80.730
Potato
1
0.714
2.626
2.711
40.583
53.367
3
1.485
2.440
2.459
54.282
39.334
Source: Tables 4.14, 4.16, 4.18, 4.20, 4.22, 4.24 and4.26.

120
explained by coffee quantity shock reduces to 88 percent,
while the proportion explained by the substitute crop price
shock increases to five percent. Enders (1995, 1996) argues
that if Xi shocks do not explain any of the forecast error
variance in X2 in a VAR at all time horizons, then the X2
sequence is exogenous. This means that the X2 sequence
evolves independently of shocks in Xi. On the other hand, if
Xi shocks explain all the forecast error variance in X2
sequence at all forecast horizons, then X2 is said to be
endogenous. Neither of these two extreme cases are observed
in Table 4.12.
With respect to the issues of endogeniety/exogeniety
and the results reported in Table 4.12, one point must be
noted. With only a few exceptions, shocks in fertilizer
price and wages appear to explain relatively small
proportions of the forecast error variance in quantity. The
exceptions are pimento, orange and to some extent cocoa-
bean. This is also true for the price of the substitute
crop in the cases of coffee, cocoa-bean and potato.

121
Table 4.13: Responses of Sugar Quantity to Shocks
in the Sugar VAR.
Responses to Shock in banana price
Period
dlbanpr
dlferp
dlwage
dlsugpr
dlsugq
1
0.220
0
0
0
-0.165
2
0.112
-0.007
0.041
0.081
0.076
Res
ponses to Shock in fertilizer price
Period
dlbanpr
dlferp
dlwage
dlsugpr
dlsugq
1
0
0.150
0
0
-0.039
2
0.014
0.023
-0.047
-0.007
0.018
Responses to Shock in wage
Period
dlbanpr
dlferp
dlwage
dlsugpr
dlsugq
1
0
0
0.055
0
0.002
2
-0.018
0.020
0.031
-0.014
0
Responses to Shock in sugar price
Period
dlbanpr
dlferp
dlwage
dlsugpr
dlsugq
1
0
0
0
0.210
-0.162
2
0.171
0.020
-0.051
0.071
0.092
Notes: The prefix 'd' to the variable names
denotes change in the variable. All variables
previously defined.
Table 4.14: Variance Decomposition Percentage of One-period
and Three-period Forecast Error Variance—Sugar VAR.
Forecast error
variance in:
Percentage forecast error
explained by shocks in:
Banana price
dlbanpr
dlferpr
dlwage
dlsugpr
dlsugq
1 Yr
100
0
0
0
0
3 yrs
42.880
0.149
0.340
21.262
35.368
Fertilizer price
1 Yr
0
100
0
0
0
3 yrs
1.119
71.220
1.999
11.999
13.663
Wage
1 Yr
0
0
100
0
0
3 yrs
18.196
24.341
32.961
24.430
0.072
Sugar price
1 Yr
0
0
0
100
0
3 yrs
10.395
0.105
0.613
75.712
13.176
Sugar quantity
1 Yr
19.940
1.105
0.004
19.407
59.543
3 yrs
20.094
1.147
0.006
21.416
57.337

122
Table 4.15: Responses of Coffee Quantity to Shocks
in the Coffee VAR.
Responses to Shock in banana price
Period
dlbanpr
dlferp
dlwage
dlcofpr
dlcofq
1
0.223
0
0
0
-0.006
2
0.004
-0.034
0.023
0.003
0.021
Responses to Shock in fertilizer price
Period
dlbanpr
dlferp
dlwage
dlcofpr
dlcofq
1
0
0.150
0
0
0.004
2
-0.006
0.014
-0.058
-0.082
0.007
Responses to Shock in wage
Period
dlbanpr
dlferp
dlwage
dlcofpr
dlcofq
1
0
0
0.073
0
-0.016
2
-0.014
0.032
0.034
-0.005
0.009
Responses to Shock in co:
:fee price
Period
dlbanpr
dlferp
dlwage
dlcofpr
dlcofq
1
0
0
0
0.178
0.021
2
0.009
-0.015
0.014
-0.034
-0.005
Notes: The prefix 'd' to the variable names
denotes change in the variable. All variables
previously defined.
Table 4.16: Variance Decomposition Percentage of One-period
and Three-period Forecast Error Variance—Coffee VAR.
Forecast error
variance in:
Percentage forecast error
explained by shocks in:
Banana price
dlbanpr
dlferpr
dlwage
dlcofpr
dlcofq
1 Yr
100
0
0
0
0
3 yrs
87.482
0.080
1.039
0.180
11.219
Fertilizer price
1 Yr
0
100
0
0
0
3 yrs
4.411
87.721
5.209
1.034
1.625
Wage
1 Yr
0
0
100
0
0
3 yrs
8.878
38.064
50.623
2.393
0.041
Coffee price
1 Yr
0
0
0
100
0
3 yrs
1.898
16.105
0.419
76.390
5.188
Coffee quantity
1 Yr
0.394
0.150
2.727
4.564
92.165
3 yrs
4.732
0.805
2.857
3.950
87.655

123
Table 4.17: Responses of Pimento Quantity to Shocks
in the Pimento VAR.
Responses to Shock in banana price
Period
dlbanpr
dlferp
dlwage
dlpimpr
dlpimq
1
0.234
0
0
0
-0.006
2
0.001
-0.037
0.025
0.019
0.078
Responses to Shock in fertilize price
Period
dlbanpr
dlferp
dlwage
dlpimpr
dlpimq
1
0
0.150
0
0
-0.021
2
-0.006
0.013
-0.056
-0.048
-0.090
Responses to Shock in wage
Period
dlbanpr
dlferp
dlwage
dlpimpr
dlpimq
1
0
0
0.071
0
-0.023
2
-0.008
0.030
0.035
-0.063
0.019
Responses to Shod
< in pimento price
Period
dlbanpr
dlferp
dlwage
dlpimpr
dlpimq
1
0
0
0
0.199
0.022
2
0.004
-0.010
0.015
-0.017
0.033
Notes: The prefix 'd' to the variable names
denotes change in the variable. All variables
previously defined.
Table 4.18: Variance Decomposition Percentage of One-period
and Three-period Forecast Error Variance—Pimento VAR.
Forecast error
variance in:
Percentage forecast error
explained by shocks in:
Banana price
dlbanpr
dlferpr
dlwage
dlpimpr
dlpimq
1 Yr
100
0
0
0
0
3 yrs
97.005
0.622
0.354
0.108
1.911
Fertilizer price
1 Yr
0
100
0
0
0
3 yrs
5.401
87.773
4.943
0.456
1.427
Wage
1 Yr
0
0
100
0
0
3 yrs
11.946
35.878
48.819
2.842
0.516
Pimento price
1 Yr
0
0
0
100
0
3 yrs
0.913
8.551
10.265
79.974
0.298
Pimento quantity
1 Yr
0.070
0.839
0.948
0.917
97.226
3 yrs
7.285
10.403
2.914
1.959
77.439

124
Table 4.19: Responses of Yam Quantity to Shocks
in the Yam VAR.
Responses to Shock in cassava price
Period
dlcaspr
dlferp
dlwage
dlyampr
dlyamq
1
0.119
0
0
0
-0.008
2
0.046
0.044
-0.033
0.051
0.018
Responses to Shock in fertilizer price
Period
dlcaspr
dlferp
dlwage
dlyampr
dlyamq
1
0
0.147
0
0
-0.019
2
-0.037
0.002
-0.047
-0.018
0.026
Responses to Shock in wage
Period
dlcaspr
dlferp
dlwage
dlyampr
dlyamq
1
0
0
0.064
0
-0.046
2
0.007
0.022
0.035
0.007
0.014
Responses to Shock in yam price
Period
dlcaspr
dlferp
dlwage
dlyampr
dlyamq
1
0
0
0
0.179
-0.087
2
-0.013
-0.031
0.038
-0.023
0.100
Notes: The prefix 'd' to the variable names
denotes change in the variable. All variables
previously defined.
Table 4.20: Variance Decomposition Percentage of One-period
and Three-period Forecast Error Variance—Yam VAR.
Forecast error
variance in:
Percentage forecast error
explained by shocks in:
Cassava price
dlcaspr
dlferpr
dlwage
dlyampr
dlyamq
1 Yr
100
0
0
0
0
3 yrs
84.304
8.617
0.466
1.210
5.404
Fertilizer price
1 Yr
0
100
0
0
0
3 yrs
7.353
83.702
2.931
4.949
1.064
Wage
1 Yr
0
0
100
0
0
3 yrs
17.798
20.621
43.241
17.782
0.558
Yam price
1 Yr
0
0
0
100
0
3 yrs
6.503
1.294
0.321
83.129
8.752
Yam quantity
1 Yr
0.127
0.653
3.691
13.330
82.199
3 yrs
1.869
2.100
3.244
25.605
67.183

125
Table 4.21: Responses of Orange Quantity to Shocks
in the Orange VAR.
Responses to Shock in grapefruit price
Period
dlgrpr
dlferp
dlwage
dlorpr
dlorq
1
0.523
0
0
0
0.061
2
-0.103
0.027
-0.038
0.081
-0.148
Responses to Shock in fertilizer price
Period
dlgrpr
dlferp
dlwage
dlorpr
dlorq
1
0
0.148
0
0
0.025
2
-0.088
0.012
-0.045
-0.122
0.023
Responses to Shock in wage
Period
dlgrpr
dlferp
dlwage
dlorpr
dlorq
1
0
0
0.072
0
0.097
2
-0.052
0.042
0.039
-0.107
-0.026
Responses to Shock in orange price
Period
dlgrpr
dlferp
dlwage
dlorpr
dlorq
1
0
0
0
0.477
-0.038
2
0.124
0.023
0.042
-0.201
0.210
Notes: The prefix 'd' to the variable names
denotes change in the variable. All variables
previously defined.
Table 4.22: Variance Decomposition Percentage of One-period
and Three-period Forecast Error Variance—Orange VAR.
Forecast error
variance in:
Percentage forecast error
explained by shocks in:
Grapefruit price
dlgrpr
dlferpr
dlwage
dlorpr
dlorq
1 Yr
100
0
0
0
0
3 yrs
84.559
2.263
1.511
10.961
0.705
Fertilizer price
1 Yr
0
100
0
0
0
3 yrs
4.577
83.741
7.159
4.251
0.272
Wage
1 Yr
0
0
100
0
0
3 yrs
12.886
22.566
50.088
13.979
0.480
Orange price
1 Yr
0
0
0
100
0
3 yrs
2.306
5.298
3.512
83.017
5.867
Orange quantity
1 Yr
6.461
1.110
16.195
2.465
73.769
3 yrs
21.606
1.548
5.842
43.470
27.534

126
Table 4.23: Responses of Cocoa-bean Quantity to
Shocks in the Cocoa-bean VAR.
Responses to Shock in banana price
Period
dlbanpr
dlferp
dlwage
dlcobpr
dlcobq
1
0.225
0
0
0
0.028
2
0.013
-0.048
0.022
-0.030
-0.005
Responses to Shock in fertilizer price
Period
dlbanpr
dlferp
dlwage
dlcobpr
dlcobq
1
0
0.146
0
0
0
2
-0.015
0.025
-0.056
0.093
0.006
Responses to Shock in wage
Period
dlbanpr
dlferp
dlwage
dlcobpr
dlcobq
1
0
0
0.072
0
0.031
2
-0.011
0.027
0.037
0.073
-0.046
Res
ponses to Shock in cocoa-bean price
Period
dlbanpr
dlferp
dlwage
dlcobpr
dlcobq
1
0
0
0
0.275
0.007
2
0.032
-0.031
-0.004
-0.083
-0.062
'Jotes: The prefix 'd' to the variable names
denotes change in the variable. All variables
previously defined.
Table 4.24: Variance Decomposition Percentage of One-period
and Three-period Forecast Error Variance—Cocoa-bean VAR.
Forecast error
variance in:
Percentage forecast error
explained by shocks in:
Banana price
dlbanpr
dlferpr
dlwage
dlcobpr
dlcobq
1 Yr
100
0
0
0
0
3 yrs
90.042
1.424
0.760
2.612
5.163
Fertilizer price
1 Yr
0
100
0
0
0
3 yrs
8.454
81.851
3.994
3.370
2.331
Wage
1 Yr
0
0
100
0
0
3 yrs
11.070
35.763
48.791
2.466
1.910
Cocoa bean price
1 Yr
0
0
0
100
0
3 yrs
0.857
11.874
5.585
77.946
3.738
Cocoa-bean
quantity
1 Yr
2.026
0
2.451
0.141
95.383
3 yrs
1.588
0.133
6.502
11.047
80.730

127
Téúale 4.25: Responses of Potato Quantity to Shocks
in the Potato VAR.
Responses to Shod
c in cassava price
Period
dlcaspr
dlferp
dlwage
dlpotpr
dlpotq
1
0.125
0
0
0
-0.026
2
0.045
0.049
-0.036
0.050
-0.028
Responses to Shock in fertilizer price
Period
dlcaspr
dlferp
dlwage
dlpotpr
dlpotq
1
0
0.148
0
0
0.049
2
-0.037
0.001
-0.046
-0.020
-0.012
Responses to Shock in wage
Period
dlcaspr
dlferp
dlwage
dlpotpr
dlpotq
1
0
0
0.062
0
-0.050
2
0.003
0.025
0.030
0.008
0.021
Responses to Shock in potato price
Period
dlcaspr
dlferp
dlwage
dlpotpr
dlpotq
1
0
0
0
0.187
-0.194
2
-0.017
-0.030
0.039
-0.029
0.177
Notes: The prefix 'd' to the variable names
denotes change in the variable. All variables
previously defined.
Téüale 4.26: Variémce Decomposition Percentage of One-period
and Three-period Forecast Error Variémce—Potato VAR.
Forecast error
variance in:
Percentage forecast error
explained by shocks in:
Cassava price
dlcaspr
dlferpr
dlwage
dlpotpr
dlpotq
1 Yr
100
0
0
0
0
3 yrs
90.133
7.985
0.168
1.591
0.123
Fertilizer price
1 Yr
0
100
0
0
0
3 yrs
8.889
84.140
2.715
3.706
0.550
Wage
1 Yr
0
0
100
0
0
3 yrs
19.565
19.787
36.307
22.798
1.543
Potato price
1 Yr
0
0
0
100
0
3 yrs
6.058
1.034
0.269
89.234
3.405
Potato quantity
1 Yr
0.714
2.626
2.711
40.583
53.367
3 yrs
1.485
2.440
2.459
54.282
39.334

128

1
dlbanpr
\
dfferp
\
dlwage
\
dlcofpr
\
dlcofq
0.4
0 2 -
0.0
-0.2 —1 1 1 1 1 1 1 1 1 . r
0 2 4 6 8 10
Figure 4.21: Impulse Responses to an
Innovation in Wage—Coffee VAR.
1.0
0.8
0.6
0.4
0.2
0.0
-0.2
Figure 4.22: Impulse Responses to an
Innovation in Coffee Price—Coffee VAR.
129

Innovation in Banana Price—Pimento VAR.
VAR.
130

131

Innovation in Yam Quantity—Yam VAR.
Innovation in Cocoa-bean Quantity—Cocoa-
bean VAR.
Innovation in Orange Quantity—Orange VAR.
Innovation in Potato Quantity—Potato VAR.
132

Innovation in Fertilizer Price—Orange VAR.
133

VAR.
VAR.
134

Innovation in Potato Price--Potato VAR.
135

136
4.6 Summary
This chapter provides empirical evidence of long-run
equilibrium or cointegrating relationships between crop
output and price incentives for eight crops grown in
Jamaica. It was found that the data series relevant to the
estimation of the long-run relationships are non¬
stationary. This statistical quality of the data has not
been taken into consideration in previous studies on crop
supply responses in Jamaica's agricultural sector. To deal
with this data problem, an error correction modeling
approach based on cointegration theory was adopted, within
an estimation framework developed by Johansen (1988) .
The results from the estimation found low own-price
elasticities as well as low adjustment coefficients.
Additionally, impulse functions and forecast error
variances were estimated in order to analyze the short-run
dynamics of the output-price incentive relationships. It
was found that overall, it appears that short-run crop
responses are low, and respond more to crop price
variations than on input-price changes. Slow adjustments in
the short-run may be a sign of constraining government
regulations, inadequate supportive infrastructure, lags
between planting and harvesting, and other reasons.

CHAPTER 5
SUPPLY RESPONSE AND COUNTERFACTUAL ANALYSIS
In this chapter the before-after approach and counter-
factual analysis are used to assess the impact of the
economic reforms on crop supply response in Jamaica. The
before-after analysis is done in Section 5.1, and the
counterfactual analysis in Section 5.2.
5.1 Before-After Analysis
The before-after approach compares the performances of
target economic variables during a particular program
regime with their performances prior to the program. The
approach assumes that any changes in the variables in the
reform period are attributed solely to the reforms in that
period. Two before-after estimators, which have been used
extensively in the literature to evaluate Fund/Bank
economic reform programs, are the mean change in the target
variable and its variance (Khan, 1990) . The mean estimator
can be expressed as AY = Y2 - Yi, where Y2 and Yi are the
means of the variable in the reform and pre-reform periods,
respectively. If AY > 0, this is interpreted to mean that
the reforms impacted positively on the variables.
137

138
Price-variability is
measured
by the
coefficient
of
variation, expressed
s
as CV = —,
X
where s,
and x are
the
standard deviation
and mean
respectively. A brief
justification for using these estimators is warranted here.
One of the primary aims of the economic reforms in
agriculture is to provide farmers with prices that are
remunerative enough to induce output growth. Indeed, the
observation that government interventions have been
historically biased against agriculture (Schiff and Valdés,
1992a) , led naturally to the belief by the IMF and World
Bank that
the
economic reforms would reduce
government
distortion
of
prices, and
improve price and
non-price
incentives
to
agriculture
(World Bank, 1996)
In this
regard, it seems logical to investigate whether the reforms
have ushered-in stimulative prices in agriculture.
However, a related but antecedent question is what
constitutes a stimulative price shift? Given the biological
lag between planting and harvesting in agriculture, and the
observation that in developing countries few farmers can
fully mitigate against temporal price risk because of lack
of resources, modern microeconomic theory provides little
predictive confidence on agricultural production response
to a given shift in the price distribution. However, micro-

139
economic theory of the firm under price uncertainty does
postulate a positive relationship between output and the
mean of output price, and a negative relationship between
output and the variance of output price (Baron, 1970;
Sandmo, 1971; Ishii, 1977).
The issues raised above therefore justify an empirical
examination of the first two moments of the price and
quantity series used in the crop supply response models. In
order to facilitate comparative analysis of variability
among the different crops, the coefficient of variation is
used instead of the variance. As mentioned in chapter 1,
the period 1962-1979 is considered as the pre-reform
period, and 1980-1999 the reform period. Two sets of
estimates of the means and coefficients of variation for
all eight crops' output and prices are computed. The first
set covers the entire pre-reform and the reform periods,
and the second set uses observations from the last five
years in the pre-reform, and the first five years in the
reform periods, respectively. These estimates are reported
in Tables 5.1 and 5.2, and are robust with respect to the
time periods used.
With respect to quantity, Table 5.1 shows that the
mean output was higher in the reform period for four of the
eight crops analyzed. These are coffee, pimento, yam and

140
TéJale 5.1: Descriptive Price and Quantity Statistics
-1962-1979 and 1980-1999.
Variable
Pre-Reform Period
1962-1979
Reform Period
1980-1999
Mean
CV
Mean
CV
banq
banpr
134.771
2.978
0.189
0.398
98.274
8.436
0.086
0.242
sugq
sugpr
175.214
0.419
0.204
0.169
93.300
5.542
0.088
0.577
cofq
cofpr
70.603
41.021
0.218
0.219
92.263
21.485
0.309
0.229
pimq
pimpr
189.586
63.750
0.197
0.429
339.261
21.555
0.148
0.238
yamq
yampr
74.936
7.297
0.357
0.141
135.259
9.227
0.226
0.593
orq
orpr
142.996
1.575
0.331
0.364
135.266
11.000
0.315
0.967
cobq
cobpr
104.918
19.774
0.133
0.464
120.861
16.600
0.240
0.494
potq
potpr
162.886
8.654
0.237
0.212
152.036
12.390
0.307
0.382
ferp
wage
9.432
283.077
0.219
0.191
8.215
246.531
0.214
0.190
Note: CV = s/x, is the coefficient of variation,
where, s and x are the standard deviation and
mean respectively. All other variables as
previously defined.
cocoa-bean. These increases in output are likely due to the
government expansion programs in these crop activities,
such as coffee and cocoa-bean rehabilitation schemes by the
government, and to the stimulus of increased prices and
export opportunities following the liberalization of prices
in the reform period. With regards to the crops whose mean
output did not increase over the reform period (banana,
sugar, orange and potato), this is likely due to the

141
Table 5.2: Descriptive Price and Quantity Statistics
—1975-1979 and 1980-1984.
Variable
Last 5 years of
Pre-Reform Period
1975-1979
First 5 years of
Reform Period
1980-1984
Mean
CV
Mean
CV
banq
105.833
0.118
109.286
0.055
banpr
4.428
0.186
7.771
0.393
sugq
141.758
0.130
89.588
0.094
sugpr
0.482
0.161
1.525
0.592
cofq
62.004
0.245
76.606
0.189
cofpr
48.084
0.165
21.103
0.152
pimq
193.388
0.241
214.163
0.425
pimpr
40.569
0.257
20.414
0.035
yamq
102.466
0.107
100.227
0.087
yampr
8.513
0.147
10.572
0.123
orq
94.350
0.145
78.325
0.017
orpr
1.116
0.591
1.061
0.089
cobq
94.169
0.095
125.868
0.207
cobpr
25.660
0.735
24.911
0.071
potq
162.018
0.266
133.925
0.321
potpr
10.746
0.112
11.568
0.139
ferp
10.020
0.246
9.175
0.219
wage
328.513
0.119
233.430
0.036
Notes: CV = s/x , is the coefficient of variation,
where, s and x are the standard deviation and
mean respectively. All other variables as
previously defined.
persistence of structural factors, which slow down the
adjustment process of output to policy changes.
Table 5.1 also shows that mean real prices for five
crops (banana, sugar, yam, orange, and potato), increased
over the reform period. The exceptions are coffee, pimento
and cocoa-bean. However, real price-variability appears to
have increased for most crop prices over the reform period.
The exceptions are banana and pimento. Thus,
the

142
coefficient of variation for sugar price was 17 percent in
the pre-reform period, but increased to 58 percent in the
reform period. Similar high coefficients of variations are
recorded for other crop prices. The observations made above
in connection to the information in Table 5.1 are in
general agreement with the information in Table 5.2.
In addition to the information in Tables 5.1 and 5.2,
Figures 5.1-5.8 show changes in nominal and real crop
prices over the period 1962-1999. For most crops, nominal
prices shifted upwards from the late 1970s and early 1980s,
but the most dramatic increases are observed from the late
1980s onwards.
While nominal prices have been increasing in the
reform period, real prices show mixed changes as shown in
Figures 5.5-5.8. Real prices for coffee and pimento have
declined steadily over the 1962-1999 period. Some price
stability and windfall gains have been observed for cocoa-
bean and potato prices but they seem to be on the descent,
since 1987, below levels observed in the earlier years. The
real prices for yam and orange increased briefly after the
announced price liberalization policies by the government
in 1985. By 1994 both prices were on the down turn but
trending above the pre-liberalized price regime. Finally,
real prices for banana and coffee showed dramatic increases

Figure 5.1: Nominal Price Changes—Banana
(BANP) and Sugar (SUGP).
143

Figure 5.5: Real Price Changes—Banana and
Sugar (1980=100).
and Potato (1980=100).
144

145
in the early 1980s, and while sugar price continued upwards
until 1991, banana price peaked in 1985, fluctuated
somewhat, and along with sugar, have been on a decline,
with some evidence of an upturn by 1998.
In effect, therefore, these observations suggest that
the pro-competitive effects that were expected to accompany
the reforms as well as from the demise of state monopolies,
may have outweighed the stability impulses of controlled
prices of the pre-reform era.
5.2 Counterfactual Analysis
One of the premises upon which the long-run supply
elasticities reported in chapter 4 is based, is that the
recent economic reforms in Jamaica influenced the variation
of the economic time series which are included in the
various crops' error correction models. On the basis of
this premise, the long-run crop supply elasticities
reported in chapter 4 could be interpreted as what has
happened to supply response under the economic reforms.
However, the true effect of the economic reforms on
crop supply response must be based upon a comparison of the
supply response estimates in chapter 4 with estimates
derived under the assumption that the economic reforms,
which were observed over the period 1980-1999 did not
occur. Crop supply response under this assumption would

146
constitute a counterfactual, against which the supply
response estimates in chapter 4 should be compared. A
relevant issue, in this regard, is what would have been the
policies pursued by the government if the reforms were not
undertaken? Answers to this question would provide the
counterfactual situations.
The most manageable way of constructing a counter-
factual using time series data, is to assume that the
policies, which the government was pursuing over the 1962-
1979 period, continued over the 1980-1999 period. Under
this assumption, the time series modeling technique would
forecast the 1980-1999 period by extrapolating techniques,
which incorporate no information other than that available
in the past. A central issue, however, is whether the
forecasts should be made on individual series or as a
system. Since in each crop supply response ECM the
variables are shown to be in long-run equilibrium, it seems
justifiable to use a systems approach to forecasting the
economic time series. As such, a VAR approach is utilized
and in each VAR the stochastic variables are the same as
those that were included in the supply response ECM in
chapter 4.
The forecasts of the economic time series are shown in
Figures 5.9-5.28. Although the graphs show clear distinct-

147
ions between the actual and forecasted variables, for some
crops the two sets of series show mixed results. For
example, Figure 5.9 shows that the forecast of banana
quantity is above the actual series over the 1980-1999
period (the exception is in 1983 when actual and forecast
quantities are equal). Similarly, Figure 5.11 shows that
actual price of sugar has been above the forecast for the
entire 1980-1999 period. Given the assumption under which
the forecasts were made, this suggests that, in the case of
Figure 5.9, if past (1962-1979) policies had continued over
the 1980-1999 period, banana output would have been higher
than what actually prevailed during the period of economic
reforms. This would have been the case despite the fact
that the forecasted banana price is lower than actual
price.
The one-step ahead forecasts shown in Figures 5.9-5.28
suffer from two short-comings. First, they reflect the data
trend of the last few years before the end of 1980.
Further, the forecasts are smooth and do not reflect the
fluctuations in the actual series. Finally, the forecasts
were plagued with large standard errors, and therefore
provide only a possible indication rather than a precise
picture of the counterfactuals.

Figure 5.9: Banana Quantity—Actual (LBANQ)
and Forecasted (FOLBANQ).
and Forecasted (FOLSÃœGPR).
148

Figure 5.13: Fertilizer Price—Actual
(LFERP) and Forecasted (FOLFERP).
Figure 5.15: Coffee Quantity—Actual
(LCOFQ) and Forecasted (FOLCOFQ).
Figure 5.14: Sugar Quantity—Actual (LSUGQ)
and Forecasted (FOLSUGQ).
and Forecasted (FOLCOFPR).
149

Figure 5.17: Pimento Quantity—Actual
(LPIMQ) and Forecasted (FOLPIMQ).
Figure 5.19: Yam Quantity—Actual (LYAMQ)
and Forecasted (FOLYAMQ).
and Forecasted (FOLPIMPR).
Figure 5.20: Yam Price—Actual (LYAMPR) and
Forecasted (FOLYAMPR).
150

Figure 5.21: Cassava Price—Actual (LCASPR)
and Forecasted (FOLCASPR).
and Forecasted (FOLORPR).
Figure 5.22: Orange Quantity--Actual (LORQ)
and Forecasted (FOLORQ).
151

Figure 5.25: Cocoa-bean Quantity—Actual
(LCOBQ) and Forecasted (FOLCOBQ).
and Forecasted (FOLORPR).
Figure 5.26: Cocoa-bean Price—Actual
(LCOBPR) and Forecasted (FOLCOBPR).
Figure 5.28: Grapefruit Price—Actual
(LGRPR) and Forecasted (FOLGRPR).
152

153
A second approach was used to construct a
counterfactual in order to assess the impact of the
economic reforms in Jamaica. In this approach, the actual
observations were divided into two separate sub-samples,
1962-1979 and 1980-1999, corresponding to the pre-reform
and reform periods, respectively. Supply functions for each
crop were estimated for each of the two sub-samples. The
estimation procedure followed the Johansen method, which
was outlined in chapter 4. Fitted values from the 1962-1979
period were then forecasted over the 1980-1999 period and
these forecasts were used as the counterfactuals against
which the fitted values from the 1980-1999 supply functions
were compared graphically.
With respect to the forecasting exercise, the BOXJENK,
CORRELATE, and FORECAST instructions in the RATS computer
software were used to identify, estimate and forecast the
fitted series which were used as the counterfactual series.
After some experimentation, the forecasts were modeled as
mixed autoregressive-moving average processes, denoted as
ARMA(p, q) , where p and q are the autoregressive and moving
average orders, respectively (Pindyck and Rubinfeld, 1991).
Following the Box-Jenkins methodology (Enders, 1995;
Doan, 1996) , the orders of p and q were determined by
examining the sample autocorrelation functions (ACF),

154
partial auto-correlations functions (PACF), and graphs of
the fitted series. For each series, several models were
fitted to the data and the final choice was determined on
the basis of parsimony, and various diagnostic statistics.
The test criteria used are the root-mean-square error
(RMSE), Theil's Inequality Coefficient (Theil U) , and the
Ljung-Box Q-statistics. Low RMSE is a desirable quality
from a forecasting model. The Theil U statistic is
independent of the units of measurement. It is the ratio of
the RMSE for the forecast model to the RMSE for a "naive"
forecast of no change in the dependent variable from the
previous value (Doan, 1996) . Values of Theil U that are
less than one indicate that a model performs better (i.e.,
has a lower RMSE) than the naive model. Doan (1996) claims
that a Theil U of 0.8 or less is reasonable for a
univariate forecast model.
The values of (p, q) were determined by the ACF and
PACF. For an ARMA(p, q) process, the ACF begins to decay at
lag q, and the PACF begins to decay at lag p. In addition
to the ACF and PACF, the Ljung-Box Q-statistic was used in
model selection. The Q-statistic is a chi-square (%2) test
used to test the null hypothesis of no significant
autocorrelations. The diagnostic statistics for the fore¬
casts of fitted values are reported in Table 5.3.

155
Table 5.3: Diagnostic Statistics for Forecasts
of Fitted Values.
Crop
RMSE
Theil U
Q (2)a
Q(4)a
Banana
0.167
0.353
3.397
3.963
Sugar
4.799
0.524
0.062
0.266
Coffee
0.378
0.880
0.061
0.368
Pimento
0.094
0.838
2.304
3.559
Yam
0.190
0.538
0.177
3.596
Orange
0.710
0.598
2.324
5.878
Cocoa-bean
5.769
1.188
0.902
4.558
Potato
0.325
0.262
1.957
3.095
a Ljung-Box Q-statistics. %] and X4 critical
values at 95% level are 5.99 and 9.49 respectively.
RMSE = Root Mean Square Error.
The estimated a and 3 vectors for each crop for the
two sample periods are reported in Tables 5.4 and 5.5,
respectively. From Table 5.4 the most economically
plausible equations (i.e., equations that meet a priori
(sign) expectations) were selected to generate fitted
values for the dependent variable in each crops' ECM for
the 1962-79 period. For example, the fitted values for
banana for the 1962-79 period were estimated as follows:
BANFIT = .300 + 0.781*Po - 0.242*Pa - 0.758*W
- 0.854*F
where all coefficients are taken from the CE #1 relation
for 3 for banana in Table 5.4. These fitted values were
then forecasted, using the methodology described above,
over the 1980-99 period, and used as the counterfactual for
banana.

156
In a manner similar to that described above, the
fitted values for each crops' ECM were estimated for the
1980-99 period. For example, the fitted values for banana
for the 1980-99 period were estimated as follows:
BANFITA = -4.140 + 0.172*Po - 0.103*Pa - 0.040*W
- 0.003*F
where all coefficients are taken from the CE #2 relation
for p for banana in Table 5.5. The 'A' appended to
'BANFITA' is used to denote that the actual observations,
Po, Par W and F, were used to estimate the fitted series.
The two sets of fitted values, the (forecasted)
'counterfactual' and 'actual', for each crop, were then
plotted on the same graph. These graphs are shown in
Figures 5.29-5.36. From a visual inspection of those graphs
the following observations are noted. First, banana is the
only crop in the sample for which the fitted values from
the reform period are clearly above the fitted values for
the counterfactual over the entire 1980-1999 period.
Excluding the years 1987-1988, this observation is also
applicable for pimento. The interpretation of this
observation, based upon the assumptions that have been used
to generate the fitted and forecasted series, is that if
there were no change in the policy regime in Jamaican over

157
Table 5.4: Estimated Long-run and Adjustment Coefficients,
P's, a's for all Crops, 1962-1979.
Crops
Q
Po
Pa
W
F
C
Banana
CE #1
P's
1
0.781
-0.242
-0.758
-0.854
0.300
a's
-0.110
(-1.711)c
-0.267
(-1.372)°
0.368
(2.018)b
0.174
(4.691)a
-0.008
(-0.084)
CE #2
P's
1
0.196
0.017
1.385
-0.170
-12.477
a's
-0.140
(-1.560)c
0.166
(0.611)
0.327
(1.288)
-0.-109
(—2.122)b
0.560
(4.270)a
Sugar
P's
1
0.195
-0.654
-0.282
-0.697
-2.053
a's
-0.153
(-1.792)c
-0.542
(-2.327)b
0.276
(1.099)
0.209
(3.929)a
0.123
(0.763)
Coffee
CE #1
P's
1
0.516
-0.176
-0.310
-0.108
-7.862
a's
-0.350
(-1.597)°
-0.242
(-1.040)
1.052
(5.714)a
0.089
(1.118)
0.067
(0.604)
CE #2
P's
1
-0.778
-0.101
-0.980
2.325
-0.938
a's
-0.066
(-0.880)
0.082
(1.039)
0.065
(1.035)
0.011
(0.393)
-0.190
(-5.061)a
Pimento
CE #1
P's
1
-0.018
-0.070
0.757
0.261
-9.838
a's
-0.846
(-4.823)3
0.346
(1.956)b
0.145
(0.622)
-0.008
(-0.127)
-0.371
(-3.379)a
CE #2
P's
1
0.364
-0.640
-0.685
-1.397
-14.232
a's
-0.391
(-2.821)3
-0.631
(-4.509)a
0.253
(1.375)°
0.051
(1.045)
0.114
(1.317)
Yam
CE #1
P's
1
-3.303
5.672
1.637
1.685
-30.596
a's
-0.083
(-2.914)3
0.135
(2.284)b
-0.119
(-3.671)a
-0.000
(-0.005)
-0.111
(-2.373)b
CE #2
P's
1
1.879
-1.587
-1.223
-2.433
-27.495
a's
-0.134
(-4.415)a
-0.016
(-0.261)
0.043
(1.260)
0.910
(0.385)
0.090
(1.814)b
Orange
CE #1
P's
1
2.356
-3.034
-2.910
-0.949
15.925
a's
-0.062
(-0.897)
0.227
(1.549)°
-0.086
(-0.502)
0.028
(1.327)
0.142
(4.666)a
CE #2
P's
1
0.723
-1.417
0.570
-1.791
-3.193
a's
-0.203
(-1.916)b
0.210
(0.939)
-0.169
(-0.649)
-0.030
(-0.946)
0.123
(2.662 b

158
Table 5.4: Cont'd.
Crops
Q
Po
Pa
W
F
C
Cocoa-
beam
CE #1
P's
1
-0.313
0.035
0.603
1.018
-9.430
a's
-1.218
(-4.036)3
1.270
(2.750)b
-0.771
(-1.848)b
-0.055
(-0.448)
-0.257
(-1.756)°
CE #2
P's
1
2.802
-1.204
-3.546
-0.148
8.720
a's
-0.033
(-0.994)
-0.112
(-2.227)b
0.022
(0.479)
0.003
(0.244)
0.103
(6.455)*
Potato
CE #1
P's
1
1.396
-2.698
0.820
-1.183
-6.713
-0.099
(-0.653)
-0.018
(-0.201)
0.212
(5.015)a
-0.006
(-0.218)
0.242
(4.597)a
CE #2
P's
1
1.592
-7.377
-4.172
-3.103
31.003
a's
-0.048
(-0.667)
-0.034
(-0.825)
0.034
(1.726)b
0.044
(3.386)a
0.055
(2.226)b
Note: Figures in parentheses are t-values.
CE indicates 'cointegration equation'.
a, b, c, indicate statistical significance at one, five and
ten percent levels, respectively.
Q = quantity; P0 = own-price; Pa = price of the alternative
crop; W = average wage rate in agriculture sector; F =
average price of fertilizer; C = constant (intercept).

159
Table 5.5: Estimated Long-run and Adjustment Coefficients,
P's, a's for all Crops, 1980-1999.
Crops
Q
Po
Pa
W
F
C
Banana
CE #1
P's
1
-0.249
0.255
-0.006
0.921
-6.383
a's
-0.135
(-1.534)
0.093
(0.230)
-1.942
(-3.814)3
0.344
(2.532)b
-0.805
(-3.891)3
CE #2
P's
1
0.172
-0.103
-0.040
-0.003
-4.140
a's
-1.047
(-4.456)3
-2.308
(-2.150)b
1.043
(0.769)
0.310
(0.858)
0.559
(1.015)
CE #3
P's
1
0.306
0.069
0.029
0.916
-7.344
a's
-0.064
(-2.788)b
-1.064
(-2.880)3
-0.176
(-0.376)
-0.162
(-1.301)
-1.122
(-5.918)3
Sugar
CE #1
P's
1
0.255
-0.249
-0.006
-0.921
5.879
a's
-0.135
(-2.534)b
-1.942
(-3.814)3
0.093
(0.230)
0.344
(2.5 32)b
-0.805
(-3.891)3
CE #2
P's
1
0.103
-0.172
-0.040
-0.003
-3.636
a's
-1.047
(-4.456)3
-1.043
(-0.769)
2.308
(2.150)b
0.310
(0.858)
0.559
(1.015)
CE #3
P's
1
0.069
0.306
0.029
0.916
-6.839
a's
-0.064
(-0.788)
-0.176
(-0.376)
-1.064
(-2.880)3
-0.162
(-1.301)
-1.122
(-5.918)3
Coffee
CE #1
P's
1
-1.140
2.259
0.956
6.905
-25.382
a's
-0.127
(-1.792)c
0.087
(1.467)
-0.143
(-2.749)b
-0.018
(-0.600)
-0.193
(-7.023)3
CE #2
P's
1
1.108
-0.231
-0.440
-0.116
-6.146
a's
-0.016
(-0.066)
-0.711
(-3.460)3
0.717
(3.972)3
0.031
(0.301)
0.223
(2.339)b
Pimento
CE #1
P's
1
-0.190
1.410
0.907
1.680
-15.165
a's
-0.048
(-0.446)
0.286
(4.779)3
-0.380
(-3.760)3
-0.015
(-0.281)
-0.127
(-2.Ill)b
CE #2
P's
1
0.282
-0.406
-1.181
-0.388
-9.764
a's
-0.769
(-5.208)3
-0.096
(-1.186)°
0.087
(2.629)b
0.021
(0.293)
0.010
(0.125)

160
Table 5.5: Cont' d.
Crops
Q
Po
Pa
W
F
C
Tam
CE #1
P's
1
2.543
-3.168
-0.052
-0.543
-7.369
a's
-0.131
(-2.183)b
-0.391
(-3.944)a
0.030
(2.464)b
0.075
(1.301)
0.012
(0.100)
CE #2
P's
1
0.356
-0.248
1.805
0.503
-18.315
a's
-0.253
(-3.344)*
-0.744
(-5.960)a
0.384
(4.703)a
-0.179
(-2.470)*
-0.193
(-1.246)
CE #3
p's
1
-0.301
0.469
0.122
0.980
-11.136
a's
-0.336
(-3.850)a
0.564
(3.928)*
-0.099
(-1.049)
-0.024
(-0.286)
-0.453
(-2.533)b
Orange
CE #1
P's
1
-5.652
4.301
-4.485
-4.097
32.444
a's
-0.047
(-1.001)
0.089
(0.816)
-0.306
(-5.225)a
0.059
(2.889)*
0.049
(1.496)
CE #2
P's
1
0.224
-0.477
-0.115
-1.609
-0.501
a's
-0.271
(-1.789)°
-1.347
(-3.846)a
1.369
(7.267)a
0.017
(0.260)
0.237
(2.240)b
CE #3
p's
1
0.050
-0.307
1.185
0.468
1.084
a's
-0.197
(-1.147)
-0.576
(-1.446)
0.453
(2.114)b
0.049
(0.655)
-0.527
(-4.348)*
Cocoa-
bean
CE #1
P's
1
0.016
-1.159
-0.505
-2.212
-0.353
a's
-0.153
(-0.884)
-0.384
(-2.326)b
0.432
(2.107)b
0.098
(1.027)
0.603
(4.688)*
CE #2
P's
1
-0.302
-0.362
1.106
0.471
-10.230
a's
-1.831
(-5.336)a
0.558
(1.708)°
0.532
(1.312)
-0.179
(-0.952)
-0.216
(-0.851)
Potato
CE #1
P's
1
0.971
-1.274
-2.240
-0.702
17.404
a's
-0.081
(-0.942)
-0.083
(-1.295)
0.136
(5.318)*
0.032
(1.928)°
0.049
(1.019)
CE #2
P's
1
-0.008
0.612
0.030
0.303
-6.541
a's
-0.917
(-1.239)
0.472
(0.855)
-0.225
(-1.016)
-0.179
(-1.263)
-0.138
(-0.331)
Note: Figures in parentheses are t-values.
CE indicates 'cointegration equation'.
a, b, c, indicate statistical significance at one, five and
ten percent levels, respectively.

(BANFITC).
Figure 5.31: Coffee—Fitted Output for
Reform Period (COFFITA) and Counterfactual
(COFFITC).
(SUGFITC).
(PIMFITC).
161

Figure 5.33: Yam—Fitted Output for Reform
Period (YAMFITA) and Counterfactual
(YAMFITC).
Reform Period (COBFITA) and Counterfactual
(COBFITC).
(ORFITC).
(POTFITC).
162

163
the 1980-1999 period, then the output of banana and
pimento, with given price incentives, would have been lower
than that which were actually observed under the reform
policy regime. In other words, it appears that the reforms
impacted positively on these two crops over the 1980-1999
period.
Second, for some crops there are two distinct periods
over 1980-1999 in which the fitted values from the reform
period are either above or below the counterfactual. For
example, over the 1980-1992 period the fitted values for
sugar for the reform period were below those for the
counterfactual. The reverse is observed for the series
since 1992. Similarly, fitted values for coffee for the
reform period have been above the counterfactual over the
1980-1995 period but the situation reversed after 1995.
Similar observations can be made for orange and cocoa-bean.
Finally, the cases of yam and potato are different
from the two patterns observed for the other crops. Yam
shows an oscillating pattern. Fitted values for the
counterfactual were below fitted values for the reform
period over 1980-1985. Potato showed similar oscillations
but with relatively wider gaps within each oscillation.
In effect, therefore, based on the counterfactual
analysis, the data seem to suggest that the impact of the

164
reforms are crop and time specific. For some crops it would
appear that the reforms impacted negatively on output
trends in the 1990s. This is the case of coffee, orange,
cocoa-bean and potato. For banana and pimento the reforms
appear to impact positively on output responses. For yam
and potato, similar straightforward conclusions cannot be
made since the reforms appear to impact positively on
output response in some periods and negatively in others.
5.3 Summary
Comparative analysis of crop output and prices between
the pre-reform and reform periods suggests a mixed record
regarding the impacts of economic reforms in Jamaican
agriculture. For some crops, mean output increased under
the reform periods, while for others, output contracted.
Clear evidence of greater price variability is observed in
the reform period compared to the pre-reform era. The
counterfactual analysis suggests that the effects of the
reforms are crop and time dependent.
A

CHAPTER 6
CONCLUSIONS AND POLICY IMPLICATIONS
Jamaica embarked upon a process of economic policy re¬
orientation in the late 1970s, at a time when the economy
was showing signs of acute economic crisis. Mainly the IMF
and World Bank sponsored the economic reforms in
consultation with the Jamaican government. Although, at
various points in time, there were clear signs of
opposition by the government towards the economic reforms
emanating from the Bretton Woods Institutions, by 1990,
Jamaica had, to a large degree, embraced most of the reform
policies.
Essentially pragmatic, the reforms highlighted the
need for a more outward orientation of the Jamaican
economy. Much emphasis has been placed on being competitive
in world markets and also upon relying on market forces to
determine prices and to allocate increasingly scarce
capital resources. The reforms also entailed a critical re¬
assessment of the role of the state in terms of its size,
overall efficiency, and its relationships with the private
sector. As a result of these considerations, topics such as
165
A

166
deregulation, privatization and economic liberalization
acquired increased attention in the policy debates over the
past two decades.
The reforms that began in the late 1980s engendered an
entirely new economic environment for agriculture. However,
it posed numerous challenges to economic agents such as
small farmers and other agricultural producers generally,
who had hitherto operated under various subsidies and
guaranteed marketing and pricing schemes. In this regard,
one government agency reports that
Since the liberalization of the economy and the sub¬
sequent restructuring of government support for
essential services in the agricultural sector, farmers
have had some problems remaining competitive.
(Jamaica, Planning Institute of Jamaica, 1996, p.7.12)
The quotation above is instructive insofar as it may
have identified an important aspect of economic reality in
Jamaica, namely, the difficulty of farm producers to make a
successful transition between policy regimes, which in many
respects make differential demands on economic actors. This
conclusion achieves some degree of plausibility in light of
the observation that the transformation or re-ordering of
economic policy priorities constitutes a fundamental
departure from the basic policy directives that had been
adopted in Jamaica from the 1960s through to the late
1970s. These policies encouraged inward looking industrial-

167
ization; state-sponsored import substitution and national¬
ization of foreign enterprises; extensive use of price
controls and subsidies; and, generally, a relatively strong
populist approach to the solution of poverty, low income,
unemployment, and other socio-economic problems.
One of the principal tasks of this study has been to
investigate the impact of the economic reforms on crop
supply response in Jamaica. The estimates on crop supply
responses presented in this study suggest that long-run
equilibrating relationships exist which link crop output
and price incentives for the eight crops studied. This
result has important policy implications insofar as it
indicates a significant relationship between agricultural
supply and price incentives. In particular, and on the
basis of the analytical method used in this study, the link
between output and price incentives means that these
variables move together over time and more importantly,
respond to the same shocks in the system, albeit with
varying degrees. Further, as markets become more
competitive, this link becomes stronger since market
signals are transmitted more effectively and efficiently.
Although economic theory is generally silent as to
what constitutes a stimulative price shift for agricultural
producers, the econometric evidence presented in this study

168
indicates a positive relationship between output supply and
own-price. Crops such as banana and sugar, whose prices are
still determined by preferential agreements, enjoyed
increased real prices over the reform period. Jamaican
sugar and banana have guaranteed access to the European
Union (EU) market via the Lome Conventions. The prices paid
for Jamaican exports in this market are significantly
higher than world market prices. However, the EU's Common
Agricultural Policy (CAP) requires progressive reductions
of subsidies to African Caribbean and Pacific (ACP)
countries. This is expected to reduce the prices paid for
ACP exports.
In addition, with respect to bananas, EU's
arrangements with the ACP countries are constantly being
challenged by Central and Latin American (Dollar) banana
countries. Proposals are for the individual ACP quotas to
be replaced by a single global ACP group quota. While the
tariff preference is expected to continue in the future, it
is under constant review. The main implication is that
Jamaican farmers must reduce cost of production in order to
compete in the EU market.
The twin issues of reducing cost of production and
achieving international competitiveness appear to be major
goals in the government's National Industrial Policy which

169
was tabled in 1996. In this regard, J$750 million in
international funding was sourced and allocated to various
projects in response to the frequent debates in the EU to
reduce preferential treatments of ACP exports (Jamaica,
Planning Institute of Jamaica, 1998) . In the case of
banana, various projects have been implemented since the
mid-1990s. These include programs for improved management
of water systems and the banana disease, Sigatoka. For
small and medium size farms, high cost and unavailability
of inputs are major constraints to improving productivity.
For these farmers the government has invested J$38 million
to set up a supply store and has established a revolving
loan scheme for improvements in irrigation and other
infrastructural works. In addition, other projects have
been implemented which aim at improving banana quality,
such as experiments in de-handling, deflowering, bunch
hardening, and packaging and storage.
There is strong evidence to suggest that there are
considerable constraints in the economic system, which slow
down the adjustment process for economic variables. This is
evidenced by the prevalence of low adjustment coefficients,
which were estimated along with the long-run relationships.
Own-price and quantity adjustments appear to exercise the
major weight in the adjustment of the short-run to the

170
long-run equilibrium process. While significant elas¬
ticities were observed for the input prices, generally
these were low, but fall within the range of estimates
reported in other studies on supply response in Jamaica.
The low input price elasticities are not surprising since
inputs such as wages and fertilizer constitute important
components in the production process. However, if these
inputs are not to appear as constraints to production,
policy initiatives have to be undertaken to make them
appealing to producers.
Given the biological lag between planting and
harvesting seasons in agriculture, adjustments in the
sector are necessarily slow processes. Additional factors
that contribute to slow adjustments are the institutional
and structural framework within which agricultural
producers must operate. Slow adjustments in the short-run
are signs of constraining government regulations,
inadequate supportive infrastructure, and lack of credit to
farmers. Although not explicitly incorporated in the
modeling framework used in this study, these factors have
been listed by farmers in coffee, cocoa-bean, citrus and
yam production as major constraints to increasing output
(Jamaica, Planning Institute of Jamaica, 1994, 1996, 1998).

171
In an earlier section of this study, some of these
factors were identified as the conduits through which
policies are transmitted to target variables. It appears
that the government has been aware of these problem as can
be gleaned from the following quotation:
For positive growth in the sector to be sustained,
there is urgent need for infrastructural development
and a strengthening of research and extension
services... This need has assumed an increasing sense of
urgency in light of pending changes on the global
economic landscape with the new General Agreement on
Tariffs and Trade (GATT). (Jamaica, Planning Institute
of Jamaica, 1994, p.7.1)
To the extent that this is true, further improvements
in crop supply response would be conditional upon the
intensification and broadening of the reform process in
agriculture. Indeed, the general decline in agricultural
production since the mid-1990s, prompted the government to
outline policy guidelines aimed at correcting some of the
problems impacting the sector. These guidelines emphasized
improving the competitiveness of traditional and non-
traditional exports in international markets; creating an
enabling environment for farmers through additional capital
funding over the long, medium and short-run period;
implementation of projects/programs aimed at improving
production and productivity for local and export markets;
and improving support services (Jamaica, Planning Institute
of Jamaica, 1998).

172
While it may appear that government policies are being
implemented to facilitate the farmer to operate in the new
policy environment, budgetary constraints may not always
permit an effective solution. For example, in light of
increasing demands by farmers for agricultural credit at
concessionary rates, the government offered agricultural
loans at subsidized rate (13 percent). However, loans
allocated to farmers have dropped almost 50 percent over
the past five years because of budgetary constraints.
The empirical evidence provides only partial support
for the hypothesis that the reforms impacted significantly
on agricultural supply responses. In particular, even when
controlling for policy regime changes by constructing
counterfactuals against which the reforms could be
compared, it was found that for the eight crops included in
this study, the effects are crop and time dependent.
Although nominal crop prices have increased
dramatically during the reform period, real crop prices
have remained stable or have declined generally. Movements
in the exchange rate and the general price level seem to be
the immediate causes for these phenomena. It appears also
that price variability increased significantly during the
reform period. Given that farmers in Jamaica, as in most
developing countries, can rarely mitigate against temporal

173
price risk, the evidence seems to suggest that the pro-
competitive effects of the reforms may have acted more as a
depressive factor on supply response, compared to the
stability of the pre-reform period characterized by stable
and guaranteed markets and prices by government commodity
boards.
A relevant question which may be posed here is how
representative is the sample of crops which was chosen for
the study. The eight crops were not randomly chosen but
were selected the same way a researcher might have chosen
to study just one of the crops. However, the crops included
in the study include the most important traditional export
crops of Jamaica. It is unlikely that any of these crops
would have been insulated from the economic reforms over
the past two decades. However, whether the conclusions from
this study can be generalized for the other crops that are
grown in Jamaica, is an empirical question.
Further insights into the impact of the reforms on
crop supply responses in Jamaica would be obtained by more
intensive study of single crops. This will allow more in-
depth study, and guard against using the same modeling
framework to accommodate crops whose supply dynamics may be
different.

APPENDIX A
PRICE DATA USED IN CROP'S ECM ON JAMAICA, 1962-1999 (J$)
Year
banana
sugar
coffee
pimento
yam
orange
Cocoa-
bean
potato
1962
28
5
542
1050
80
18
210
99
1963
29
5
549
1052
86
21
219
99
1964
29
o
560
1065
88
25
228
102
1965
30
6
570
1075
90
27
232
102
1966
30
6
570
1,102
91
33
236
104
1967
34
6
641
2,121
96
38
263
107
1968
34
6
425
1,047
101
31
276
114
1969
29
6
572
1,102
107
29
265
120
1970
30
6
569
1,165
118
31
272
133
1971
51
7
613
1,102
140
31
344
141
1972
49
8
674
1,146
141
25
350
132
1973
73
9
839
1,005
176
37
369
242
1974
159
11
1,272
1, 397
273
57
451
309
1975
188
19
1,784
1,947
342
26
559
350
1976
196
19
1, 966
1,994
399
64
566
466
1977
196
21
2,483
1, 837
424
78
616
573
1978
175
32
3,560
1,900
424
21
3,590
651
1979
347
45
3,000
2, 010
551
42
3,000
816
1980
520
71
2,205
2,100
1,168
104
2,208
1,230
1981
640
102
2,021
2,205
1,224
108
2,991
1,100
1982
700
130
2,572
2,425
1,168
122
2,991
1,320
1983
1,190
239
2, 940
2,425
1,345
122
2,991
1,587
1984
1,998
483
3,123
3,417
1,499
208
4,228
1,852
1985
2,065
762
5,144
5,071
1,631
380
5,104
2,601
1986
2,249
908
8,157
7,716
1,764
454
6,022
4,189
1987
2,866
1,249
7,716
8,047
2,557
381
6,482
5,754
1988
2,668
1,354
7,716
8,818
3,527
536
6,941
5,225
1989
2,337
1,849
7,800
9,149
5,842
3,960
6,556
5,688
1990
3,748
4,943
8,000
9,590
4,410
2,860
6,756
4,670
1991
7,046
5,324
11,684
11,464
7,980
8,800
7,706
10,780
1992
8,818
9,743
27,925
23,148
12,050
30,800
8,000
16,950
1993
11,000
10,957
27,355
27,550
14,700
44,660
9, 000
19,880
1994
17,637
14,825
30,433
33,460
15,000
45,000
10,000
20,000
1995
20,073
18,120
43,333
34,171
11,400
45,522
21,000
15,200
1996
17,813
20,456
43,333
36,376
11,562
45,521
21,987
16,982
1997
17,680
17,170
43,333
45,194
11,256
45,987
21,558
16,989
1998
18,510
16,752
46,065
50,706
11,789
45,887
22,154
17,586
1999
22,522
17,635
48,221
52,602
12,156
45,978
22,546
17,892
Source: (1) 1962-65, 1995-1999: Jamaica, Statistical
Institute of Jamaica, (STATIN), (Data Files) (2) 1966-1995:
Food and Agricultural Organization (FAO). FAOSTAT
Agricultural Data, http://fao.org. August, 1998.
174

APPENDIX B
QUANTITY DATA USED IN CROPS' ECM, 1962-1999 (METRIC TONNES)
Year
banana
sugar
coffee
pimento
yam
orange
Cocoa-bean
potato
1962
221741
523215
1721
3012
61245
73125
1542
10257
1963
210523
521000
1785
3012
62546
74552
1945
11450
1964
226331
512663
1798
3111
62946
75489
1854
12453
1965
231046
505621
1852
3125
63012
78562
1945
10254
1966
240,400
508,247
1,861
3,200
63,486
79,772
2,174
13,608
1967
234,200
455,815
1, 991
4,500
74,580
82,236
1,642
13,297
1968
198,000
451,895
1, 972
4,800
66,306
85,085
1,864
7,454
1969
196,000
389,124
1,589
3,800
62,556
73,304
• 1,857
8,717
1970
195,000
376,057
1, 805
2,671
80,825
65,142
2,163
8,440
1971
187,000
385,029
1,358
3,510
122,864
72,726
1, 872
13,207
1972
198,000
379,214
1, 040
3,290
127,489
83,160
2,378
16,132
1973
169,000
331,181
1,146
3, 410
127,650
23,408
1, 952
8,941
1974
132,000
372,391
1, 486
3,570
131,323
39,309
1, 618
14,606
1975
127,000
360,578
1,186
3,320
134,254
40,849
1,780
13,617
1976
140,000
368,737
1, 929
4,970
119,566
41,234
1, 646
7, 643
1977
160,000
295,237
1,208
3,320
124,941
42,042
1, 692
7,579
1978
160,000
291,541
1,477
3,830
149,767
49,665
1,797
11,794
1979
170,000
283,084
958
2,340
157,169
31,224
1,366
11,331
1980
140,000
231,800
2,216
1,840
132,893
43,159
1,752
6, 848
1981
150,000
225,000
1,379
4,266
136,410
33,000
1,752
12,428
1982
160,000
195,531
1,516
2,759
116,978
32,879
2,408
6, 667
1983
160,000
198,202
1,632
4,729
130,633
31,686
2,302
7,603
1984
155,000
187,791
1,745
6,109
149,060
28,298
2,812
12,310
1985
150,000
225,492
1,274
5,390
163,763
42,119
2,380
7, 075
1986
145,000
206,136
1, 697
4,162
165,633
41,850
2, 406
5,439
1987
140,000
188,985
1, 658
4,412
175,628
75,460
3,186
9,443
1988
135,000
221,706
2,231
3,834
166,864
50,820
1,549
9,893
1989
130,000
204,915
1,260
5, 444
133,281
57,750
1,386
10,818
1990
127,660
208,592
1,560
3,770
161,462
82,236
2,104
14,296
1991
134,000
235,904
2,280
3,750
186,104
61,908
1,750
7,548
1992
130,000
228,024
1, 920
5,730
214,386
70,000
2,478
6, 935
1993
125,000
219,046
1,500
7,120
221,928
72,000
2,522
9,134
1994
120,000
223,041
2,460
9,370
233,912
72,546
2,576
12,188
1995
130,000
248,558
2,580
9,630
240,371
72,892
2,538
17,036
1996
130,000
237,943
2,580
10,370
253,371
73,452
1, 407
13,775
1997
130,000
236,510
2, 887
10,400
212,567
74,561
1,653
12,661
1998
130,000
179,251
3,104
10,542
198,402
74,987
1,687
12,875
1999
130,000
223,000
3,412
11,221
201,354
75,984
1,702
13,256
Source: (1) 1962-65, 1995-1999: Jamaica, Statistical
Institute of Jamaica, (STATIN), (Data Files) (2) 1966-1995:
Food and Agricultural Organization (FAO). FAOSTAT Agri¬
cultural Data, http://fao.org. August, 1998.
175

APPENDIX C
SELECTED ECONOMIC STATISTICS ON JAMAICA
Year
Adef
ferp
rwage
Cpi80
gdef
ER
1962
13.50
144
208.74
11.60
0.40
0.71
1963
13.80
138
209.58
11.95
0.40
0.71
1964
14.10
139
210.28
12.20
0.40
0.71
1965
14.50
138
212.34
12.71
0.40
0.71
1966
15.90
124
216.42
14.60
0.50
0.71
1967
16.70
137
255.69
15.16
0.50
0.83
1968
17.40
132
289.11
16.43
0.60
0.84
1969
19.70
142
300.95
17.78
0.80
0.83
1970
20.50
146
289.13
19.01
0.90
0.84
1971
20.40
144
299.33
19.58
0.90
0.78
1972
20.10
149
325.47
21.03
0.90
0.85
1973
23.90
193
307.26
24.81
1.10
0.91
1974
31.37
229
302.97
31.53
1.40
0.91
1975
38.13
298
362.47
37.04
1.80
0.91
1976
44.75
339
361.25
40.64
1.90
0.91
1977
52.40
4 92
354.16
45.02
2.20
0.91
1978
72.09
778
326.04
60.97
2.70
1.70
1979
86.54
1, 007
264.19
78.79
3.20
1.78
1980
100.00
1,158
234.02
100.00
3.80
1.78
1981
103.67
1,228
221.24
112.75
4.10
1.78
1982
110.58
983
231.40
119.92
4.50
1.78
1983
133.06
966
244.66
113.86
5.20
3.28
1984
198.67
1,149
235.83
170.91
7.00
4.93
1985
249.27
1,582
262.21
215.14
9.20
5.48
1986
316.83
1, 820
277.54
247.41
10.80
5.48
1987
354.07
2, 013
316.75
263.75
12.00
5.50
1988
413.28
1, 913
340.99
285.66
13.60
5.48
1989
442.73
2, 697
314.55
326.27
15.30
6.48
1990
546.16
4,636
268.05
398.45
18.90
8.04
1991
835.26
6,776
182.63
601.99
27.90
21.49
1992
869.95
7,704
175.35
1,066.93
44.40
22.19
1993
893.32
7,725
179.30
1,302.79
59.70
32.48
1994
804.54
13,904
206.60
1,759.36
78.40
33.20
1995
841.98
19,017
218.66
2,109.56
100.00
39.62
1996
879.42
18,136
213.69
2,666.53
121.10
34.87
1997
886.67
21,176
238.27
2,934.30
134.30
41.02
1998
897.56
20,254
270.68
2,942.14
154.00
42.40
1999
909.78
21,924
298.20
3,023.26
159.00
42.84
Sources: (1) Data files, Jamaica, Statistical Institute of
Jamaica (STATIN); (2) Jamaica, Planning Institute of
Jamaica (various issues).
Notes: Adef = Agricultural deflator (1980=100); ferp =
Average price of fertilizer; rwage = index of average
agricultural real wage (1980=100); CPI80 = consumer price
index (1980=100); gdef = GDP deflator (1980=100); ER =
Jamaican exchange rate ($J/US).
176

APPENDIX D
DIAGNOSTIC STATISTICS FOR MODEL SPECIFICATION AND TEST
STATISTICS FOR COINTEGRATION
Table D-l: Tests of Cointegration Rank for Sugar
Eigen
values
Xmax
Trace
Ho: r
P-r
Xmax 90
Trace 90
0.649
37.66
80.99
0
5
21.73
71.66
0.369
16.56
43.33
1
18.03
49.92
0.332
14.54
26.77
2
3
14.09
31.88
0.220
8.96
12.23
3
2
10.29
17.79
0.087
3.27
3.27
4
1
7.50
7.50
Table D-2: Diagnostic Statistics for Residual
Tests for the Sugar ECM.
Tests
Dfa
x2
p-value
For normality
of residuals
10
15.108
0.10
For autocorrelation
L-B(9)
195
124.208
0.21
LM (1)
25
23.084
0.57
LM (4)
25
21.299
0.68
For ARCH(2)
lsugq
2
1.825
lsugpr
2
0.441
lbanpr
2
1.555
lwage
2
0.919
lferp
2
4.600
Notes: The critical value for %2(2) = 5.99
At the 95 percent significance level.
a Df = degrees of freedom.
177

178
Table D-3: Tests of Cointegration Rank for
Coffee ECM.
Eigen
value
iMax
Trace
Ho: r
P-r
A.Max 90
Trace 90
0.615\
40.44
88.74
0
5
21.74
71.66
0.497
24.72
48.30
1
4
18.03
49.92
0.289
12.30
23.58
2
3
14.09
31.88
0.189
7.56
11.28
3
“IT
10.29
17.79
0.098
3.72
3.72
4
1
7.50
7.50
Table D-4: Diagnostic Statistics for Residual
Tests for the Coffee ECM.
Tests
Dfa
x2
p-value
For normality
of residuals
10
11.349
0.33
For autocorrelation
L-B(9)
190
238.190
0.20
LM (1)
25
38.821
0.06
LM (4)
25
26.303
0.39
For ARCH(2)
Lcofq
2
2.763
Lcofpr
2
4.747
Lbanpr
2
0.086
Lferp
2
2.478
Lwage
2
1.345
Notes: The critical value for X2(2) = 5.99 at the
95 percent significance level.
a Df = degrees of freedom.
Table D-5: Tests of Cointegration Rank for
Pimento ECM.
Eigen
value
A.Max
Trace
Ho: r
P-r
A.Max 90
Trace 90
0.608
33.75
81.34
0
5
21.74
71.66
0.438
20.74
47.59
1
4
18.03
49.92
0.284
12.00
26.85
2
3
14.09
31.88
0.263
10.99
14.85
3
2
10.29
17.79
0.102
3.86
3.86
4
1
7.50
7.50

Table D-6: Diagnostic Statistics for Residual
Tests for the Pimento ECM.
Tests
Df a
x2
p-value
For normality
of residuals
10
19.394
0.09
For autocorrelation
L-B(9)
190
212.972
0.06
LM (1)
25
23.861
0.53
LM (4)
25
36.915
0.06
For ARCH(2)
Lpimg
2
3.255
Lpimpr
2
1.131
Lbanpr
2
2.209
Lferp
2
9.196
Lwage
2
4.575
Notes: The critical value for x2(2) = 5.99 at the
95 percent significance level.
a Df = degrees of freedom.
Table D-7: Diagnostic Statistics for Residual
Tests for the Yam ECM.
Tests
Df a
x2
p-value
For normality
of residuals
10
7.472
0.68
For autocorrelation
L-B(9)
185
214.194
0.07
LM (1)
25
28.044
0.31
LM (4)
25
22.418
0.61
For ARCH(2)
Lyamq
2
0.337
Lyampr
2
2.800
Lcaspr
2
3.434
Lferp
2
5.101
Lwage
2
0.065
Notes: The critical value for x2(2) =
95 percent significance level.
a Df = degrees of freedom.
5.99 at the

180
Table D-8: Tests of Cointegration Rank for the
Yam ECM.
Eigen
value
XMax
Trace
Ho:r
P-r
XMax 90
Trace 90
0.682
41.23
110
0
5
21.74
71.66
0.549
28.66
68.67
1
4
18.03
49.92
0.462
22.31
40.01
2
3
14.09
31.88
0.246
10.17
17.70
3
2
10.29
17.79
0.189
7.53
7.53
4
1
7.50
7.50
Table D-9: Diagnostic Statistics for Residual
Tests for the Orange ECM.
Tests
Df a
x2
p-value
For normality
of residuals
10
5.691
0.10
For autocorrelation
L-B(9)
195
217.078
0.12
LM (1)
25
26.658
0.37
LM (4)
25
30.412
0.21
For ARCH(2)
Lorq
2
0.088
Lorpr
2
0.405
Lgrpr
2
6.827
Lferp
2
0.797
Lwage
2
6.931
Notes: The critical value for %2(2) = 5.99 at the
95 percent significance level.
a Df = degrees of freedom.
Table D-10: Tests of Cointegration Rank for
Orange ECM.
Eigen
value
XMax
Trace
Ho: r
P-r
XMax 90
Trace 90
0.679
40.90
75.90
0
5
21.74
71.66
0.341
15.01
34.99
1
4
18.03
49.92
0.315
13.63
19.99
2
3
14.09
31.88
0.142
5.53
6.36
3
2
10.29
17.79
0.021
0.82
0.82
4
1
7.50
7.50

181
Table D-ll: Diagnostic Statistics for Residual
Tests for the Cocoa-bean ECM.
Tests
o
t-h
0)
x2
p-value
For normality
of residuals
10
15.772
0.09
For autocorrelation
L-B(9)
195
171.387
0.10
LM (1)
25
18.188
0.83
LM (4)
25
35.577
0.08
For ARCH(2)
Lcobq
2
0.910
Lcobpr
2
0.438
Lbanpr
2
1.278
Lferp
2
6.781
Lwage
2
3.322
Notes: The critical value for X2(2) = 5.99 at the
95 percent significance level.
a Df = degrees of freedom.
Table D-12: Tests of Cointegration Rank for
Cocoa-bean ECM.
Eigen
value
XMax
Trace
Ho: r
P-r
A,Max 90
Trace 90
0.665
39.41
84.17
0
5
23.72
82.68
0.404
18.65
44.77
1
4
19.88
58.96
0.298
12.74
26.12
2
3
16.13
39.08
0.263
10.98
13.38
~3~
2
12.39
22.95
0.064
2.39
2.39
4
1
10.56
10.56
Table D-13: Tests for Cointegration Rank for
Potato ECM.
Eigen
value
A.Max
Trace
Ho: r
P-r
A.Max 90
Trace 90
0.713
44.92
93.40
0
5
21.74
71.66
0.415
19.30
48.47
1
4
18.03
49.92
0.302
12.92
29.17
2
3
14.09
31.88
0.254
10.55
16.25
3
2
10.29
17.79
0.146
5.70
5.70
4
1
7.50
7.50

Table D-14: Diagnostic Statistics for Residual
Tests for the Potato ECM.
Tests
Dfa
x2
p-value
For normality
of residuals
10
14.295
0.16
For autocorrelation
L-B(9)
190
218.901
0.07
LM (1)
25
24.219
0.51
LM (4)
25
22.342
0.62
For ARCH(2)
Lpotq
2
1.816
Lpotpr
2
3.973
Lcaspr
2
1.283
Lferp
2
1.237
Lwage
2
1.401
Notes: The critical value for %2(2) = 5.99 at the
95 percent significance level.
a Df = degrees of freedom.

REFERENCES
Alogoskoufis, G., and R. Smith. "On Error Correction
Models: Specification, Interpretation, Estimation.
Journal of Economic Survey. 5(1991):97-128 .
Anderson, A., and M. Witter. "Crisis, Adjustment and Social
Change: A Case Study of Jamaica." Consequences of
Structural Adjustment: A Review of the Jamaican
Experience. Elsie Le Franc, ed., pp.1-55. U.W.I.,
Jamaica: Canoe Press, 1994.
Askari, Hossein, and John Thomas Cummings. Agricultural
Supply Response: A Survey of the Econometric Evidence.
New York: Praeger Publishers, 1976.
Baron, David P. "Price Uncertainty, Utility, and Industry
Equilibrium in Pure Competition." International
Economic Review. 11 (1970) :463-480.
Bates, Robert. "The Reality of Structural Adjustment: A
skeptical Appraisal." Structural Adjustment and
Agriculture: Theory and Practice in Africa and Latin
America. Simon Commander, ed., pp. 221-227. London:
Overseas Development Institute, 1989.
Behrman, Jere R. "Macroeconomic Policies and Rural Poverty:
Issues and Research Strategies." Research Paper Series
RP-145, Williams College, Williamstown, Massachusetts,
1990.
Bernanke, Ben. "Alternative Explanation of Money-Income
Correlation." Carnegie-Rochester Conference Series on
Public Policy 25. 1986:49-100.
Binswanger, H. "The Policy Response of Agriculture."
Proceedings of the World Bank Annual Conference on
Development Economics. Supplement to the World Bank
Economic Review and the World Bank Research Observer,
pp. 231-271. Washington D.C., 1989.
183

184
Binswanger, H., and K Deininger. "Explaining Agricultural
and Agrarian Policies in Developing Countries."
Working Paper Series, World Bank, Washington D.C.,
1997.
Bonnick, G. "Liberalization to Centralization and Back."
World Economic Growth. Arnold C. Harberger, ed., pp.
265-291. San Francisco: Institute for Contemporary
Studies Press, 1984.
Boyd, Derick. Economic Management, Income, Distribution,
and Poverty in Jamaica. New York and London: Praeger,
1988.
Brown, L. "Crisis, Adjustment and Social Change: The Middle
Class under Adjustment." Consequences of Structural
Adjustment: A Review of the Jamaican Experience. Elsie
Le Franc, ed., pp. 56-117. U.W.I., Jamaica: Canoe
Press, 1994.
Campbell, H., and Howard Stein, eds. The IMF and Tanzania.
Harare, Zimbabwe: Southern Africa Political Economy
Series Trust, 1991.
Chadha, R., S. Pohit, A. Deardorff, and R. Stern, eds. The
Impact of Trade and Domestic Policy Reforms in India:
A CGE Modeling Approach. Ann Arbor, Michigan: The
University of Michigan Press, 1998.
Chhibber, A. "The Aggregate Supply Response: A Survey."
Structural Adjustment and Agriculture: Theory and
Practice in Africa and Latin America. Simon Commander,
ed., pp.55-68. London: Overseas Development Institute,
1989.
Commander, Simon, ed. Structural Adjustment and
Agriculture: Theory and Practice in Africa and Latin
America. London: Overseas Development Institute, 1989.
Connors, Thomas A. "The Apparent Effects of Recent IMF
Stabilization Programs." International Finance
Discussion Paper 135. Washington: Board of Governors of
the Federal Reserve System, April 1979.
Davis, Carlton, Ballayram, Edward Evans, and C.Y. Thomas.
"Trade Liberalization and Small Economies in the
Americas: The Case of the Caribbean Community
(CARICOM)." International Working Papers IW 99-1, Food

185
and Resource Economics Department, University of
Florida, Gainesville, February 1999.
Doan, T.A. RATS: User's Manual, Version 4.2. Estima,
Illinois. 1996.
Donovan, Donal J. "Macroeconomic Performance and Adjustment
Under Fund-Supported Programs: The Experience of the
Seventies." IMF Staff Papers. 29(1982):171-203.
—. "Real Responses Associated with Exchange Rate Action in
Selected Upper Credit Tranche Stabilization Programs."
IMF Staff Papers. 28 (1981):698-727 .
Doornik, J.A., and H. Hansen. "An Omnibus Test for
Univariate and Multivariate Normality." Working Paper,
Nuffield College, Oxford, 1994.
Enders, W. Applied Econometric Time Series. John Wiley and
Sons: New York. 1995.
—. RATS Handbook for Econometric Time Series. John Wiley
and Sons: New York, 1996.
Engle, Robert F., and Clive Granger. "Cointegration and
Error Correction: Representation, Estimation, and
Testing." Econometrics. 55(1987):251-276.
Food and Agriculture Organization (FAO). FAO Production
Yearbook. Rome: FAO, various years.
—. FAOSTAT Agricultural Data. http://fao.org. August,
1998.
Fuller, W.A. Introduction to Statistical Time Series. New
York: Wiley, 1976.
Gafar, John. "Price Responsiveness of Tropical Agricultural
Exports: A Case Study of Jamaica, 1954-1972." The
Developing Economies. 18 (1980):288-297 .
—. "The Supply Response of Aggregate Agricultural Output
in Jamaica." Agricultural Economics. 16(1997):205-217.
Gardner, B.L. The Economics of Agricultural Policies. New
York: Macmillian Publishing Company, 1987.

186
Garrity, Michele. "Jamaica: Business Associations and the
Public Policy Process." Policy Reforms for Sustainable
Development in the Caribbean. M. Garrity, and L.A.
Picard, eds., pp. 39-57. Amsterdam: ISO Press, 1996.
Goldstein, Morris, and Peter J. Montiel. "Evaluating Fund
Stabilization Programs with Multi-country Data: Some
Methodological Pitfalls." IMF Staff Papers.
33(1986):304-344.
Green, Reginald H. "Articulating Stabilization Programmes
and Structural Adjustment: Sub-Saharan Africa."
Structural Adjustment and Agriculture: Theory and
Practice in Africa and Latin America. Simon Commander,
ed. pp. 35-54. London: Overseas Development Institute,
1989.
Greene, William H. Econometric Analysis, 2nd ed. New York:
Macmillian Publishing Company, 1993.
Guitián, Manuel. Fund Conditionality: Evolution of
Principles and Practices. IMF Pamphlet Series.
No.38(1981).
Gujarati, Damodar N. Basic Econometrics, 3rd ed. New York:
McGraw-Hill, Inc., 1995.
Gylfason, Thorvaldur. "Credit Policy and Economic Activity
in Developing Countries with IMF Stabilization
Programs." Princeton Studies in International Finance,
No.60, Princeton University Press, 1987.
Hall, Bronwyn H., and Clint Cummins. Time Series Processor,
Version 4.4: Reference Manual. TSP International: Palo
Alto, California, 1998.
Hansen, Henrik, and Katarina Juselius. CATS in RATS:
Cointegration Analysis of Time Series. Estima:
Evanston, Illinois, 1995.
Hendry, D.F., A.R. Pagan, and J.D. Sargan. "Dynamic
Specification." Handbook of Econometrics, Vol.II. Z.
Griliches, and M.D. Intriligator, eds., pp.1023-1100.
Amsterdam: Elsevier, 1984.
Hylleberg, S., and G.E. Mizon. "Cointegration and Error
Correction Mechanisms." Economic Journal (Supplement).
99(1989):113-125.

187
Inter-American Development Bank (IDB). Latin America After
a Decade of Reform. Washington, D.C., 1997.
International Monetary Fund (IMF). International Financial
Statistics. Washington, D.C., 1998.
Ishii, Yasunori. "On the Theory of the Competitive Firm
Under Price Uncertainty: Note." American Economic
Review. 67(1977):768-769
Jaeger, W. "The Impact of Policy in African Agriculture: An
Empirical Investigation." Working Papers, WPS-640,
Policy, Research and External Affairs, World Bank,
Washington, D.C., 1991.
Jamaica, Agency for Public Information (API), Michael
Manley. "The Need for a new Economic Path." Broadcast
to the Nation, Sunday 3, February, 1980.
Jamaica, Bank of Jamaica. The Balance of Payments.
Kingston, Jamaica: Bank of Jamaica, 1985.
—. Statistical Digest. Kingston, Jamaica: Bank of Jamaica,
various issues.
Jamaica, Planning Institute of Jamaica. Economic and Social
Survey: Jamaica. Kingston, Jamaica: Planning Institute
of Jamaica, various years.
Jamaica, Statistical Institute of Jamaica (STATIN). Census
of Agriculture. Kingston, Jamaica: Statistical
Institute of Jamaica, various years.
—. National Income and Product. Kingston, Jamaica:
Statistical Institute of Jamaica, 1970.
—. Production Statistics. Kingston, Jamaica: Statistical
Institute of Jamaica, various years.
—. Statistical Yearbook of Jamaica. Kingston, Jamaica:
Statistical Institute of Jamaica, various years.
Jefferson, 0. The Post-war Economic Development of Jamaica.
Mona, Jamaica: University of the West Indies,
Institute of Social and Economic Research, 1972.

188
Johansen, S0ren. Likelihood-based Inference in Cointegrated
Vector Autoregressive Models. Oxford University Press:
New York, 1995.
—. "Statistical Analysis of Cointegration Vectors."
Journal of Economic Dynamics and Control. 12(June-
September 1988):231-254 .
Johansen, Seren, and Katarina Juselius. "Maximum Likelihood
Estimation and Inference on Cointegration — with
Applications to the Demand for Money." Oxford Bulletin
of Economics and Statistics. 52(1990):169-210.
—. "Testing Structural Hypotheses in a Multivariate
Cointegration Analysis of the PPP and the UIP for UK."
Journal of Econometrics. 53 (1992):211-244.
Kaufman, M. Jamaica under Manley: Dilemmas of Socialism and
Democracy. London, U.K.: Zed Books Ltd., 1985.
Khan, Mohsin S. "The Macroeconomic effects of Fund-
Supported Adjustment Programs." IMF Staff Papers
37:2 (1990) :195-231.
Khan, Mohsin, and Malcolm Knight. "Fund-Supported
Adjustment Programs and Economic Growth." Occassional
Paper 41, International Monetary Fund, 1985.
—. "Stabilization Programs in Developing Countries: A
Formal Framework." IMF Staff Papers. 29(January
1981):1-53.
Krueger, A.O. The Political Economy of Agricultural Pricing
Policy, Volume 5: A Synthesis of the Political Economy
in Developing Countries. Baltimore: Johns Hopkins
University Press, 1992.
La Guerre, John, ed. Structural Adjustment, Public Policy
and Administration in the Caribbean. St. Augustine
Campus, Trinidad and Tobago: University of the West
Indies, School of Continuing Studies, 1994.
Le Franc, E., ed. Consequences of Structural Adjustment: A
Review of the Jamaican Experience. U.W.I., Jamaica:
Canoe Press, 1994.

189
Lora, Edwardo. "A Decade of Structural Reforms in Latin
America: What Has Been Reformed and How to Measure it."
OCE Working Paper #348, IDB, Washington, D.C., 1997.
Loxley, John. The IMF and the Poorest Countries. Ottawa,
Canada: North South Institute, 1984.
Lütkepohl, H., and H.E. Reimers. "Impulse Response Analysis
of Cointegrated Systems". Journal of Economic Dynamics
and Control. 16:1(1992):53-78.
Mamingi, N. "The Impact of Prices and Macroeconomic
Policies on Agricultural Supply: A Synthesis of
Available Results." Agricultural Economics.
16(1997):17-34.
Nallari, R. "Jamaica: The Search for Stabilization and
Adjustment." World Bank Staff Paper, Washington, D.C.,
World Bank, 1992.
Nelson, C.R., and C.I. Plosser. "Trends and Random Walks in
Macroeconomics Time Series." Journal of Monetary
Economics. 10(1982):140-162.
Newman, M., and Elsie Le Franc. "The Small-Farm Sub-Sector:
Is there Life after Structural Adjustment?"
Consequences of Structural Adjustment: A Review of the
Jamaican Experience. Elsie Le Franc, ed. pp.118-181.
U.W.I., Jamaica: Canoe Press, 1994.
Nickell, S. "Error Correction Partial Adjustment and all
That: An Expositionary Note." Oxford Bulletin of
Economics and Statistics. 47(1985):119-129.
Osterwald-Lenum, Michael. "Practitioner's Corner: A Note
with Quantiles for the Asymptotic Distribution of the
Maximum Likelihood Cointegration Rank Test Statistic."
Oxford Bulletin of Economics and Statistics.
54(1992):461-471.
Pantula, Sastry G., Graciela Gonzalez-Farias, and Wayne A.
Fuller. "A Comparison of Unit-Root Test Criteria."
Journal of Business and Economic Statistics. 12(October
1994) : 449-459.
Pastor, Manuel. "The Effects of IMF Programs in the Third
World: Debate and Evidence for Latin America." World
Development. 15 (1987):249-262.

190
Pindyck, Robert S., and Daniel L. Rubinfeld. Econometric
Models and Economic Forecasts 3rd ed. New York: McGraw-
Hill, Inc., 1991.
Phillips, P.C.B., and Pierre Perron. "Testing for a Unit
Root in Time Series Regression." Biometrika.
75(1988):335-346.
Pollard, S.K., and D.H. Graham. "Price Policy and
Agricultural Export Performance in Jamaica." World
Development. 13(1985):1067-1075.
Reichmann, Thomas M., and Richard T. Stillson. "Experience
with Programs of Balance of Payments Adjustment: Stand¬
by Arrangements in the Higher Credit Tranches, 1963-
72." IMF Staff Papers. 25 (1978):293-309.
Robinson, S., and C. Gehlhar. "Impacts of Macroeconomics
and Trade Policies on a Market-Oriented Agriculture."
Egypt's Agriculture in a Reform Era. Lehman B.
Fletcher, ed., pp. 271-293. Iowa: Iowa State University
Press, 1996.
Sandmo, Agnar. "On the Theory of the Competitive Firm Under
Price Uncertainty." American Economic Review.
61(1971):65-73.
Schiff, Maurice, and Alberto Valdés. "A Synthesis of the
Economics in Developing Countries." The Political
Economy of Agricultural Pricing Policy (Vol.4): A World
Bank Comparative Study. Baltimore: Johns Hopkins
University Press, 1992a.
—. The Plundering of Agriculture in Developing Countries.
Washington, D.C.: The World Bank, 1992b.
Serieux, John. "Democratization and Structural Adjustment
in Sub-Saharan Africa: An Empirical Investigation."
Canadian Journal of Development Studies. 20(1999):521-
544.
Sharpley, Jennifer. "Jamaica, 1972-80." The IMF and
Stabilization: Developing Country Experiences. Tony
Killick, ed., pp.115-163. New York: St. Martin's Press
Inc., 1984.
Sims, Christopher. "Macroeconomics and Reality."
Econometrica. 48(1980):1-49.
\

191
Singh, R. "The Impact of Structural Adjustment Policies on
the Performance of Agriculture: The Case of Jamaica."
Structural Adjustment and the Agricultural Sector in
Latin America and the Caribbean. John Weeks, ed., pp.
229-257. New York: St. Martin's Press Inc., 1995.
Stephens, E.H., and J.D. Stephens. Democratic Socialism in
Jamaica: The Political Movement and Social
Transformation in Dependent Capitalism. Princeton, New
Jersey: Princeton University Press, 1986.
Thomas, Desmond. "Anatomy of a Stabilization Process: The
Case of Jamaica, 1984 to the Present. Canadian Journal
of Development Studies. 20(1999):159-179.
Tsakok, Isabelle. Agricultural Price Policy: A Practioner's
Guide to Partial-Equilibrium Analysis. Ithaca, New
York: Cornell University Press, 1990.
Tshibaka, T.B., ed. Structural Adjustment and Agriculture
in ¡Vest Africa. Dakar, Senegal: Council for the
Development of Social Science Research in Africa
(CODESRIA), 1998.
Valdés, Alberto, and Kay Muir-Leresche, eds. Agricultural
Policy Reforms and Regional Market Integration in
Malawi, Zambia, and Zimbabwe. Washington, D.C.:
International Food Policy Research Institute (IFPRI),
1993.
Weeks, John. Structural Adjustment and the Agricultural
Sector in Latin America and the Caribbean. New York:
St. Martin's Press, 1995.
World Bank. Adjustment in Africa: Reforms, results, and the
road ahead. New York: Oxford University Press, 1994.
—. Trends in Developing Economies. Washington, D.C.: World
Bank, 1996.

BIOGRAPHICAL SKETCH
Ballayram of Essequibo Coast, Guyana, worked in the
primary educational and civil service systems in Guyana
before pursuing higher education. He completed his B.S.Soc.
degree with distinction in economics at the University of
Guyana in 1980. After three years as an Assistant Lecturer
in the Department of Economics at the University of Guyana,
he proceeded to pursue a Master of Arts degree in
development economics, under a Fulbright Scholarship, at
the University of Notre Dame, Indiana. On completion of his
master's degree he lectured first in the Economics
Department at the University of Guyana, and then
subsequently, in the Economics Department at the University
of the West Indies, Mona Campus, Jamaica. Ballayram was
awarded a Graduate Research Assistantship, in 1994, to
pursue a Ph.D. in the Food and Resource Economics
Department at the University of Florida. His areas of
specialty are environmental economics, and international
trade and economic development.
192

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^^Doctor of Philosophy.
Dr. Carlton^G. Davis, Chair
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 oj
Dr. Clyde F. Kiker
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 Doctor of Ph^lqsophy.
Dr. Richard L. Kilmer
Professor of Food and Resource
Economics
I certify that I have read this study and that in my
opinion it conforms to acceptablestandards of scholarly
presentation and is fully ade^uata/Vln scope and' quality,
as a dissertation for the degree jz<£ pyctffar j6f Philosophy.
Robert 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.
Dr. David A. Denslow, J,
Distinguished Service
Professor of Food and
Resource Economics
This dissertation was submitted to the Graduate
Faculty of the College of Agricultural and Life Sciences
and to the Graduate School and was accepted as partial
fulfillment of the requirement for the degree of Doctor of
Philosophy.
May 2001
Dean, Graduate School




PAGE 1

ECONOMIC REFORMS AND AGRICULTURAL SUPPLY RESPONSE IN JAMAICA By BALLAYRAM 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 2001

PAGE 2

ACKNOWLEDGEMENTS First, I would like to express my deepest gratitude to my wife for all her support throughout my studies here at the University of Florida. Her support in this process was overwhelming and unselfish. My daughters were also very understanding, supportive, and encouraging, and I thank them sincerely for the sacrifices they had to make so that I could complete my graduate studies. My parents and in laws (all now deceased) gave me encouragement and support, both financial and otherwise, in my quest to achieve a higher education. Without their love, confidence and moral support, none of this would have been possible. I owe them an eternal debt of gratitude, because their support came at a time when the cost of higher education was far beyond the reach which their economic resources could accommodate. I would also like to express my deepest gratitude to Dr. Carlton G. Davis, chairman of my supervisory committee. He opened the door for me at the University of Florida, and provided me with countless hours of indi victual attention during my course of studies. On numerous occasions he went beyond what would normally be the call of duty of an ii

PAGE 3

academic supervisor, to provide me with the support that enabled me to continue my studies. I would also like to acknowledge the contributions from the other members of my supervisory committee: Dr. Robert Emerson, Dr. Clyde Kiker, Dr. Richard Kilmer and Dr. David Denslow, who provided me with valuable suggestions during the preparation of this dissertation. I also benefited immensely, in my course of studies at U. F., from discussions with Dr. Max Langham, Dr. Charles Moss, Dr. Ronald Ward and Dr. James Seale, Jr. Financial support from the Food and Resource Economics Department is greatly acknowledged. Dr. Carlton Davis' efforts in securing this funding for me are greatly appreciated. Finally, I wish to acknowledge the excellent computer guidance and support services which Alex Heyman, Roger and Laura Clemons, Ed Howard and Juan Carlos, of the Food and Resource Economics Support Center, provided throughout my studies here at U.F. iii

PAGE 4

TABLE OF CONTENTS page ii ACKNOWLEDGEMENTS LIST OF TABLES LIST OF FIGURES ABSTRACT CHAPTERS . . . . . . . . . . vi 1. INTRODUCTION 1.1 Background to Jamaica's Economic Reforms . . . . . . 1.2 Problem Statement and Study Justification . . . . . 1. 3 Objectives of the Study . . 1.4 Hypotheses . . . . 1.5 Conceptual Issues . . . . . 1. 6 Procedural and Methodological Considerations . . . . 1. 7 Summary . . . . . . . 2. ECONOMIC STRUCTURE, GROWTH, AND POLICY REFORMS IN THE JAMAICAN ECONOMY, 1962-1999 2.1 Structure of the Economy and Economic Growth: An Overview 2.2 Output Trends and Policies in Jamaican Agriculture . . 2.3 The Economic Reforms . . . 2.4 Summary . . . . 3. METHODOLOGICAL AND EMPIRICAL ISSUES IN ECONOMIC REFORMS AND SUPPLY RESPONSE 3.1 Review of the Economic Reform Literature iv . . . . ix xv 1 2 8 10 . 11 11 . 20 22 24 . 26 . 30 38 . 46 47 47

PAGE 5

3 .1.1 3 .1. 2 3 .1. 3 Before-After Approach .. With-Without Approach .. Comparison of Simulations Approach ...... Preliminary Issues in Modeling Supply Response in Jamaica ..... 51 52 54 3.2 3.3 3.4 Error Correction Model . . . . 55 60 3.5 3.6 An Error Correction Model for Crop Supply Response in Jamaica ..... The Data . . . . . . . Summary ..... 4. EMPIRICAL ESTIMATION OF CROP SUPPLY RESPONSE. 4.1 Motivations for Using Cointegration Analysis . . . . . . 4.2 Definition of Variables . . . . 4.3 Testing for Stationarity . . . 4.4 Estimating Long-run Supply Response 4.5 Analysis of Short-run Dynamics 4 6 Summary . . . . . . . . 5. SUPPLY RESPONSE AND COUNTERFACTUAL ANALYSIS . . 63 71 74 75 75 79 87 90 106 136 137 5.1 Before-After Analysis ... ........ 137 5.2 Counterfactual Analysis ........... 145 5.3 Summary . . . . . . . . 164 6. CONCLUSIONS AND POLICY IMPLICATIONS APPENDICES A. PRICE OF SELECTED CROPS PRODUCED IN JAMAICA, . 165 1962-1999 (J$) ............. ... 174 B. QUANTITY OF SELECTED CROPS PRODUCED IN JAMAICA, 1962-1999 (METRIC TONES) 175 C. SELECTED ECONOMIC STATISTICS ON JAMAICA 176 D. DIAGNOSTIC STATISTICS FOR MODEL SPECIFICATION AND TEST STATISTICS FOR COINTEGRATION REFERENCES BIOGRAPHICAL SKETCH. V .. 177 183 192

PAGE 6

LIST OF TABLES Table page 2.1 Sectoral Contribution to Real Gross Domestic Product (Period Averages, Percentages) .................. 27 2.2 GDP and Sectoral Growth (1995=100) (Percentage) . . . . . ..... 28 2.3 Growth Rates of Total Agricultural and Broad Agricultural Aggregate Indexes ...... 30 2.4 Composition of Agricultural Output (J$M, 1995=100), Selected Years . ..... 32 2.5 Agriculture Sub-sectors as a Percentage of Total Agriculture (Period Averages) ..... 32 2.6 Growth Rates of Agricultural Components (Percentages) . . . 34 2.7 Average Rates of Growth of Farmgate and F.O.B. Prices, Output and the Nominal Protection Coefficient (NPC) for the Period 1970-78, Jamaica . . ... 36 2.8 Net Barter Terms of Trade and Agricultural Income Terms of Trade (1980=100) (Period Averages) ................ 38 2.9 Outcomes of Economic Reform Policies in Jamaica, 1986 vs. 1995 . . ...... 45 3.1 Empirical Studies on Agriculture Supply Responses in Jamaica .. 4.1 Alternative (Substitute) Crops in Each ECM 4.2 Unit Root Tests--Prices and Quantities 56 8 4 in Levels . . . . . . . . . 88 vi

PAGE 7

4.3 Unit Root Tests--Prices and Quantities in First Difference. . . . ..... 89 4.4 Diagnostic Statistics for Residual Tests for the Banana ECM 4.5 Tests of Cointegration Rank for Banana ECM 4.6 Estimated Long-run and Adjustment . 91 95 Coefficients, P's, a's--Banana ECM ....... 96 4.7 Estimated Long-run and Adjustment Coefficients, P's, a's for all Crops 4.8 Endogenous and Exogenous Variables in Crops' . 100 Impulse Response Functions . . ..... 109 4.9 Responses of Banana Quantity to Shocks in the Banana VAR .............. 111 4.10 Variance Decomposition Percentage of One period and Three-period Forecast Error Variance--Banana VAR .............. 114 4.11 Responses of Quantity to Shocks in Exogenous Variables . . . . . . ..... 117 4.12 Summary of Forecast Error Variance in Crop Quantities Explained by Variables in the VAR (Percentage) .......... 119 4.13 Responses of Sugar Quantity to Shocks in the Sugar VAR ...... 4.14 Variance Decomposition Percentage of One period and Three-period Forecast Error . . 121 Variance--Sugar VAR . . . . . . 121 4.15 Responses of Coffee Quantity to Shocks in the Coffee VAR ............... 122 4.16 Variance Decomposition Percentage of One period and Three-period Forecast Error Variance--Cof fee VAR . . . . . . . 122 4.17 Responses of Pimento Quantity to Shocks in the Pimento VAR . . . . . . 123 vii

PAGE 8

4.18 Variance Decomposition Percentage of One period and Three-period Forecast Error Variance--Pirnento VAR ........... 123 4.19 Responses of Yarn Quantity to Shocks in the Yarn VAR ......... 4.20 Variance Decomposition Percentage of One period and Three-period Forecast Error . . 12 4 Variance--Yarn VAR . . . . . ... 124 4.21 Responses of Orange Quantity to Shocks in the Orange VAR . . . . ... 125 4.22 Variance Decomposition Percentage of One period and Three-period Forecast Error Variance--Orange VAR .............. 125 4.23 Responses of Cocoa-bean Quantity to Shocks in the Cocoa-bean VAR . . ...... 126 4.24 Variance Decomposition Percentage of One period and Three-period Forecast Error Variance--Cocoa-bean VAR. . . . ... 126 4.25 Responses of Potato Quantity to Shocks in the Potato VAR . . . . . . 127 4.26 Variance Decomposition Percentage of One period and Three-period Forecast Error Variance--Potato VAR ............. 127 5.1 Descriptive Price and Quantity Statistics --1962-1979 and 1980-1999 ........... 140 5.2 Descriptive Price and Quantity Statistics --1975-1979 and 1980-1984 . . . ... 141 5.3 Diagnostic Statistics for Forecasts of Fitted Values . . . . . ..... 155 5.4 Estimated Long-run and Adjustment Coefficients, P's, a's for all Crops, 1962-1979 . . . ..... 5.5 Estimated Long-run and Adjustment Coefficients, P's, a's for all Crops, 157 1980-1999 . . . . ........ 159 viii

PAGE 9

LIST OF FIGURES Figure page 1.1 Conceptualization of the Jamaican Agricultural Sector .............. 14 2.1 GDP and Sectoral Contribution to GDP in Constant ( 1995) Dollars . . . . . 27 2.2 GDP Growth (Percentage Change Over Previous Year, 1995=100) . . 2.3 2.4 2.5 2.6 2.7 4.1 4.2 4.3 4.4 4.5 4.6 4.7 Growth of Total Agriculture, Industry and Services (Percentage Change Over Previous Year, 1995=100) ..... Total Agricultural Output Index (1995=100) Output Indexes for Broad Agricultural Aggregates (1995=100) ...... Sub-sectors as a Proportion of Agriculture Selected Economic Policy Reforms in Jamaica, 1977-98 ..... Residuals for Banana Price in Banana ECM . Residuals for Banana Quantity in Banana ECM Residuals for Sugar Price in Banana ECM Residuals for Wage in Banana ECM . . Crossand Autocorrelograms in Banana ECM Plot of Eigenvalues for the Banana ECM . Impulse Responses to an Innovation in Sugar Price--Banana VAR . . . . . . ix . . . . . . . . . 29 29 31 31 33 43 92 92 93 93 94 94 112

PAGE 10

4.8 Impulse Responses to an Innovation in Fertilizer Price--Banana VAR. . ... 112 4.9 Impulse Responses to an Innovation in Wage --Banana VAR . . . . . 112 4.10 Impulse Responses to an Innovation in Banana Price--Banana VAR .... 4.11 Impulse Responses to an Innovation in Banana ... 112 Quantity--Banana VAR .............. 113 4.12 Impulse Responses to an Innovation in Sugar Quantity--Sugar VAR. .......... 113 4.13 Impulse Responses to an Innovation in Coffee Quan ti ty--Coffee VAR . . . . . . 113 4.14 Impulse Responses to an Innovation in Pimento Quantity--Pimento VAR .............. 113 4.15 Impulse Responses to an Innovation in Banana Price--Sugar VAR. . . . . ..... 128 4.16 Impulse Responses to an Innovation in Fertilizer Price--Sugar VAR .... 128 4.17 Impulse Responses to an Innovation in Wage --Sugar VAR. . . . . . ... 128 4.18 Impulse Responses to an Innovation in Sugar Price--Sugar VAR ..... 4.19 Impulse Responses to an Innovation in Banana . 12 8 Price--Cof fee VAR . . . . . . 12 9 4.20 Impulse Responses to an Innovation in Fertilizer Price--Coffee VAR. . .. 129 4.21 Impulse Responses to an Innovation in Wage --Coffee VAR . . . . ... 129 4.22 Impulse Responses to an Innovation in Coffee Price--Coffee VAR .... 4.23 Impulse Responses to an Innovation in Banana . 12 9 Price--Pimento VAR ............. 130 X

PAGE 11

4.24 Impulse Responses to an Innovation in Fertilizer Price--Pirnento VAR . . ... 130 4.25 Impulse Responses to an Innovation in Wage --Pimento VAR. . . . . ... 130 4.26 Impulse Responses to an Innovation in Pimento Price--Pirnento VAR ... 4.27 Impulse Responses to an Innovation in Cassava . 130 Price--Yarn VAR. . . . . ..... 131 4.28 Impulse Responses to an Innovation in Fertilizer Price--Yarn VAR . . ... 131 4.29 Impulse Responses to an Innovation in Wage --Yarn VAR . . . . . . . 131 4.30 Impulse Responses to an Innovation in Yam Price--Yarn VAR. . . . ... 131 4.31 Impulse Responses to an Innovation in Yam Quantity--Yarn VAR ............... 132 4.32 Impulse Responses to an Innovation in Orange Quantity--Orange VAR ........... 132 4.33 Impulse Responses to an Innovation in Cocoabean Quantity--Cocoa-bean VAR ......... 132 4.34 Impulse Responses to an Innovation in Potato Quantity--Potato VAR .............. 132 4.35 Impulse Responses to an Innovation in Grapefruit Price--Orange VAR. . ..... 133 4.36 Impulse Responses to an Innovation in Fertilizer Price--Orange VAR. . ... 133 4.37 Impulse Responses to an Innovation in Wage --Orange VAR . . . . ... 133 4.38 Impulse Responses to an Innovation in Orange Price--Orange VAR .... 4.39 Impulse Responses to an Innovation in Banana . 133 Price--Cocoa-bean VAR ............. 134 xi

PAGE 12

4.40 Impulse Responses to an Innovation in Fertilizer Price--Cocoa-bean VAR .... 134 4.41 Impulse Responses to an Innovation in Wage --Cocoa-bean VAR. . . . . ... 134 4.42 Impulse Responses to an Innovation in Cocoa-bean Price--Cocoa-bean VAR . 134 4.43 Impulse Responses to an Innovation in Cassava Price--Potato VAR . . . . . . 135 4.44 Impulse Responses to an Innovation in Fertilizer Price--Potato VAR .. 4.45 Impulse Responses to an Innovation in Wage --Potato VAR ........ 4.46 Impulse Responses to an Innovation in Potato Price--Potato VAR ..... 5.1 Nominal Price Changes--Banana (BANP) . . 135 135 135 and Sugar (SUGP) ................ 143 5.2 Nominal Price Changes--Coffee (COFP) and Pimento (PIMP) ............... 143 5.3 Nominal Price Changes--Yam (YAMP) and Orange (ORP) ....... . 143 5.4 Nominal P~ice Changes--Cocoa-bean (COBP) and Potato ( POTP) . . . . . 143 5.5 Real Price Changes--Banana and Sugar (1980=100) . . . . ......... 144 5.6 Real Price Changes--Coffee and Pimento (1980=100) . . . . ..... 144 5.7 Real Price Changes--Yam and Orange (1980=100) ........ . . 144 5.8 Real Price Changes--Cocoa-bean and Potato (1980=100) ................... 144 5.9 Banana Quantity--Actual (LBANQ) and Forecasted (FOLBANQ) ............ 148 xii

PAGE 13

5.10 Banana Price--Actual (LBANPR) and Forecasted (FOLBANPR) 5.11 Sugar Price--Actual (LSUGPR) and Forecasted (FOLSUGPR) 5.12 Wage--Actual (LWAGE) and Forecasted 148 148 (FOLWAGE) . . . . ...... 148 5.13 Fertilizer Price--Actual (LFERP) and Forecasted (FOLFERP) . . . ...... 149 5.14 Sugar Quantity--Actual (LSUGQ) and Forecasted (FOLSUGQ) .............. 149 5.15 Coffee Quantity--Actual (LCOFQ) and Forecasted (FOLCOFQ) .............. 149 5.16 Coffee Price--Actual (LCOFPR) and Forecasted (FOLCOFPR) . . . . . .. 149 5.17 Pimento Quantity--Actual (LPIMQ) and Forecasted (FOLPIMQ) . . . ...... 150 5.18 Pimento Price--Actual (LPIMPR) and Forecasted (FOLPIMPR) ........... 150 5.19 Yam Quantity--Actual (LYAMQ) and Forecasted (FOLYAMQ) .............. 150 5.20 Yam Price--Actual (LYAMPR) and Forecasted ( FOLYAMPR) . . . . . 150 5.21 Cassava Price--Actual (LCASPR) and Forecasted (FOLCASPR) . . ....... 151 5.22 Orange Quantity--Actual (LORQ) and Forecasted (FOLORQ) ...... 151 5.23 Orange Price--Actual (LORPR) and Forecasted (FOLORPR) . ...... 151 5.24 Grapefruit Price--Actual (LGRPR) and Forecasted (FOLGRPR) . . . .... 151 5.25 Cocoa-bean Quantity--Actual (LCOBQ) and Forecasted (FOLCOBQ) .............. 152 xiii

PAGE 14

5.26 Cocoa-bean Price--Actual (LCOBPR) and Forecasted (FOLCOBPR) ............. 152 5.27 Potato Quantity--Actual (LPOTQ) and Forecasted (FOLPOTQ) .............. 152 5.28 Potato Price--Actual (LPOTPR) and Forecasted (FOLPOTPR) ..... 5.29 Banana--Fitted Output for Reform Period . . 152 (BANFITA) and Counterfactual (BANFITC) ..... 161 5.30 Sugar--Fitted Output for Reform Period (SUGFITA) and Counterfactual (SUGFITC) ..... 161 5.31 Coffee--Fitted Output for Reform Period (COFFITA) and Counterfactual (COFFITC) ..... 161 5.32 Pimento--Fitted Output for Reform Period (PIMFITA) and Counterfactual (PIMFITC) ..... 161 5.33 Yam--Fitted Output for Reform Period (YAMFITA) and Counterfactual (YAMFITC) ..... 162 5.34 Orange--Fitted Output for Reform Period (ORFITA) and Counterfactual (ORFITC) . .. 162 5.35 Cocoa-bean--Fitted Output for Reform Period (COBFITA) and Counterfactual (COBFITC) ... 162 5.36 Potato--Fitted Output for Reform Period (POTFITA) and Counterfactual (POTFITC) ..... 162 xiv

PAGE 15

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 ECONOMIC REFORMS AND AGRICULTURAL SUPPLY RESPONSE IN JAMAICA By Ballayram May 2001 Chairman: Dr. Carlton G. Davis Major Department: Food and Resource Economics A number of economic reform programs have been undertaken in Jamaica, over the past two decades. Designed largely by the International Monetary Fund ( IMF) and the World Bank, these reforms focused on correcting internal policy weaknesses and creating an environment conducive to sustained growth. The reforms emphasized liberal trade and exchange rate regimes, a less intrusive and smaller public sector, and reliance on market forces to determine agricultural prices and quantities. Against this background, this study investigates the impact of these recent economic reforms on agricultural crop supply responses in Jamaica. The estimation technique used an error correction modeling framework based on xv

PAGE 16

cointegration theory, within an estimation framework developed by Johansen. The results of the crop supply response estimation confirm that there is a long-run relationship between agricultural crop output and price incentives. Most of the estimated crop price elasticities are low, statistically significant, and fall within the range estimated by other studies on Jamaica. The adjustment process of the short-run to the long-run was found to be slow for some crops and higher for others. Using a counterfactual, which assumed no change in policy regime, fitted series of supply response functions from the pre-reform period were forecasted within a univariate ARMA(p,q) framework and compared to fitted series of supply responses from the reform period. The results are mixed. It was found that the impacts of the economic reforms in Jamaica are crop and time specific. Mean output was higher in the reform period for four of the eight crops analyzed. Higher real price shifts were observed in the reform period for some crops but these price shifts were also accompanied by higher price variability. This suggests that the pro-competitive effects that were expected to accompany the reforms may have outweighed the stability impulses of administered prices in the pre-reform era. xvi

PAGE 17

CHAPTER 1 INTRODUCTION The past two decades have seen a shift in economic policies throughout Latin American and Caribbean countries. Generally, these policy changes have been in response to debt crises, persistent payments-imbalances, and years of negative and slow growth. By the 1990s a new orthodoxy on development thinking could be discerned. This is manifested in reductions in public sector activities and greater reliance on private sector initiatives and market forces, a departure from the policy regimes of the 1960s and 1970s. In Jamaica, as elsewhere, the new orthodoxy has influenced economic policies in agriculture. In particular, policies have shifted away from far-reaching government interventions and control, to increased reliance on market signals to allocate resources and form prices. In addition, economy-wide macro-economic policy changes have been made which can potentially influence agricultural incentives, income, and output. Designed largely by the International Monetary Fund (IMF) and the World Bank in consultation with 1

PAGE 18

2 the Government of Jamaica, these reforms 1 are expected to favor Jamaican agriculture under the presumption by the Bretton Woods institutions that previous policies severely discriminated against the sector. (This point is elaborated in Chapter 2, section 2. 4) This study pertains to the impacts that these economic policy reforms have on agricultural supply responses in Jamaica, as the agricultural sector emerges from a heavily regulated sector to one that is more open 2 and liberal. This chapter discusses the problems that this study addresses. 1.1 Background to Jamaica's Economic Reforms For purposes of this study, the reform period is defined as the years 1980-1999, while the pre-reform period is 1962-1979. The year 1962 is sufficiently far back to facilitate a meaningful analysis of economic policies and conditions that led up to the dramatic changes in policies in the reform period. The reform period marks a re orientation of economic policy. This change arose from the stabilization programs of the IMF in Jamaica in the late 1970s. It gathered momentum in the 1980s and 1990s in the In this study, "policy reforms," "economic policy reforms," "economic reforms," and "structural reforms," are used interchangeably. 2 Openness is defined here in terms of trade/GDP ratio. A high trade/GDP ratio can coexist with high import tariffs, export subsidies, and non-tariff barriers, signaling a non-liberal trade regime.

PAGE 19

3 form of structural adjustment programs cum reforms for economic liberalization, under the direction of the IMF, the World Bank and other leading external funding agencies, notably the Inter-American Development Bank (IDB) and the United States Agency for International Development (USAID). In effect, therefore, for the purposes of this study, economic reforms refer to the policies stipulated in the stabilization and structural adjustment programs which Jamaica has been pursuing since the late 1970s. When the People's National Party (PNP) took office in 1972 under Prime Minister Michael Manley, it was under pressure to keep its election campaign promises to increase real wages and government spending, and to reduce social inequalities. The responses of the new administration to this situation, domestic events, coupled with subsequent external and were to plunge the economy into a sustained recession and major disequilibrium During the previous administration under the Jamaica Labor Party ( JLP) the performance of the economy was impressive. GDP growth rates were positive and relatively high (5.1 percent per annum over the 1960-1970 period), inflation was less than five percent per annum and the balance of payments showed modest surpluses (Jefferson, 1972; IMF, 1998). Over the 1970-1980 period, however, real GDP declined by -0. 9

PAGE 20

4 percent. In fact, with the exception of a negligible O. 7 percent GDP growth in 1978, negative growth rates were recorded for all other six years between 1974 and 1980. Similarly, in the productive sectors, agricultural value added growth declined from 6.7 percent in 1970-1972 to -0.4 percent in 1973-1980. For these same periods, declines in growth rates were recorded for services from 9.2 percent to -0.08 percent, manufacturing from 6.7 percent to -4.2 percent, construction from 6.0 to -10.6 percent, and mining from 13.2 percent to 1.3 percent, respectively (Singh, 1995; IMF, 1998). The external account reflected the production crisis described above. Between 1972 and 1980, the accumulated deficit on the balance of payments was US$ 67 9. 2 million, compared to an accumulated surplus of US$95 million between 1960 and 1971 (Jamaica, Bank of Jamaica, 1985). Further, the deficit on the current account increased almost three fold from US$570.2 million in 1962-1970 to US$1,620.0 million in 1972-1980, and the capital account became negative in 1977, signaling capital flight. At the same time, the dwindling net capital inflows could not match the current account deficit. This depleted international reserves, which plummeted sharply from US$103.l million in 1972 to minus US$461 million in 1980 (Sharpley, 1984). The

PAGE 21

5 foreign exchange shortage harmed the economy given its relative openness and dependence on foreign intermediate imports. Finally, the inflation rate, which was 4.2 percent per annum over the 1960-1970 period, recorded double-digit rates throughout the following decade. Domestic prices, as measured by the consumer price index, rose sharply, recording an annual increase of 25 percent and 47 percent over the 1974-1975 and 1978-1979 periods respectively (Sharpley, 1984). By the late 1970s economic conditions had deteriorated to such an extent that, when combined with manifest signs of social distress (such as a dramatic increase in violent crimes and emigration of skilled workers), they generated a sense of instability and crises (Stephens and Stephens, 1986; Kaufman, 1985). The crises indicated a need for policy changes, and it appears that the government was thinking along those lines as reflected in a Stand-by Agreement it negotiated with the IMF just before the December 1976 elections. The Agreement included, inter alia, a wage freeze, fiscal restraint and a currency devaluation of 20 to 40 percent (Boyd, 1988). However, in the December 1976 election the PNP was returned to office and the new administration promptly rejected the IMF Stand-by Agreement as being inconsistent with the

PAGE 22

6 mandate obtained in the recent election. Instead, and as a way of avoiding agreements with the IMF (Boyd, 1988), and after significant slippages in the external and domestic accounts (Sharpley, 198 4) the government implemented its own adjustment policy measures in January 1977. These included (Boyd, 1988) (1) The reversal of the previous wage indexation policy; ( 2) Implementation of import and exchange controls, a dual exchange rate system and foreign exchange rationing; (3) Suspension of foreign debts for 18 months; and, (4) A search for funding from sympathetic governments. However, it became increasingly clear to the government that in order to attract foreign capital some formal agreement with the IMF was necessary. In this regard, the government signed three agreements with the IMF between August 1977 and June 1979. The first was suspended after four months for failure to meet the stipulated quarterly tests. The second, signed in May 1978, was pursued in all its aspects by the government for a year. Although economic performance under this agreement was poor (Sharpley, 1984; Boyd, 1988), it was re-negotiated in June 1979. However, the agreement collapsed in December 1979

PAGE 23

after the government performance criteria. failed to meet the 7 program's In 1980, in light of increasing social unrest, continued slippages in the fiscal deficits and external accounts, severe foreign exchange shortages, and rising inflation (Sharpley, 1984; Boyd, 1988; Thomas, 1999), Prime Minister Manley called a general election so that the nation could decide on a path of economic development and "what part the IMF should play; or whether it should play any part at all" (Jamaica, API, Manley, 1980). Dubbed "the IMF election", the October 1980 election was won by the strongly pro-IMF JLP, led by Edward Seaga. Expectedly, therefore, both the IMF and the government began negotiations on funding and policies aimed at ameliorating the country's economic and social crises. The new administration relied heavily on funding from external sources. Several loan agreements were negotiated with the leading collaborating lending institutions, the IMF, World Bank, IDB and USAID which have developed a system of cross-conditionalities on their loans, given their basic set of shared objectives on economic policies and on Jamaica (Anderson and Witter, 1994). The prescription by the IMF and the World Bank (Fund/Bank) on how best to deal with the economic crisis in

PAGE 24

8 Jamaica was a series of economic policy reforms. These reforms went beyond the expenditure-switching and expenditure-reducing policies which macroeconomic policy would normally prescribe under these circumstances, to include a built-in bias in favor of a more liberal economic system. The latter aimed at, inter alia, fiscal discipline, liberalizing the domestic market and the external trade sector, privatizing of State Owned Enterprises (SOEs) and other social services, and generally, greater reliance on market signals to allocate resources. Over all, between 1981 and 1992, IMF financial support totaled US$1,036.9 million (under 10 agreements), and that of the World Bank OS$360.4 million (under 8 agreements). 1.2 Problem Statement and Study Justification The cumulative effect of the reforms mentioned above is that the Jamaican agricultural sector is now expected to operate within an economic framework that differs vastly from that of the 1960s and 1970s. Historically, the sector has played an important role in the economy. Agricultural output to GDP ratio is about eight percent, agricultural food export in total export is approximately 20 percent, and agriculture's share of the total labor force exceeds 26 percent (Davis et al., 1999). Given the importance of agriculture in the economy, how the sector performs in this

PAGE 25

9 new policy environment demands urgent attention. In this study, agricultural sector performance is evaluated in terms of crop supply responses. Quan ti tati ve analyses of the impact of these reforms on Jamaican agricultural supply responses are both scarce and generally not particularly robust. In addition, the literature on similar economic reforms in other countries suggests that the outcomes of these reforms are ambiguous, which justifies more empirical work (Khan, 1990). Additionally, this study is justified on the following grounds. First, the economic reforms have generated major changes in the long-term prospects for the national economy within which agriculture will operate. After two decades a good set of data on these policy reforms is now available which should facilitate empirical research on the impact of these reforms on agricultural crop supply responses. This evaluation can throw light on courses of action which policy makers in Jamaica and elsewhere can consider as they seek to reactivate and sustain long-term growth in the agricultural sector. Second, while a few studies have analyzed specific aspects of the impact of the economic policy reforms on Jamaican agriculture (Singh, 1995; Anderson and Witter, 1994; Newman and Le Franc, 1994; and Brown, 1994), what is

PAGE 26

10 lacking is an over-all analytically rigorous assessment of agricultural supply responses within the context of the new policy framework as defined by the recent economic reforms. Finally, analysis of the impact of these reforms on agricultural supply responses is both timely and urgent in light of an on-going debate. A number of writers in the Caribbean and elsewhere have been advancing the view that the kinds of economic reforms implemented in Jamaica and in other countries have harmed agriculture (Anderson and Witter, 1994; Green, 1989; Bates, 1989). In contrast a recent series of studies sponsored by the World Bank suggests that policy reforms similar to Jamaica's should boost agriculture in countries which have traditionally discriminated against the sector (Schiff and Valdes, 1992a; Krueger, 1992). 1.3 Objectives of the study The general objective of this research is to evaluate the impact of economic reforms on selected agricultural crop supply responses in Jamaica over the 1980-1999 period. The specific objectives are to provide ( 1) A statistical evaluation of the performance of the agricultural sector in the pre-reform (1962-1979) and reform (1980-1999) periods.

PAGE 27

11 (2) A descriptive, diagnostic and statistical analysis of the economic reforms that have taken place over the past two decades within the agricultural sector. (3) An econometric (modeling) evaluation of selected agricultural crop supply responses over the 19621999 period, with the number and types of crops chosen for study to be determined by data availability. ( 4) A comparative analysis of agricultural crop supply responses in the reform period with a counter factual, the latter defined as crop supply responses that would have resulted in the absence of the reforms. 1.4 Hypotheses The following hypotheses are advanced in the study: (1) Structural reforms significantly improve agricultural supply responses. (2) Sustained improvements in agricultural supply responses require further broadening and deepening of the reform process. 1.5 Conceptual Issues Agriculture features prominently in policy discussions when (1) it is believed that agricultural policies and institutions help precipitate an economic problem which

PAGE 28

12 policy makers are addressing; and ( 2) when agriculture is an important contributor to employment, GDP, export earnings, domestic food supply, and revenue for the government (Binswanger and Deininger, 1997). Various publications by the government of Jamaica suggest that both of these factors have featured prominently in the recent reforms, while ( 2) was a major consideration in the pre reform (1962-1979) period. Consequently, the conceptual ization is based on two perspectives that are believed to have influenced the design of the policy reforms: ( 1) The Jamaican government perspective. From numerous government publications, it would appear that this perspective is clustered around three main goals: (a) increasing food supply; (b) developing rural areas; and (c) increasing agriculture's contribution to the over-all economy, through, inter alia, employment, contribution to GDP, and export earn ings; and, (2) The IMF /World Bank perspective, whose focus on the agricultural sector clusters around (a) getting prices right; (b) improving efficiency; and (c) rationalizing public expenditure in the agricultural sector.

PAGE 29

13 These two perspectives are not mutually exclusive. On the contrary, the choice of policies in a particular program package results from extensive discussions between officials of the IMF/World Bank and the Jamaican government. The program policy mix therefore reflects the particular economic situation in the country as well as the preferences of the government (Khan, 1990). Despite this, a review of the literature on Jamaican agricultural policy formation suggests that since the late 1970s the influence of the IMF/World Bank perspective has exceeded the Jamaican government's. Given the two perspectives mentioned above, and a review of the content of the reforms since the late 1970s, the evidence suggests that the reforms have focused on three inter-related issues of relevance to the agricultural sector: (i) agricultural (inter-sectoral) terms of trade; (ii) agricultural growth, its adjustment, and supply response; and (iii) efficiency in the agricultural sector (Nallari, 1992; Singh, 1995; World Bank, 1996). Figure 1.1 conceptualizes the Jamaican agricultural sector, dividing the factors impacting agriculture into external (e.g., world prices) and internal. The latter can be separated into those that are exogenous (weather, terrain) and those that are policy induced. Another useful

PAGE 30

,..._ P4 pg P6 P3 P4 P7 Bxternal Trada imports exports Bxogenoua: Weather, Terrain. PS bogenOU8: World Prices P2 P4 PS P6 l I Pl I i PB Pl0 1_,___ ._______. Non-l'ara BOU8e Bolds: resources L---. Agriculture l Production consumption rarm Bouae Bolds Input resources consumption i I---+ :+crops livestock fisheries i H Agri. Product Marketing Non-Agri. Production/ Marketing Agricultural :Input Marketing L------......J~ PB, PlO I Policy Ina~ta: Pl. Agriculture Projects: infrastructure, land development, mechanization, crop improvement P2. Rural industries P3. Monetary Policies: interest rates, exchange rates. P4. Tax Policies: income, indirect, customs duties. PS. Public consumption, rural invest. P6. Rural health, education, welfare. P7. Import/Export controls, tariffs, subsidies. PB. Price controls, input subsidies. P9. Agriculture credit Pl0.Marketing: infrastructure, information. i Agriculture Sector Per~ormance :Indicators: 1. Economic Outcomes (a) quantity flows supply response agri. in GDP i agri. growth rate (b) agri. multi-factor productivity (c) relative prices (TOT, RER). 2. Accumulation Indicators coefficient of investment capital formation agri. employment agri. tax revenues rural nutrition farm incomes a 14 i Figure 1.1: Conceptualization of the Jamaican Agricultural Sector.

PAGE 31

distinction is between prices and non-price 15 factors (including exogenous shocks) that influence agricultural output and income. At the micro level, the price variables include prices of the particular crops, prices of alternative crops, input prices, and the general price level. Government policies exercise both direct and indirect influence on agricultural prices and thus on agricultural performance. Direct (sectoral) policies such as price ceilings, guaranteed prices and trade taxes/subsidies provide incentives to shift resources among crops and sectors of the economy (Binswanger, 198 9) Government policies also can influence output indirectly through macroeconomic (or economy-wide) policies. The most critical components in this regard are fiscal policies and exchange rate policies (Mamingi, 1997). The terms of trade as a policy variable turns on the hypothesis that if all prices were determined in markets, and effective rates of taxation were equalized across commodities, then agriculture's supply responses, growth, income, and contribution to GDP would be higher compared to situations of negative effective protection of the sector (Schiff and Valdes, 1992b). A negative effective protection of agriculture arises when:

PAGE 32

16 ( 1.1) where Pis a general price deflator, Pi denotes the producer price of crop i, P! is the free-market price for crop i, and o is a subsidy to producers as a proportion of cost, Ci. For many developing countries, agricultural pricing policies have consistently kept Pi below and subsidies have not sufficiently (Schiff and Valdes, counterbalanced this 1992b). Given severe disparity budgetary constraints, to achieve at least zero effective protection, the policy implication of (1.1) is to raise the real producer price, or to liberalize all markets thereby eliminating the disparity between Pi and In effect, improving agriculture's terms of trade. The exchange rate sets an upper limit on agricultural export earnings and, when combined with input taxes and subsidies, affects input prices and competing agricultural imports. Changes in the real exchange rate can affect agricultural output and growth by altering the terms of trade between agriculture and non-agricultural sectors. Further, policies that lead to over-valuation of the exchange rate can adversely affect export crops, encourage

PAGE 33

17 rent-seeking activities and generate unproductive uses of resources (Jaeger, 1991). In addition to the price policies, non-price factors constitute an important influence on agricultural output. These factors reflect the material conditions of production in the agricultural sector. Some of these factors include, inter alia, government expenditure and investment in the agricultural sector, construction of rural infrastructure (roads, drainage and irrigation works), extension services, rural credit institutions, and dissemination of relevant scientific information to farmers. The aim of agricultural sector-specific programs is to find a mix of policies for increasing efficiency and productivity in the sector. It is assumed that these two outcomes are necessary (if not sufficient) conditions for increasing agricultural output and growth by reducing costs of production and/or increasing product prices. These policies focus on using public investment in agriculture efficiently, reducing marketing and transport bottlenecks, improving agricultural extension services, health and education (capital formation in agriculture), rationalizing input prices, and enhancing the efficiency of parastatals in the agriculture sector. Behrman (1990) has advanced an important perspective on the channels through which policies affect performance

PAGE 34

18 indicators. He argues that analysis of the impact of policy reforms on performance indicators must explicitly consider the conduits through which the effects of the policies are transmitted to the observed (or desired) outcome. These umeso-level" variables, identified as markets (product, input and financial) and infrastructure (economic and social) are the interface between the policies ( sector specific and economy-wide) and targets. In Figure 1.1, for example, various policies, i = 1,2,3, ... 10, are identified. Some are macroeconomic (e.g., monetary policies and exchange rate adjustments) whereas others are sector specific (e.g., agriculture credit, tariff reductions in the trade sector and elimination of price controls in agriculture) These policies are then combined to achieve specific targets, either at the sectoral or macroeconomic levels. Behrman (1990) emphasizes, however, that the meso level variables condition the effectiveness of policies on the target variables and that analyses of policy impacts must also evaluate how policies have influenced these meso level variables. For example, if poor transportation, lack of effective irrigation, or in-efficient research and extension services restrain farmers' responses to higher

PAGE 35

19 prices, then improving these meso-level variables may do more for the farmer than price increases (Chhibber, 1989). The influence of the factors mentioned above on agricultural supply responses has been well documented in the empirical literature. However, the recent economic reforms have had an enduring effect on these factors. This points to an important issue raised in the literature, viz., how to measure the magnitude of reforms (IDB, 1997). Economic statistics such as exchange rate differentials, inflation rates, tax changes and so on deal with outcomes rather than the policies that gave rise to them. In order to address this issue, structural policy indexes have been constructed by Lora (1997) and others. The index by Lora ( 1997) was constructed for twenty countries in Latin America and the Caribbean. This index seeks to measure the extent of market freedom accorded to economic policies in areas of trade, tax, finance, privatization and labor. In each of these areas indices of market freedom are identified. For example, in trade policy the indices are average tariffs and tariff spreads; in tax policy the indices include, inter alia, tax rates on companies and on individuals; and on financial policy, indices include freedom of interest rates on deposits, loans, reserves on bank deposits, etc. (IDB, 1997). The

PAGE 36

20 structural policy index is a simple average of the indices in the five areas. The index can range from Oto 1, based on the worst and best observations respectively, on market freedom in the country. Further, the index is an important indicator of the extent to which countries are departing from past ways of operating their economies. Jamaica's structural policy index has shown continuous movement towards market freedom, increasing from O. 426 in 1985 to 0.684 in 1995. These index values are higher than the average reported for the twenty Latin American and Caribbean countries in the sample (Lora, 1997). 1.6 Procedural and Methodological Considerations In approaching the general and specific objectives of the study, critical analysis of the Jamaican agricultural sector's performance is first undertaken. To achieve the general objectives and, more specifically, objectives (2), (3) and (4), requires four specific tasks. The first task is to identify and evaluate the structural reforms undertaken since the late 1970s. Both the stabilization and the structural reform policies are evaluated according to their ( 1) theoretical bases; ( 2) breath of vision; and ( 3) logical consistency. Second, supply response models are developed and estimated. Finally, specific objective (4) is

PAGE 37

21 addressed within the context of simulating an alternative path to that of the reforms. Any appraisal of the reforms must compare their impacts not to the pre-reform indicators but rather to some specified, hypothetical alternative. This requires some simulation exercises. The idea here is to generate simulated time series over the reform period for the performance indicators. The simulated series then constitute the counterfactual to which the actual outcomes in the reform period are compared. For the counterfactuals, at least three scenarios appear to be logical extensions. Scenario 1 assumes that the policies pursued over the pre reform period continued into the reform period, i.e., that the reforms over the 1980-1999 period were not instituted. Scenario 2 assumes policies based on the Jamaican government's critique of the IMF /World Bank programs were implemented. Finally, Scenario 3 assumes that IMF/World Bank assistance came with no conditionalities. These components form the basis of the data series for the simulation exercise. The three scenarios appear logical and plausible options given (1) the initial defiance by the government of the mandated reforms in the late 1970s; (2) the frequent failures to meet conditionality tests because the reforms

PAGE 38

22 were not implemented; and (3) government's own protest ations over the years of what are considered "acceptable" reforms. While numerous instances of disagreements exist between the Jamaican government and the IMF/World bank with respect to appropriate reform policies, the clearest statement to this effect is captured in the following quotation: Even though lender and borrower the ref ore shared the common concern regarding the need for economic reform, there is still much debate about appropriate social policies and about the specific effects of manipulating major economic variables such as exchange rates and interest rates. This was illustrated by the lengthy and tortuous negotiations between the government of Jamaica and the IDB and World Bank on the 1989-90 Agricultural Sector Adjustment Loan. (Anderson and Witter, 1994, p.14) 1.7 Summary It is generally believed that developing countries have historically discriminated against their agricultural sectors in favor of industrial development. It is further felt that policies that reverse that discrimination should boost agricultural output and income Over the past two decades far-reaching economic reforms have been undertaken in Jamaica leading to a new economic framework within which the agricultural sector must now operate. This study seeks to assess the impact of these reforms on agriculture crop supply responses, an undertaking that is both timely and

PAGE 39

23 urgent given the importance of the sector in the economy and the absence of rigorous analytical studies on the Jamaican experience with these reforms

PAGE 40

CHAPTER 2 ECONOMIC STRUCTURE, GROWTH, AND POLICY REFORMS IN THE JAMAICAN ECONOMY, 1962-1999 It is useful, at the outset, to situate the discussions in this chapter against the backdrop of three distinct phases of economic policy that characterize the economic history of Jamaica over the period 1962-1999. In the first phase, 1962-1972, the Jamaica Labor Party (JLP) emphasized free markets. However, Bonnick (1984) argues that despite the rhetoric of economic liberalism, in reality the government pursued an import-substitution (IS) strategy based on protectionism, trade restrictions, and price controls. In the second phase, 1972-1980, the People's National Party (PNP) under Prime Minister Michael Manley pursued economic populism and state directed control (dirigisme) in an effort to build democratic socialism. This phase was marked by extensive government intervention, which included nationalization of major industries, price controls, and subsidization of basic foods and some agricultural imported inputs. These policies made significant demands on national resources and created a large bureaucratic economic 24

PAGE 41

25 structure. Consequently, the government was unable to respond to the crisis in the world economy in the mid1970s, and was forced to approach the IMF for stabilization funding in 1977. The third phase, 1980-present, is directly linked to the policies of the previous phases, particularly the 1977 IMF stabilization funding. Phase three can be viewed as having two sub-phases. The first is the return of the JLP to government in 1980-1989 and the implementation of various Fund/Bank stabilization and structural adjustment programs. The second, 1989-present, marks the return to government by the PNP, the continuation of the market led policies of the preceding JLP administration, but more importantly, the intensification of the country's commit ment to liberalization of the economy. The rest of this chapter is organized as follows: Section 2 .1 provides an overview of the structure of the economy and its growth performance. Section 2.2 highlights trends and policies in the agriculture sector. Sector 2. 3 describes the economic reforms, and Section 2.4 provides a summary of issues raised in this chapter.

PAGE 42

26 2.1 Structure of the Economy and Economic Growth: An Overview The relative contribution of economic sectors to GDP in Jamaica has not changed much over the past two decades. The data presented in Table 2.1 and Figure 2.1 show that agriculture's share has been fairly stable, ranging from an average of 8.4 percent over the period 1969-1979 to 8.0 percent over 1990-1998. Although agricultural contribution to GDP is lower than that of the other sectors, these comparatively small percentage contributions are deceptive. Davis et al. ( 1999) have shown that the sector plays an important part in the country's employment, food production, and foreign exchange earnings. The industrial sector, which includes mining, manufacturing and construction, contributed on average 39.1 percent to GDP in the period 1969-1979 but declined to 37.3 percent in 1990-1998. The manufacturing sub-sector also shows a decline from 18.4 percent in 1969-1979 to 16.3 percent in 1990-1998. Finally, services constitute the largest sector of the Jamaican economy, accounting for over 50 percent of GDP over the period 1969-1998. Its contribution to real GDP has increased from 48.5 percent in 1969-1979 to 54.7 percent in 1990-1998.

PAGE 43

Table 2.1: Sectoral Contribution to Real Gross Domestic Product (Period Averages, Percentage). 1969-79 1980-89 1990-98 1969-98 Agriculture 8.4 6.4 8.0 7.6 Industry 39.1 34.5 37.3 37.0 Manufacturing 18.4 20.9 16.3 17.1 Services 48.5 59.1 54.7 53.9 Source: Computed using data from Jamaica, Planning Institute of Jamaica (various issues). Note: The 1998 data used were preliminary estimates. 2000 1600 .-1200 a 41> ., 800 400 0 1969 1973 1977 1981 1985 1989 1993 Years 1969-1998 I -+Real GDP -+-Agri. --Indus. -.Serv. I 1997 27 Figure 2.1: GDP and Sectoral Contribution to GDP in Constant 1995 Dollars. Real sectoral and GDP growth rates are estimated and shown in Table 2.2. The estimation uses a log-linear model, Yt = o + P*Time + Et and corrected for autocorrelation whenever it exists. Over the 1969-1998 period, growth rates were positive but consistently low, while sub-period growth reveal mixed results. GDP growth rates were just over two

PAGE 44

Table 2.2: GDP and Sectoral Growth (1995=100) (Percentage) 1969-79 1980-89 1990-98 1969-98 GDP 2. 3a 2. 4 -0.2 1.10 Agriculture -0.3 1.1 2. 3 0. 6 Industry 0.9 4. 3 -3. Ba 0. 8a Services 4. 6a 0.1 2. 4 1. 6a Source: Computed using data from Jamaica, Planning Institute of Jamaica (various issues). Note: The 1998 data used were preliminary estimates. a, b, indicate statistical significance at the five and 10 percent levels, respectively. 28 percent in the 1969-1979 and 1980-1989 periods. Services growth was reduced from 4.6 percent over 1969-1979 to 2.4 percent over 1990-1998, and was insignificant over the 1980-1989 period. Industrial growth, which was 4.3 percent in the 1980-1989 period, declined significantly to -3.8 percent over the 1990-1998 period. Agriculture growth over the 1990-1998 period was 2.3 percent, compared to its insignificant growth in the two previous sub-periods. Over the entire 1969-1998 period agriculture and industry recorded less than one percent growth, while GDP and services grew just over one percent, respectively. These periodic growth rates mask the highly volatile annual growth rates in these economic series as shown in Figures 2.2 and 2.3. Over the period 1969-1998, annual growth rates for GDP ranged between -9.0 and 13.0 percent, and were negative in 12 of the 30 years. For the

PAGE 45

15 10 5 t,I II .IJ i:: 0 0 J,f -5 Pl -10 -15 1970 1974 1978 1982 1986 Years 1969-98 1 -.-GDP Growth I 29 1990 1994 1998 Figure 2.2: GDP Growth (Percentage Change Over Previous Year, 1995-100). 40 30 20 t,I 10 Ill .IJ i:: 0 0 J,f -10 Pl -20 -30 -40 1970 1974 1978 1982 1986 1990 1994 1998 Years 1969-98 1 -+-Agri. Growth -aInd. Growth _.,_Ser. Growth Figure 2.3: Growth of Total Agriculture, Industry and Services (Percentage Change over Previous Year, 1995=100). 1969-1998 period, agriculture growth rates ranged between -12.6 and 29.3 percent, and were negative in 15 years.

PAGE 46

30 Similar annual volatilities characterize the industrial and services growth rate series. 2.2 Output Trends and Policies in Jamaican Agriculture The growth rates of volume indexes for total agriculture and broad aggregates (food, crops, livestock and cereals) and graphs for these series are shown in Table 2. 3 and Figures 2. 4 and 2. 5. With the exception of the cereals index, which declined over 1971-1998 at an annual rate of -3. 2 percent, the other indexes grew but at less than two percent over the 1971-1998 period. Table 2.3: Growth Rates of Total Agricultural and Broad Agricultural Aggregate Indexes. 1971-79 1980-89 1990-98 1971-98 Total Agriculture 1. 6a 1.7 1. 9a 1.7a Food 1. 6a 1.6 1. 9a 1. 7a Crops 2. 6 1. 5a -3.6 1.7a Livestock 1. 6a -16.5 0.7 1. 3a Cereals 16. ga -9.0 -4. 5 -3. 2 Source: Computed using data from Food and Agriculture Organization, FAO Production Yearbook (various issues). a b, indicate statistical significance at five and ten percent levels, respectively. The composition of agricultural output is shown in Table 2.4. Export agriculture, domestic agriculture, and livestock, forestry and fishing constitute the aggregative components of agriculture in GDP. The data in Table 2.5 and the graph in Figure 2.6 show that domestic agriculture

PAGE 47

31 0 120 0 It) 90 OI OI 60 M 't1 s:: 30 H 4.1 ::, 0 4.1 ::, 1971 1976 1981 1986 1991 1996 0 Years 1971-98 1 -+-Total Agriculture I Figure 2.4: Total Agricultural Output Index (1995=100). 250 0 0 ... 200 It) OI OI 150 M 100 ,:j s:: 1-1 50 4.1 ::, '11 0 4.1 ::, 1971 1976 0 1981 1986 1991 1996 Years 1971-1998 J ---.-crops -+-Livestock --cereal -+-Food Figure 2.5: Output Aggregates (1995=100). Indexes for Broad Agricultural

PAGE 48

32 Table 2. 4: Composition of Agricultural Output (J$M, 1995=100), Selected Years. Agriculture and Subsectors 1969 1975 1980 1985 1990 1995 1998 Total 95.9 105.9 101. 3 72.2 104.4 154.4 120.4 Agriculture Export 34.3 25.9 14.8 13.9 13.6 15.4 12.3 Agriculture Sugar 22.6 17.7 9.9 7.8 9.2 9.4 7.9 Cane Other 11. 6 8.2 4.8 6.1 4.4 5.9 4.4 Main Exports Domestic 33.0 47.1 56.9 38.0 66.2 114.5 82.8 Agriculture Root 14.4 26.9 25.3 14.6 35.1 62.8 42.7 Crops Other 18.6 20.2 31. 7 23.4 31.1 51. 7 40.1 Primary Products Livestock, 28.6 32.9 29.5 20.3 24.6 24.5 25.3 Forestry & Fishing Source: Jamaica, Planning Institute of Jamaica (various issues). Table 2.5: Agriculture Sub-sectors as a Percentage of Total Agriculture (Period Averages). Agriculture Sub-sectors 1969-79 1980-89 1990-98 1969-98 23.47 15.95 11. 21 7.29 Export Agriculture (5.50) (2.24) (1.46) (6.26) 44.78 57.21 68.79 56.13 Domestic Agriculture (5.32) (6.35) (3.62) (11.17) Livestock, Forestry & 31. 73 26.85 20.03 26.59 Fishing ( 1. 35) (4.49) (2.50) (5.65) Source: Computed using data from Jamaica, Planning Institute of Jamaica (various issues). Note: Figures in parentheses are standard deviation.

PAGE 49

80 60 t,I .. i:: 4 0 u l,f 20 0 1969 1973 1977 1981 1985 1989 1993 1997 Years 1969-98 -+-Export Agriculture -Dorne3tic Agriculture --.-Livestock, Fishing & Forestry Figure 2.6: Sub-sectors as a Proportion of Agriculture. 33 constitutes the largest proportion of total agriculture in the 1969-1998 period and other sub-periods. In 1969-1979 the domestic agriculture sub-sector averaged 44. 8 percent in total agriculture and increased to 68.8 percent in 19901998. Both of the other two sub-sectors, export agriculture and livestock, forestry and fishing, have declined over the period 1990-1998 compared to the two previous decades. In terms of growth rates, Table 2.6 shows that domestic agriculture and its sub-groups are the only sub-sectors within the agri-sub-sector that have positive growth rates over the 1969-1998 period. This has compensated somewhat for the negative and low growth in the other agriculture sub-sectors.

PAGE 50

Table 2.6: Growth Rates of Agricultural (Percentage). Agriculture Sub-sectors 1969-79 1980-89 1990-98 Export Agriculture -4. 4a -1. 3 -4 .1 D Sugar -4.2a -2. 6a -5.4a Other main exports -4. 6a 1. 3 1.0 Domestic Agriculture 1. 8 11. 2a 25. 6a Root crops -2.7 19. 7a -1.2D Other domestic crops 4. oa 3. 4 -1.4 Livestock, forestry, & -1.5 -3. 3a 0.1 fishing Jamaica, 34 Components 1969-98 -2 .1 a -2. oa -2.7a 2. 6a 2 6 2. 7a -1. 7a Planning Source: Computed using data from Institute of Jamaica (various issues). a, b' indicate statistical significance percent levels, respectively. at the five and ten The stagnation of agriculture in Jamaica, which can be inferred from the growth rates in Table 2.6 reflects, to a large degree, the state of the agricultural sectors in many less developing countries (LDCs) A plausible explanation for this, is that the development literature in the 1950s and 1960s viewed agriculture as a static sector, from which resources could be shifted to promote industry, considered as the dynamic sector. To a large extent, this view has influenced the kinds of agricultural policies that have been pursued in the past by LDCs. In this regard, Schiff and Valdes (1992a, p.59) state: In many developing countries, the high rate of agricultural taxation has been part of an explicit or implicit policy of industrialization-led growth, justified in part by the belief that industry was the dynamic sector while agriculture was static and not very responsive to incentives. That means that economic growth could be accelerated by shifting

PAGE 51

35 resources from agriculture to industry, by taxing agriculture directly, and by protecting the industrial sector. Many governments in the developing countries have intervened in their agricultural sectors, both directly through agricultural sector policies, and indirectly, through economy-wide policies such as industrial protection. Direct interventions take numerous forms. Some of these include procurement measures (e.g., government marketing boards as sole buyers of agricultural output, and suppliers of major agricultural input) ; quotas and direct taxation on various agricultural export crops; subsidies on farm credit and farm inputs; and quantitative restrictions and tariffs on imported agricultural imports. While some of the direct interventions have benefited agricultural producers, some are tantamount to an implicit tax on agriculture, depressing farmgate prices and farm incomes below levels that would otherwise prevail (Schiff and Valdes, 1992a). Indirect forms of interventions affect agricultural production incentives via macroeconomic policies (e.g., overvaluation of the exchange rate) and industrial protection policies (Krueger, 1992). Various measures have been employed by the Jamaican government to extract surpluses from the agricultural sector. These include taxing agricultural exports,

PAGE 52

36 controlling farmgate prices by the state marketing boards, over valuating exchange rates and reducing internal agricultural terms of trade relative to manufacturing. Studies by Gafar (1980, 1997) and Pollard and Graham (1985) strongly support this proposition. Table 2.7 shows that Tabla 2 7 : Average Ra tas of Growth of Farmgata and F. O. B. Prices, output and the Nominal Protection Coa~~icient (NPC) for the Period 1970-1978, Jamaica. Growth Rates (%)1970-78 a Farmgate F.O.B. Nominal Protection Commodities Prices Prices Output Coefficient b Sugar Cane -1. 25 4.13 -2.90 0.82 Banana 5.06 7.77 -7.43 0.77 Cocoa 1. 64 14.06 -0.54 0.72 Coffee 9.80 8.34 2.02 0.61 Coconut 7.06 4.67 -21.00 n.a. a Adapted from Pollard and Graham (1985), Table 2. b Based on the data in Pollard and Graham (1985), Table 4 for 1970-1979. The NPC is defined as the ratio of farmgate prices (Pr) to F.O.B. prices (Pw) received minus marketing and processing costs (C): NPC =Pr/ (Pw C). over the 1970-1978 period, the nominal protection coefficient (NPC), for Jamaican agriculture was less than one, indicating the extent to which the sector was taxed. For example, a NPC of O. 82 means that the commodity is taxed at a rate of 18 percent. Subtracting one from the NPC gives the nominal protection rate, NPR, which, according to Table 2. 7, is negative, suggesting that producers of the crops reported were not supported, but were taxed instead.

PAGE 53

37 It should be emphasized also, that developing countries have a tendency to overvalue their domestic currencies, so that the official exchange rates used to calculate the NPC overstate internal prices. Hence government taxation of agriculture is even greater than is usually captured by the NPC (Gardner, 1987). More recent data compiled for this study also support the proposition that Jamaican agriculture has been taxed. Table 2. 8 reports on the net barter and income terms of trade between agriculture and manufacturing for Jamaica. The net barter terms of trade, N, is defined here as where PA and PM are the price indexes of agriculture and manufacturing, respectively. Values of N greater than one indicate that prices of agricultural commodities have risen relative to those in manufacturing (Tsakok, 1990). The estimates in Table 2. 8 reveal that over the 1966-1978 and 1989-1999 periods, the net barter terms of trade were unfavorable to Jamaican agriculture, but were favorable over the decade 1979-1988. Although the net barter terms of trade may move against agriculture, the sector can still increase its purchasing power if agricultural output increases pro portionately more than the decrease in agricultural prices.

PAGE 54

38 The purchasing power of agriculture is captured in the net income terms of trade, I, estimated as: I is the agricultural output index. The index I is therefore the net barter terms of trade adjusted for marketed amounts. Increases in I indicate a rising purchasing power of agriculture for buying manufactured goods (Tsakok, 1990) The data in Table 2. 8 suggest that agriculture's ability to purchase manufactured goods have peaked in the 1979-1988 period. Table 2.8: Net Barter Terms of Trade and Agricultural Income Terms of Trade (1980=100; Period Averages). 1966-78 1979-88 1989-98 Net Barter Terms of Trade 0.60 1. 39 0.86 Income Terms of Trade 58.20 159.18 134.03 Source: Compiled by author. 2.3 The Economic Reforms After the late 1970s, Jamaica undertook major changes in economic policies, both sector-specific and economy wide. These policy changes have been an integral part of the conditionalities usually attached to program packages by external funding agencies. The reforms vary in intensity, both inter-temporally and in terms of the areas in which policy reforms were enacted, such as foreign trade, taxation, and liberalization measures. Underlying

PAGE 55

39 the reforms has been a conviction by funding agencies that freer markets allow for a more productive and efficient use of scarce resources, a condition not necessarily accepted uncritically by the Jamaican government. Nevertheless, a number of market restrictions were eliminated, regulations deemed necessary were simplified and made more transparent, and private enterprises were encouraged. The most conspicuous reforms were the liberalization efforts to facilitate foreign trade and financing activities through exchange rate and tariff adjustments. The reform period, 1980-1999, has been punctuated by successive IMF/World Bank stabilization and structural adjustment programs, as well as programs financed by other international funding agencies. The stabilization and structural adjustment programs of the IMF and the World Bank for Jamaica were developed as programs of policy reforms, aimed at reversing, within a few years, the economic crisis in the country. Three facets of this crisis emphasized in the IMF/World Bank literature are structural factors, external factors and poor domestic policies, including an anti-agricultural bias in pricing policies. However, it is a matter of public record that these IMF/World Bank reforms have been resisted by the Jamaican government at various times, and have been hotly debated in

PAGE 56

the popular media regarding the social impacts 40 of externally imposed structural adjustment programs. Despite this apparent dissatisfaction, the government has utilized operational policies along the lines of those prescribed by the Bretton Woods institutions. While in principle the early reforms were aimed at stabilizing the economy, from the outset there was a built-in bias in the policies for a more liberalized economic system. For example, the three year Extended Fund Facility {EFF), which the Jamaican government signed with the IMF in May 1978, contained specific conditions for exchange rate devaluation as well as the liberalization of domestic prices {Boyd, 1988). More extensive economic reforms were advanced by the new administration under Prime Minister Edward Seaga who held office from 1980-1989. Under the 1982 and 1983 structural adjustment programs, quantitative restrictions {QR), on trade were replaced by tariff equivalents, and by 1984 all QRs were eliminated. Exchange rate devaluation was pursued in 19841985, followed by the unification of the official and parallel rates. At the same time, the government began to reduce the public sector via the divestment of state owned enterprises (SOEs).

PAGE 57

41 The 1990s witnessed the most protracted reforms aimed at deregulation and liberalization of the Jamaican economy. The PNP, which took off ice in 198 9 under Prime Minister Michael Manley, had campaigned in the general elections on a platform similar to the JLP, vowing to continue the JLP's market oriented policies. This was an opportunity for a fresh departure for the PNP, which was noted for its previous support for democratic socialism and extensive government intervention. Garrity (1996, p.52) argues that ... the Manley-led government made the decision in 1990 to t~uly embrace the market and proceed with economic liberalization reforms. For Manley ... [t] o continue the reforms of the previous JLP government would not be sufficient to bring about the needed structural changes, nor were there sufficient resources to continue the previous reforms .... Based on the limited available options, Manley's commitment to the liberalization process represented a turning point in state-society relations in Jamaica. In particular, the administration sought an accommodative pattern of governance between state and social actors by energizing the private sector. In this sense, the liberalization process can be viewed as "state-sponsored disengagement from a command role in the economy ... and the creation of an enabling environment for private sector-led growth and development." (Garrity, 1996, p.52)

PAGE 58

42 Against this background, the sectoral adjustment loan, which the government signed with the World Bank in 1989, focused on liberalizing the agriculture sector along the following lines: (1) Deregulation of the coffee and cocoa boards by 1990; (2) Shifts in government focus from production and markets to support services and infrastructure; ( 3) Continuation of the di vestment of agro-enterprises (e.g., sugar); (4) Elimination of subsidies on agricultural credit; (5) Reduction of tariffs; and (6) Elimination of the Generalized Food Subsidies; As a result, by 1991 the generalized food subsidies were eliminated, and complete liberalization of the exchange market and credit interest rate were achieved. Consequently, the overall liberalization process, since the first agreement with the IMF in 1977, opened markets to more competition and reduced the role of the public sector. As would be expected, the conditions that accompanied the loans from the Bretton Woods Institutions involved reforms at both the macroeconomic and the sectoral levels. Figure 2.7 is a schematic presentation of the kinds of economic reforms and a partial list of specific tasks undertaken since the late 1970s. For convenience, the

PAGE 59

\ Overall Objectives Means Components I Domestic 1 Economic Stability, Growth, Improved Living Standard 1 Fiscal Economic/Structural Reforms Monetary/Fin. l Trade Deregulation Reforms Reforms Liberalization Specific Changes Elimination of govt. marketing board Removal of price controls Improve land tenure Sell/lease state land Divestment of govt. assets New pricing system for coffee, cocoa, citrus and pimento and more appropriate sharing of export earnings a, Wage restraint Reduced subsidies Cost recovery Tax changes Expansion of agriculture non-price incentives Elimination of subsidies on interest, loans and food Introduction of Generalized Consumption Tax ', Devaluation Elimination Exchange rate of liberalization quantitative Open capital restrictions account Tariff Banking reductions reforms Elimination Interest rate of exports and credit restraints liberalization Rationaliz* Capital market ing of Reforms import* Creation of licensing Agriculture system Credit Bank Elimination and National of subsidies Development on imported Bank foods. 43 l Other Reforms Social safety net program Land capability classificati on Soil conservation Irrigation rehab. Planned rehab. of the sugar Industry Increase availability of foreign inputs Agro-21 program Hillside project Figure 2.7: 1977-98. Selected Economic Policy Reforms in Jamaica,

PAGE 60

44 policy reforms are organized into five blocks: (1) Domestic Deregulation; (2) Fiscal Reforms; (3) Monetary/Financial Reforms; ( 4) Trade Liberalization and ( 5) Other Reforms. Within each block, specific tasks are listed, reflecting implicitly and explicitly policy changes, which seek to enhance the role of market forces, encourage private sector initiatives and reduce government regulation and intervention in the economy. It is interesting to note that in spite of intense criticisms regarding the reforms' lack of success in producing the economic recovery predicted by its proponents, the trend of economic reforms has not been reversed in periods of economic stress. Such periods have instead been met with the broadening and deepening of reform efforts. Table 2. 9 shows selected economic statistics in the pre-reform and reform periods in Jamaica. In the area of trade, average tariffs in 1986 were 56 percent but fell to 11 percent in 1995. Exchange rate systems have been an important policy instrument in Jamaica to establish restrictions on capital outflows and restrictions for repatriating export revenues and foreign exchange. Following the exchange rate liberalization, many of these restrictions have been dismantled. As evidence of the process of exchange rate unification and deregulation,

PAGE 61

45 Table 2.9 shows that the exchange rate differential (i.e., the difference between the average market price for foreign exchange--inclusive of transaction costs and exchange rate taxes--and the official rate) was 25 percent in 1986 but fell to 5 percent in 1995. Tabla 2.9: Outcomes of Economic Reform Policies in Jamaica, 1986 vs. 1995. Structural Maximum Tax Rate Policy Tariff Exchange Reduction Differential Companies Individuals Index, I Net aver. Tariffs% Percent Percent Percent O
PAGE 62

46 2.4 Summary The evidence presented in this chapter suggests that the direction of change of economic structure of the Jamaican economy that began around 1980 continued over the past two decades. Over the period 1966-1998, growth rates of GDP and economic sectors were low but positive. With few exceptions, sectoral growth rates in the reform period, 1980-1999, were higher than in the pre-reform (1962-1979) period. Some structural change within the agricultural sector was discerned from the data. In particular, export agriculture declined relative to other agri-sub-sectors. Domestic agriculture recorded impressive growth, and it is this agri-sub-sector's overall growth performance that partially compensated for the negative and low growth in the other agri-sub-sectors. Significant economic reforms were implemented, and analysis of their impact on agriculture will be undertaken in a later chapter.

PAGE 63

CHAPTER 3 METHODOLOGICAL AND EMPIRICAL ISSUES IN ECONOMIC REFORMS AND SUPPLY RESPONSE The purpose of this chapter is four-fold. First, it reviews the methods that have been used in the literature to analyze the impact of economic reforms in developing countries. Second, it reviews the literature on supply response analysis. Third, it develops a crop supply response model for Jamaica; and finally, it presents the data sources for the analysis. 3.1 Review of the Economic Reform Literature In this review of the economic reform literature two sets of issues are addressed. First, attention is directed at the way the economic reforms are conceptualized. Second, the analytical methods used to evaluate the impact of the reforms are examined. One of the most challenging problems in capturing the impact of economic reforms on target variables is how to isolate the effect of each of the reforms undertaken. Since the reforms were undertaken in various areas (trade reforms, fiscal/monetary reforms, etc), at different points in time, and with varying levels of intensities, their 47

PAGE 64

48 effects on particular variables in the economy become compounding and difficult to isolate. Khan ( 1990) argues that it is theoretically and empirically difficult to link all the policy reform measures to the ultimate targets of the policies. Hence most studies have attempted to assess the effects of the overall policy package on particular target outcomes. This is the thrust of the studies by Chadha et al. ( 1998) and some of the studies in Tshibaka (1998), Campbell and Stein (1991), La Guerre (1994), Valdes and Muir-Leresche (1993), Commander ( 198 9) and Weeks (1995). In these studies, neither the precise nature of the underlying economic relationships, nor the specific policies adopted are made explicit. Instead, the attention is on whether or not the program package, which in effect gives rise to a particular policy environment, has been "effective" in the sense of achieving broad macroeconomic objectives (Khan, 1990). Most of the studies reviewed proceeded from the premise that the economic reforms originated from the conditionalities packages. In all that accompanied country experiences Fund/Bank-supported the problem of economic instability provided the initial imperative to seek IMF assistance, whi~h, when it came took the form of stabilization policies. However, in most of the studies

PAGE 65

49 reviewed the stabilization policies are treated en passant, or not at all, and instead attention is focused exclusively on economic outcomes associated with structural adjustment 3 For most developing countries that have been pursuing the Fund/Bank economic programs, the last decade has been one in which the reforms have in tens if ied the process of liberalizing the economic system. A frequently raised question with respect to Fund/Bank-supported structural adjustment programs cum structural reforms for economic liberalization, has been whether these programs are effective in achieving stated economic objectives. A second and related question is how to measure the effects of these reform policies on the target variables identified in the programs. With respect to this latter question, Guitian (1981) has argued that economic performance under a policy package should be compared to a counterfactual. The latter is defined as economic outcomes that would have taken place in the absence of the program package. The concept of a counterfactual is intuitively appealing and is a standard yardstick widely used in economics to measure the impact of policy interventions. Since the counterfactual is Stabilization policies are designed to put the economy back on its equilibrium path, whereas, structural adjustment policies aim at putting the economy on a new (higher) equilibrium.

PAGE 66

50 unobservable, it has to be estimated, hence alternative methodologies used to evaluate the program effects are judged in terms of their estimates of the counterfactual. Although the preceding two questions have generated a large body of literature over the past two decades, there is little consensus in the economics profession either about the impact of past programs on target variables, or about how to estimate the effects of program packages (Khan, 1990). With respect to measuring the impact of reforms, there are three main approaches that have been applied in the empirical l i terature. These are ( 1) The before-after approach, which compares the behavior of key macro-economic variables before and during, or after, any particular reform period, or policy package. ( 2) The with-without approach compares the performance of macro-economic variables of non-program countries (the control group) with those from program countries. A modified version of this approach is a reduced form regression estimate that controls for initial conditions in program and non-program countries. (3) The comparison-of-simulations approach. This approach simulates performance of Fund/Bank-type

PAGE 67

51 programs then compares them with simulated outcomes from another set of policies. Approaches (2) and (3) have been used extensively in cross country studies to assess the impact of Fund/Bank-supported programs. The before-after approach has been the most prevalent in country case studies using time series data. 3.1.1 Before-After Approach The before-after approach has been the most popular in the literature on Fund/Bank-support programs. Reichmann and Stillson ( 197 8) were the first to use this approach in an examination of 79 Fund/Bank-supported programs over the period 1963-1972. They compared growth, balance of payments and inflation in the two years prior to and after the implementation of the program. Connors (1979) also used the before-after approach to evaluate 31 programs in 23 countries over the period 1973-1977. More recent studies include Singh (1995) and the papers in Le Franc (1994), on Jamaica. Similar studies on other developing countries include most of the papers in Tshibaka (1998), Campbell and Stein (1991), La Guerre (1994), Valdes and Muir-Leresche (1993), Commander (1989) and Weeks (1995). The before-after approach is viewed by some analysts as providing a relatively poor estimate of the counter f actual to program effects by assuming that non-program

PAGE 68

52 determinants of economic variables remain constant between non-program and program periods. This assumption has been questioned in the literature in light of the fact that non program determinants of economic outcomes (e.g., terms of trade variations, changes in international interest rates, weather, etc.) typically change year after year. Conse quently, Goldstein and Montiel (1986) have demonstrated that the before-after estimates of program effects are biased, since all economic changes in program period are incorrectly attributed to program factors. These authors suggested further that the before-after estimates are unsystematic over time, since in any particular year program effects will often be dominated by non-program factors. For example, if a hurricane damages infrastructural works and agricultural crops, then agricultural growth may decline, causing all programs in that year to appear to have performed poorly. 3.1.2 With-Without Approach The with-without approach seeks to overcome the shortcomings of the before-after approach by comparing economic outcomes between program and non-program countries. Since both groups of countries are subjected to the similar external environments, it is argued that this comparison cancels out the non-program determinants, so

PAGE 69

53 that any observable differences in outcomes in the two groups of countries are attributable to the Fund/Bank supported programs. In other words, the economic outcomes observed in non-program countries are taken as the counterfactual of what would have occurred in program countries in the absence of Fund/Bank-supported programs. Donovan (1981, 1982) was the first to use this approach on a sample of Fund/Bank-supported programs implemented between 197 0-198 0. Loxley ( 198 4) also applied this approach to 38 less developed economies (income of $690 or less) with Fund/Bank program during 1971-1982. Other similar studies include Gylfason ( 1987) and Pastor (1987). A major problem with the with-without approach is that countries in the sample of non-program and program countries are not randomly selected. They are program countries because of poor economic conditions prior to entering into a Fund/Bank-support program. Consequently there is a systematic difference between these two groups of countries and the non-random selection of program countries produces a biased estimate of program effects. This is the result of attributing observable differences in economic outcomes between program and non-program countries to program status, when in fact the initial economic

PAGE 70

54 position of the two groups of countries is an important determinant of economic performance. To overcome this problem requires identifying the specific differences between program and non-program countries in the pre-program period and controlling for these differences prior to comparing economic outcomes. This is the idea underlying a modified version of the with without approach. In this regard, Goldstein and Montiel (1986) proposed a generalized evaluation estimator to control for pre-program differences between program and non-program countries. This estimator is a reduced-form relationship that links changes in macroeconomic outcome (program target variables) to lagged values of the target variables, lagged values of policy variables and variables that represent exogenous effects on the target variables. These authors applied this approach to a sample of 58 developing countries in which 68 programs were implemented over the period 1974-1981. The approach was extended by Serieux (1999) to include the effect of democratization in Fund/Bank-supported program countries. 3.1.3 Comparison of Simulations Approach Finally, the simulation approach relies on simulations of economic models to make inferences on hypothetical outcomes of Fund/Bank-type policy packages. Khan and Knight

PAGE 71

55 (1981), created a simulation using panel data for 29 developing countries in a dynamic econometric model. The objective was to investigate the hypothetical effects of pursuing a stabilization program with similar policy characteristics as a Fund/Bank-supported one. The authors later (Khan and Knight, 1985) extended their simulation exercise to include comparisons of alternative packages. More recent studies include Robinson and Gehlhar (1996), and Chadha et al. (1998), who use computable general equilibrium (CGE) models for the Egyptian and Indian experiences with recent economic reforms, respectively. 3.2 Preliminary Issues in Modeling Supply Response in Jamaica A number of empirical studies have been conducted in the past to estimate crop supply response in Jamaica. A partial list of these studies is shown in Table 3.1. With the exception of Ga far ( 1997), all of the studies were conducted prior to the period 1980-1999. In addition, none of the studies addressed the issue of the impact of economic reforms on agricultural supply response. One approach to capturing the long-run and short-run changes in agriculture supply response is to use the Nerlovian-type partial adjustment models. Both quantity and prices can be modeled to adjust to their long-run or equilibrium path, and the model is capable of estimating

PAGE 72

Table 3 1: Empirical Responses in Jamaica. Studies Crop/ Category Period Author & Year Banana 1954-1972 Gafar (1980) 1961-1979 Pollard & Graham (1985) Cocoa 1954-1972 Gafar (1980) 1961-1979 Pollard & Graham (1985) Coffee 1953-1968 Williams (1972) 1954-1972 Gafar (1980) 1961-1979 Pollard & Graham (1985) Citrus 1961-1979 Pollard & Graham (1985) Sugar 1954-1972 Gafar (1980) 1961-1979 Pollard & Graham (1985) Broad Ag:ri. Ag:g:reg:ates 1964-1990 Gafar (1997) Export Domestic Livestock Forestry & Fishing Total Agri. on Agricul. ture 56 Supply Funct. Price Elasticity Form Short-run Long-run Linear 0.16 0.57 log 0.49 -2. 72 Linear 0.41 2.56 Log 0.74 0.76 Log 0.70 -0.80 Linear 0.92 1.15 Log 0.10 0.07 Log 0.24 -1.33 Linear 0.17-0.29 0.31-0.7 log 0.24 1. 41 log 0.20 0.35 0.15 1.08 0.15 0.21 0.02 0.21 0.12 0.23 Source: Adapted from Gafar (1997, p.213). both long-run and short-run parameters as well as the speed of adjustment towards the long run equilibrium. A potential complication in such an analysis of long-run and short-run changes is that most economic time series data are non stationary and usually characterized by a unit root. This means that the linear properties of the series such as its mean and variance, are not constant over the sample but change over time (Greene, 1993; Gujarati, 1995). Nelson and Plosser (1982) have shown that the economic implications of an economic time series that is

PAGE 73

57 characterized by a unit root are different from those of a stationary process. In particular, an economic time series with a unit root will have a permanent response consequent upon any shock in the system. In contrast, a stationary series will reflect only a transitory response. Therefore, the response to economic reform policies is not simply a "policy on", "policy off" or one-shot choice. Even if the reforms were temporary, so long as the economic series possesses a unit root, there will be a permanent response. Non-stationarity of variables poses problems of estimation of functional relationships using conventional econometric methods. In the first place, Fuller (1976) has shown that under non-stationarity the limiting distribution of the asymptotic variance of the parameter estimates is not finitely defined, hence the conventional t and F tests are inappropriate. Secondly, non-stationarity gives rise to spurious correlation among variables (Greene, 1993). In macroeconomic time series it is not unusual to find that a variable is non-stationary in its level 4 but stationary in first differences. In technical terms, if Yt is nonstationary, but /J.Yt (the first difference) is stationary, then Yt is integrated of order one, i.e., Yt~I(l). If two Data that have not been transformed in any way (such as logarithmic transformation, first differences, etc.), are said to be in 'level' form.

PAGE 74

58 series Yt and Xt are I(l) then a linear combination of them, e.g., Zt = Yt ClXt may also be I ( 1) However, there may also exist, a value for a that ensures that Zt is stationary. In such an event, the two series are said to be cointegrated, and the cointegrating vector is denoted (1, a). It is tempting to conclude that in order to estimate meaningful relationships among non-stationary variables, all that is necessary is to difference the variables until they achieve stationarity and then estimate the relationship. However, Johansen and Juselius ( 1990) argue that unless the difference operator is also explicitly applied to the error process, such differencing results in loss of information. In this event, resorting only to estimating the relationship in difference form captures only the short-term relationship among the effects, variables while is the left long-term undetected (Nickell, 1985) Finally, differencing the economic series may not be appropriate, such as when economic theory postulates a relationship among variables in levels, not in difference forms. To overcome these problems, econometricians have developed an approach known as error correction models based on cointegration. Hylleberg and Mizon (1989, p.124) claim that

PAGE 75

59 ... when estimating structural models it is our experience from practical applications that the error correction formulation provides an excellent framework within which, it is possible to apply both the data information and the information obtainable from economic theory. Error correction modeling requires two conditions (1) All variables must be integrated to the same order, i.e., Yi~I(d), where dis the number of times Y has to be differenced to achieve stationarity; and, (2) All variables must be cointegrated of order (d b), where b>O. The idea underlying cointegration is that one or more linear combinations of non-stationary variables are stationary. That is, cointegration approaches the station arity issue as linear combinations of economic series, rather than by differencing the series. The implication of this is that if a set of variables is cointegrated then, following the Granger Representation Theorem (Fuller, 1985) a valid error correction representation of the data exists. In effect, then, cointegration is a test of existence of a long-run relationship of variables that are integrated of the same order (Greene, 1993; Gujarati, 1995). However, an important feature of error correction models based on cointegration is that the data in both levels and differences are included, thereby facilitating investi gation of both short-run and long-run effects in the data.

PAGE 76

60 3.3 Error Correction Model The error correction model (ECM) can be derived from a re-parameterization of an Autoregressive Distributed Lagged (ADL) model (Hendry et al., 1984). Alternatively, the ECM can be derived from the dynamic optimizing behavior of economic agents. This latter approach is presented here. Following Nickell (1985), suppose economic agents optimize their behavior with respect to an inter-temporal quadratic loss function: ( 3. 1) L == f CX. 5 ~l (Yt+s Y:+s} + {Yt+s Yt+s-1 ) 2 2 02 s=O where y* is the desired or long-run equilibrium value of Y that the economic agent can control to minimize L, conditional on information at time t, and subject to movements in Y. The discounting and weighting factors are a, (OO), respectively. Minimizing ( 3 .1) with respect to Yt+s gives a second order difference equation whose solution at time tis: (3.2)

PAGE 77

61 where 1 is the stable root {i.e., the root that is < 1) from solving the characteristic equation of the general Euler equation which was derived from minimizing (3.1 ) Equation ( 3. 2) is an optimal rule and has an error correction term in brackets [.] ; the coefficient ( l-) is the speed of adjustment, i.e., the speed of closure to any discrepancy between desired and actual values of Y. In the error correction term in ( 3. 2) the long-run target is a convex combination of all target values from Y ; 1 onwards. Nickell (1985) has shown that when 8 2 = 0 in the loss function (3.1) then equation (3.2) nests the forward looking partial adjustment model (PAM): (3.3) It is important to note that the dynamic equation ( 3. 3) does not necessarily have to be of a standard PAM type. To estimate (3.3) empirically requires the parameterization of Y ; +s. One way is to model this sequence of expected target values as a stochastic process such that Y ; +s are expressed in terms of current and lagged values and substituted into (3.3). The resulting equation may not necessarily result in a PAM. Indeed, Nickell demonstrates that when

PAGE 78

62 ... we allow the target [Y ; +J to follow anything more complex than a first order autoregression, the structural equation [ 3. 3] which is fundamentally a partial adjustment model, will reduce to an error correction mechanism in terms of observable variables (Nickell, 1985, p.124). Suppose that the actual values of the series follow a second order autoregressive scheme with a unit root and a drift: ( 3. 4) where s>O, Et+s is white noise, and g is the drift. Denoting the expected value with an (*), the expectation of (3.4) is: (3.5) Following the derivations by Nickell (1985) and Alogoskoufis and Smith (1991) the solution to the second order difference equation (3.5) is: ( 3. 6) Substituting (3.6) into (3.2) yields the decision rule in the form: ( 3. 7) where c 0 2 + (1 1 X1 eJ ( ) tlYt = C + ( ( )\ tlYt + 1 AYt-l yt-1 1 + 0'. 1 1 J = g(l ~x2 ~0'.X1 X1 eJ (2 ~)2[1 + 0'. 1 (1 ~)] g(a X1 1 ) (2 ~xl 0'.) Equation (3.7) is written in the form of an ECM. The parameterization of y in terms of exogenous variables would

PAGE 79

63 reflect the long-run cointegrating relationship. According to Nickell (1985, p.124), since it is almost a stylized fact that aggregate quantity variables in economics follow second order autoregression with a root close to unity, we may find the error correction mechanism appearing in many different contexts. 3.4 An Error Correction Model for Crop Supply Response in Jamaica The dynamically unrestricted version of the ECM in (3.7) can be expressed as: ( 3. 8) where Otis output in logarithm and Wo, W1, W2 are straight forward analogues of the intercept and coefficients in (3. 7) Generally, o: is parameterized in terms of other (weakly) exogenous variables. This parameterization shows the long-run cointegrating relationship between the exogenous variables and the dependent variable, Ot. Given the unrestricted nature of (3.8) a wide range of possible processes that describe the law of motion of Ot can be accommodated. Following previous agricultural supply response models, quantity supplied is postulated as a function of the expected values of a set of variables that are believed to capture agricultural incentives. These exogenous variables are, the price of the crop, Pc, the price of substitute crops, Psi, i=l, 2. . and the prices of

PAGE 80

inputs. In this study, two inputs are 64 considered, fertilizer and labor, whose prices are denoted as F and W, respectively. With these specifications of the exogenous variables, the parameterization of the long-run equilibrium supply function of the c th crop is: ( 3. 9) c = l...n ; i = 1 2 ... ; t = 1... T ; &t ~ii d ( 0 er) where all variables are measured in logarithms (to facilitate interpretation of estimated coefficients) and have been previously defined. The superscript e denotes expected value of the variables. Given the parameterization in 3.9, the general error correction model, with all variables as previously defined, can be written as: (3.10) b.Qt = E\.o + f\.lb.Qt-1 + f\.zb.P:,t-1 + E\.31b.P:i,t + f\.4b.w:-1 + f\.5b.F; _l + f\.6(Q~-1 Qt_i) + \) To give empirical content to these models, the specification of expected values must be addressed. Clearly if economic agents had full information about the current variables at the time they are set, then actual values of the variables would be substituted for their expected values. When this is not the case, some process of expectations formation has to be assumed. The applied literature on ECM has largely ignored issues of

PAGE 81

. ~. 65 expectations. Because of its simplicity in practice, actual values are usually substituted for expected values. An alternative to this procedure is to assume that crop and input prices are determined by policy changes for one year in advance. With this assumption, expectations about current economic variables have to be based on information that is available up to the end of the previous period, (t 1). This can be expressed as: ( 3. 11) x ~e = E(x t l
PAGE 82

66 (3.13) (3.14) (3.15) Using (3.8)-(3.15), the general error correction model for the c th crop is: 2 2 (3.16) !:J.Q c ,t = O'.oc + 0'.1c,tf:J.Qc,t-l + L 0'.2jf:J.Pc,t-j + L 0'.3 i k f:J.Ps i, tk j = l k = l 2 2 + L 0'.41!:J.Wtl + L 0'. 5m f:J.F t m + aio c <5 1 P c 0 2 1 P si 1 = 1 m= l where, i=l,2, all variables as previously defined, and the random error term is suppressed. The last term in (3 .16) is the error correction term. In the empirical estimation of the ECM (3.16), the error correction term is usually specified as the residual from the cointegrating relationship. The Engle-Granger (1987) two-step method has been used extensively in the applied literature to estimate the ECM (3.16). However this method assumes that the cointegration vector is unique. Except in the bivariate model, this assumption may be violated in multivariate models. To test for, and estimate multiple cointegrating vectors, Johansen (1988) and Johansen and Juselius (1990) have devised an appropriate method within the following framework. Define a

PAGE 83

67 standard vector autoregressive (VAR) model with lag length k as: (3.17) t=l.. T where X is an Nxl vector of N endogenous variables, Et~iid ( O, A) with dimension NxN. The long-run, or cointegrat i ng matrix is: ( 3 18) I Il l II 2 ... Il k = II The number of distinct cointegrating vectors, r, which exists between the variables of X, is given by Rank ( IT ). Most economic time series appear to be integrated to the order of one, in which case, r~N-1, where N is the number of variables in the vector X. In the case of a bivariate model, N = 2, and therefore if the variables are cointegrated, then there is a unique cointegrating vector. The matrix rr is then decomposed as: (3.19) II = a.W where represents the matrix containing the r cointegrating vector, and a. is the matrix of weights with which each cointegrating vector enters each of the differenced X equations. A large a value implies that the system will respond to any deviation from the long-run equilibrium path with a rapid adjustment. If a' s are zero

PAGE 84

68 in some equations, this is a sign of a weak exogenei ty, implying that the variable does not respond to the disequilibrium in the system. The parameters a. and form an over-parameterization of the model. However, the space spanned by~, sp{~), can be estimated, and shown to be the empirical canonical variates of Xt-k with respect to AXt. This is in effect the following theorem advanced by Johansen: The maximum likelihood estimator of the space spanned by~ is the space spanned by the r canonical variates corresponding to the r largest squared canonical correlations between the residuals of Xt-k and AXt corrected for the effect of the lagged differences of the X process. {Johansen, 1988, p.233) Implementation of this theorem begins by re-para meterizing {3.17) (detailed derivations are in Johansen, (1988) and Enders, (1995)), into the following error correction model: (3.20) where ri = -I + TI1 + II2 + ... + IIi, i=l .. k Without any loss of information, the ECM in (3.20) is therefore a transformation of the VAR{k) model in equation (3.17), and is expressed in first differences and augmented by the error correction term, rkxt-k The long-run equilibrium or impact matrix is the matrix rk and is equivalent to IT = a'~ in (3 .19). The rank of IT is the basis of

PAGE 85

determining the number of 69 cointegrating relationship between the variables in the ECM (3.20). Johansen (1988) identifies three possibilities with regards to Rank(IT ) (1) rank(Il) = 0. This means that the variables are not cointegrated and the model is basically a VAR in first differences. ( 2) 0 < Rank (IT) < p. In this case, the variables are cointegrated relationship(s) and is the number less than variables, p, in the model. of the cointegrating number of (3) rank(Il) = p. This means that all variables are stationary and the model is in effect a VAR in levels. The loglikelihood representation of (3.20) is: (3.21) L(et, ~, 0) = ,ni -T/2 Johansen's procedure begins by regressing ~Xt on the lagged differences of M{t and generating fitted residuals Rot, then regressing Xt k on the lagged differences and generating fitted residuals, Rkt These fitted residuals are then used to construct the following product moment matrices: (3.22) 1 T -R j t' TL R it i = l i, j = 0, k

PAGE 86

70 The product moment matrices (3. 22) are then used to find the cointegrating vectors by solving the determinant: (3.23) This will yield the estimated eigenvalues ... An) and eigenvectors (v 1 ... Vn), which are normalized such that: (3.24) v's k'.k'. v = r where Vis the matrix of eigenvectors. The most significant eigenvectors then constitute the r cointegrating vectors, i.e. (3.25) Using (3.25), a is then estimated from (3.19). The critical issue in all of this is to determine which, and how many, of the eigenvectors in (3.24) represent significant cointegrating relationships. First, the vectors that have the largest partial correlation with Axt, conditional on the lags of AXt, are identified. Second, the eigenvectors that correspond to the r largest eigenvalues are chosen. Finally, to determine the value of r the following test statistics suggested by Johansen (1988) are employed: n (3.26) n 1 (q, n) = -T L ln(l li) i =q+l (3.27)

PAGE 87

71 The null hypothesis HO: r:$q is tested with ( 3. 2 6) while HO: r = q is tested against Hl: r = q + 1 with (3.27). The critical values for these tests are taken from OsterwaldLenum (1992). The critical values from this source recalculates and extends those critical values from Johansen (1988) and Johansen and Juselius (1990), to handle a full test sequence from full rank ( r = p, i.e., Xt is stationary) to zero rank ( r = 0, i.e., all linear combinations of X are I ( 1) ) for at most 11-dimensional systems. The cointegration technique is used to determine the long-run relationships among variables. This co-movement of variables in a long-run supply function has not been explored for Jamaica. Cointegration analysis is appropriate in this regard. It will also suggest which variables in the supply function are in the long-run equilibrium. 3.5 The Data The principal sources of annual time series data, which are used in this study, are the Food and Agriculture Organization (FAO) agricultural database (FAOSTAT), avail able on the internet, and annual publications of various government agencies in Jamaica. These include, Economic and Social Survey of Jamaica (Jamaica, Planning Institute of Jamaica); Production Statistics (Jamaica, Statistical

PAGE 88

Institute of Jamaica (STATIN)); Statistical 72 Digest (Jamaica, Bank of Jamaica), Statistical Year Book of Jamaica (Jamaica, Statistical Institute of Jamaica (STATIN)); Census of Agriculture, 1968, 1978, 1996 (Jamaica, Statistical Institute of Jamaica (STATIN)) ; International Financial Statistics (IMF); and data from published and unpublished documents from the Ministry of Agriculture in Jamaica. It should be noted that the data on Jamaica contained in the FAO database, FAOSTAT, IMF and World Bank sources are based on publications and data supplied by STATIN, Ministry of Agriculture, and other agencies of the Government of Jamaica. The data used to estimate the supply functions in this study are taken from the FAO FAOSTAT database. This is the most comprehensive data set on crop output and prices in Jamaica. However, in several areas the data are less than ideal. For example, whenever data are not available the FAO provides its own estimates based on past crop performance, other crops and other country's data. Jamaican officials claim that production data for some crops reported on the FAO data base are not monitored in Jamaica. The crop output and price data are annual series, and are collected at the farmgate through periodic production

PAGE 89

73 surveys and agricultural censuses. Farmers in the Caribbean rarely keep records of production, cost expenditures and so on. Hence, for these farmers it is difficult to recall data for production surveys that are conducted between long periods. Consequently, this casts some amount of suspicion on the reliability of the data. Nonetheless, these are the data reported as official statistics, and are used by the government as a basis for policy analysis/discussion. Several important economic aggregates were not adequately covered in the data sources. In particular, data on agricultural wages are not available on Jamaica. As a result, a proxy for this variable was constructed from data on compensation to agricultural employees recorded in the national accounts. Data on consumer price indexes--the basis for def la ting crop and input prices in this studywere not available as continuous series over the 1962-1999 period. Consequently, a combination of splicing indexes and converting the final 1962-1999 series to a 198 0 base year was undertaken. Although the data are less than ideal in several areas, and are bound to contain some noise, the advantages of using them are:

PAGE 90

7 4 (1) they are from a common source and are characterized by a common accounting/estimating procedure in their derivation; and (2) the data are the most comprehensive time series available on Jamaican crop supply in a single source. 3.6 Summary There are competing approaches in the literature on how to evaluate the impacts of economic reforms associated with Fund/Bank-type policy packages. Each approach utilizes the idea of a counterfactual against which actual economic outcomes in the program period are compared. Estimating crop supply responses over a period in which significant policy changes have occurred, requires a modeling framework that is capable of capturing both the long-run and short run changes. The Nerlovian-type supply response models have been used in the literature for this purpose. However, in the context of data series that are non-stationary, this approach can produce spurious regressions. A more appropriate analytical framework is that provided by error correction models based on cointegration analysis.

PAGE 91

CHAPTER 4 EMPIRICAL ESTIMATION OF CROP SUPPLY RESPONSE The aim of this chapter is four-fold. First, the central motivations for using error correction modeling based on cointegration theory in this study are presented. Second, the time-series data on prices and quantities to be used in the ECMs to estimate crop supply responses in Jamaica are tested for stationarity. Third, long-run supply responses are estimated for eight agricultural crops. Finally, the short-run dynamics of the crop supply responses are analyzed. 4.1 Motivations for Using Cointegration Analysis Chapter 3 provides a fairly elaborate treatment of the theoretical and statistical aspects of error correction modeling based on cointegration theory. Against that background, it is useful to recall in a cryptic, condensed and fairly non-technical way, the central motivations for using this approach to address the issues with which this study is concerned. The empirical purpose of this study is to investigate the impact of economic reforms on crop supply responses in 75

PAGE 92

76 Jamaica. Generally, supply response models in agriculture postulate a long-run relationship between output and agricultural incentives (Askari and Cummings, 197 6) Deviations from this long-run equilibrium occur in the short-run and may involve considerable adjustment costs to the economic agent. This is especially the case when significant policy changes are implemented. Jamaica provides a good case for evaluating the connections between economic reforms and supply responses. The country has traditionally been heavily dependent upon agriculture for food, employment and export earnings. A combination of excesses in state intervention and adverse world economic conditions prompted major economic reforms in the late 1970s in the form of serious macroeconomic stabilization policies under the direction of the IMF. During the early 1980s more far-ranging structural reforms were instituted in an effort to reverse the current account and fiscal deficits, reduce inflation and monetary growth to achieve financial stability, restore economic growth, and so on. Implicitly in these early reforms, but more explicitly in the late 1980s and throughout the 1990s, the aim has been to re-orient the economy towards a more liberal economic system. These reforms took expression in progressive devaluation of the Jamaican dollar, elimination

PAGE 93

77 of state marketing boards, liberalization of agricultural input and output prices and privatization of state held monopolies and public enterprises. Intuitively, therefore, the analysis of supply responses in these situations would require a modeling framework that is capable of incor porating both the long-run and short-run changes. A preliminary analysis of the Jamaican crop output and price data series over the 1962-1999 period reveals sub stantial fluctuations, especially since the early 1980s. The strongly time trended data imply a statistical problem that has not been addressed in applied analysis on Jamaican crop supply responses. This statistical problem is referred to, in the literature, as non-stationarity of the time series. The analytical approach chosen in this study to deal with the above mentioned issues, emphases the importance of considering the interactions between the variables in the system in a simultaneous equation model, and to distinguish between the short-run and long-run effects. The modeling approach differs from those previously used to model Jamaican crop supply responses in two very important ways. First, t?e data are analyzed as a full system of equations. This allows for possible interactions in determining the precise relationships among the variables in the system.

PAGE 94

78 Second, the multi variate cointegration modeling which is used in this study, is designed for this type of empirical work by explicitly classifying the non-stationary and stationary components and facilitating an inter pretation in terms of the dynamics of short-run and long run effects. There are two general motivations for using error correction models based on cointegration theory in applied economic analyses. First, when time-series data are non-stationary, i.e., their linear properties, such as mean and variance, are time dependent, then conventional econometric analysis on such data may produce "spurious" results. Second, even though a set of time series may individually be non-stationary, there may be a linear combination among them that is stationary. Such series are said to be cointegrated, that is, they have a tendency to move together in the long-run, even though in the short-run they may diverge from each other. This co-movement of related series suggests the existence of long-run relationship between them. So when data are non-stationary there is generating the additional possibility that process contains information the data about the equilibrium process that makes the process adjust toward the long-run steady state or the equilibrium path.

PAGE 95

There is, 79 therefore, additional economic insights gained from the recognition that the data are non stationary. The component jkXt-1, (= cx[3' Xt-1, since II = cx[3'), in equation (3.20), is directly related to the non stationarity in the data. Since this component contains information about the speed of adjustment, ex, to some long run relations, [3'Xt, and if the data are cointegrated, then the economic process can generally be understood within a theoretical model that assumes some adjustment behavior. This means that theoretical models such as the partial adjustment model (PAM), which assume static equilibrium, cannot be used when data are non-stationary. 4.2 Definition of Variables In this chapter, error correction models based on cointegration theory are used to test the hypothesis that long-run relationships exist between crop output and price incentives in Jamaica. The estimation procedure is based on the work of Johansen (1988) and Johansen and Juselius (1990). There are basically three steps in this estimation procedure ( 1) test the order of integration of the variables and specify the lag length of the variables using a standard vector auto-regressive (VAR) specification.

PAGE 96

80 (2) estimate the ECM and the number of cointegrating relationship(s) among the variables included in the ECM. (3) perform short-run analysis by conducting innovation accounting on the ECM. For each crop a p-variable vector autoregression (VAR) of lag k is specified. The model is re-parameterized into an error correction model (ECM) as specified in equation (3.20) and estimated according to the methodology of Johansen (1988) and Johansen and Juselius (1990). Esti mation is done using the Regression Analysis for Time Series (RATS), version 4. 3 (Doan, 1996), and Cointegration Analysis for Time Series (CATS) in RATS, (Hansen and Juselius, 1995), computer programs. The variables in each ECM include the output (quantity) of the crop of interest, the price of the crop, the price of a substitute (or alternative) crop, and two input prices, average agri cultural wage rate and average fertilizer price. The choice of a crop price variable is critical for the estimation of crop supply responses. In this regard, Askari and Cummings (1976) suggest using any one of the following prices: (1) Nominal farmgate price; (2) Farmgate price deflated by any one of the following:

PAGE 97

(a) a price index of farmer's inputs; (b) a consumer price index; and 81 (c) some index of the prices of competitive crops (or the price of the most competitive crop). An additional issue that is related to the choice of an appropriate price variable for the crop supply functions especially within the context of the IMF/World Bank programs in Jamaica, is that agricultural price incentives are influenced by various macroeconomic policies. Of particular importance, in this regard, is the real exchange rate (RER), defined as the ratio of prices of tractable (PT) to non-tractable (PN) goods (Tsakok, 1990). Krueger (1992) and Schiff and Valdes (1992a, 1992b), have shown extensively how macroeconomic policies in developing countries generally have a RER effect, which ultimately affect output price and agricultural supply. In the empirical literature the RER is usually approximated as RER=e*WPI/CPI, where WPI is the foreign wholesale price index, CPI is the domestic consumer price index, and e is the official exchange rate. The World Bank's approach to showing the link between macroeconomic policies (as represented by the RER) and real crop prices, decomposes real producer price as follows: RPP =PF/CPI= PF/Pae= e(WPI/CPI) = NPC*RER*p 5

PAGE 98

82 where RPP is real producer price, PF is the farmgate producer price, PB is the border price, NPC is the nominal protection coefficient, and PB is the real border price of the country's exports (World Bank, 1994). This definition of RPP shows that it contains information on the RER and also reveals that it is really hazardous to include both the RER and RPP in the same equation (Mamingi, 1997). In addition to the issues raised above, two additional considerations were taken into account in the choice of the crop price variable. First, since farm producers sell most of their marketable surplus immediately after reaping, farmgate prices seem to be a good approximation for prices received. Second, since farmers purchase most of their requirements from retail markets, the consumer price index seems to be an appropriate def la tor for producer prices. For these reasons, therefore, the farmgate price deflated by the CPI was used as the real producer price (RPP) variable in the supply functions. In effect, this price variable reflects not only price incentives to the producers, but also macroeconomic (reform) policies (as represented by the RER) In particular, this link between the price variable and macroeconomic policies is important in order to pursue an investigation of the hypotheses advanced on page 11.

PAGE 99

The question regarding which alternative 83 (or substitute) crop to include in a particular crop's ECM proved to be difficult. In the absence of recorded data on this issue, the final choice of the substitute crop resulted from a consideration of a number of factors. Among these were, the nature of the crops (traditional export versus domestic crops) ; the terrain where the crops are grown (mountainous versus flat lands) ; tree-crops versus annuals; and, root-crops versus vine-crops. The difficulty of identifying an appropriate substitute crop (or crops) can be demonstrated in the case of banana. Possible substitute crops for banana are coffee and sugar, since these are all traditional export crops. Banana is grown extensively on flat lands, but is also cultivated on hilly terrains as is coffee, whereas, sugar growing has been confined to the relatively flat plains in Jamaica. Hence, both sugar and coffee are plausible substitutes for banana, and the choice of either one or both becomes an empirical issue. A further complication in the choice of a substitute crop arises. In the early 1990s, this researcher observed two cases where lands which were previously used to cultivate sugar and banana were converted into papaya groves. Telephone interviews conducted with officials of

PAGE 100

84 the Ministry of Agriculture, Government of Jamaica, indicate that these are not isolated cases, and that similar land conversions have been observed into cutflowers, decorative foliage, aloe-vera, ochro, and other non-traditional export crops. Several substitute crops were initially included in each ECM but in all cases it was found that the inclusion of only one substitute crop price improved the statistical properties of the model. The alternative crops in each crop's ECM are reported in Table 4.1. Table 4.1: Alternative (Substitute) Crops in Each EOf. Crop Alternative Crops Considered Banana Sugarw, coffee, papaya Sugar Banana w, papaya Coffee Banana, pimento, sugar, orange Pimento Banana, coffee, cocoa bean Yam Cassava, potato, sweet potato Orange Grapefruit w, tangerine, coffee Cocoa Bean Bananaw, pimento, coffee Potato Cassavaw, yam, sweet potato Crop chosen as substitute crop. Two input prices are included in each crop's ECM, namely, fertilizer price and average wage in the agriculture sector. The fertilizer price variable is a weighted price index of all types of fertilizers imported into Jamaica, with quantities as weights. With respect to the wage variable, there are no data on agriculture wages

PAGE 101

85 in Jamaica. A proxy for this variable was constructed from data on compensation to agricultural employees recorded in the national income accounts. Compensation of agricultural employees is defined to include all payments by producers of wages and salaries to their employees in kind and in cash, and of contributions paid or imputed, in respect of their employees to social security schemes and to private pensions, family allowance, casualty insurance, life insurance and similar schemes. (Jamaica, National Income and Product, 1970, p.50) Total compensation to agricultural workers is divided by total agricultural labor force to yield a proxy for the average nominal wage rate. The latter is then divided by the consumer price index and used as the proxy for average real wage rate in the agricultural sector. All variables are measured in logari thrns and indexed at 1980=100. The logarithmic transformation of the data was done because of ease of interpretation of the estimated coefficients and the frequency of its use in the econometric literature. In addition, Doan (1996) has argued that preliminary data transformation helps to straighten out trends and eliminate systematic tendency of the variance of the series to depend on the magnitude of the data. To save on degrees of freedom, the smallest lag length was sought while at the same time seeking to pass the diagnostic tests on the residuals. After some

PAGE 102

86 experimentation and testing, a lag-length of k=2 was settled upon. This is the minimal lag length required for an ECM, and is sufficiently long to accommodate the issues of expectations discussed in Chapter 3. The residual analysis for adequacy of model specification combines statistical testing and inter pretation of graphs. Tests for autocorrelation utilize the Ljung-Box (L-B), LM(l) and LM(4) tests. The (L-B) test is a test of the null hypothesis that the residuals from the first T / 4 observations are not autocorrelated. The LM ( 1) and LM(4) are tests of the null hypotheses of no firstand fourth-order autocorrelation. A multivariate test using the piocedure of Doornik and Hansen (1994) is used to test the null hypothesis that the residuals from the system of equations are normally distributed. Finally, the auto regressive conditional heterosckedasticity (ARCH) test is used to test the null hypothesis that the residuals of each equation in the system are normally distributed. All of the tests mentioned above are distributed as x 2 with the appropriate degrees of freedom and corrected for small sample bias when necessary. A sample of graphs that aid in the residual analysis is presented in Section 4. 4. These residual graphs and residual test statistics for the

PAGE 103

87 tests mentioned above, are normal output from the RATS/CATS program when estimating ECMs, using the Johansen approach. 4.3 Testing for Stationarity A preliminary step in the estimation of the ECM models is to test the data for stationarity. Three tests are used in this study, the Dickey-Fuller (D-F) and Augrnented Dickey-Fuller (ADF), the Weighted Symmetric (W-S), and the Phillips-Perron (P-P) tests. The Time Series Processor (TSP), (version 4.4), was used to conduct these three tests on all quantity and price data. First, tests were conducted on the variables in levels, then on their first difference. The results are shown in Tables 4.2 and 4.3. For the Dickey-Fuller test, the following models were used: ( 4 1) ( 4 2) where Tis time, Xt is a time series, ai are coefficients to be estimated, t is white noise, and ~Xt the first difference. The ADF tests were conducted on the following equations: m ( 4 3) M{t = ao + a1 Xt-1 + J31 L Xt-1 + t i=l m ( 4 4) M{t = ao + a1Xt-1 + a2T + J31L M{t-1 + t i=l

PAGE 104

88 Table 4.2: Unit Root Tests--Prices and Quantities in Levels. Tests for Price Tests Variable W-S 3 DF" P-Pc W-S 3 Banana -2.85 -2.49 -10.22 -1. 74 (0.13) (0.34) (0.42) (0.80) Sugar -1. 66 -1.24 -4.24 -1.39 (0.83) (0.90) (0.87) (0.92) Coffee -2.91 -2.13 -9.08 -1.08 (0.11) (0.53) (0.50) (0.97) Pimento -2.96 -2.57 -15.83 -2.21 (0.09) (0.29) (0.16) (0.49) Yam -1. 96 -1.59 -7.81 -2.06 ( 0. 67) (0. 80) ( 0. 60) ( 0. 61) Orange -1.55 -1.34 -4.95 -1.33 (0.88) (0.88) (0.82) ( 0. 93) Cocoa Bean -2.36 -2.07 -7.94 -2.89 (0.38) (0.56) (0.59) (0.12) Potato -2.36 -2.02 10.74 -3.12 (0.39) (0.59) (0.39) (0. 06) Fertilizer -1. 972 -1.142 -13.460 (0.667) (0.922) (0.141) Wage -2.092 0.743 -16.639 (0.135) (1.000) (0.120) Note: Numbers in parentheses are P-values. a Weighted Symmetric Test. b Dickey-Fuller Test. c Phillips-Perron Test. for Quantity DF" P-Pc -3.70 -5.39 (0.02) (0. 79) -1.14 -5.41 (0.92) (0. 79) -3.08 -25.21 (0.11) (0.02) -2.40 -12.44 (0.38) (0.29) -4.10 -9.46 (0. 01) (0.47) -2.43 -10.63 (0.36) (0.40) -2.53 -20.69 (0.32) (0.06) -0.58 -20.94 (0.98) (0. 06) The order of the ADF test (i.e., the value of m) was chosen on the basis of residual whiteness, since the ADF test is not valid when serial correlation among the errors exists. The test statistics from equations ( 4 .1) ( 4. 4) are not distributed as at-distribution because under Ho: non stationarity, the variance is not finite. Fuller (1976) has simulated the critical values of the distribution for equations (4.1)-(4.4). The (W-S) test is sometimes recommended over the (D-F) tests due to its higher power ( Pantula et al., 1994) The

PAGE 105

89 Table 4.3: Unit Root Tests--Prices and Quantities in First Difference. Tests for Price Tests for Quantity Variable w-sa DF/AD~ P-Pc w-sa Banana -3.80 -2.12 -18.21 -4.81 (0.01) (0.54) ( 0. 10) (0.00) Sugar -3.67 -2 .11 -17.33 -3.65 (0.01) (0.54) (0.12) (0.01) Coffee -3.92 -3.61 -29.02 -3.88 (0.01) (0.03) (0.01) (0.01) Pimento -3.25 -3.00 -44.74 -3.66 (0.04) (0.13) (0. 00) (0.01) Yam -3.39 -2.97 -24.80 -3.32 (0.03) (0.14) (0.03) (0.03) Orange -2.90 -3.56 -24.54 -4.17 (0.11) (0.03) (0.03) (0.00) Cocoa Bean -3.24 -2.94 -31.13 -3.54 (0.04) (0.15) (0.01) (0.02) Potato -3.99 -3.70 -28.38 -2.47 (0.00) (0.02) (0.01) (0.31) Fertilizer -6 .112 -5.180 -42.821 (0.063) (0.052) (0.000) Wage -6. 4 72 -6.639 -43.081 (0.024) (0.017) (0.000) Note: Numbers in parentheses are P-values. a Weighted Symmetric Test. b Dickey-Fuller/Augmented Test. c Phillips-Perron Test. OF/AD~ P-Pc -3.23 -18.32 (0.08) (0.10) -3.36 -42.73 (0.06) (0.00) -3.65 -43.70 (0.03) (0.00) -3.47 -39.44 (0.04) (0.00) -4.05 -24.77 (0.01) (0.03) -3.90 -42.52 (0. 01) (0.00) -3.29 -39.41 (0.07) (0.00) -3.12 -28.49 (0.10) (0.01) (W-S) test uses a weighted double-length regression. The first half of the regression regresses Yt on Yt-1 and lags of ~Yt, with weights (t-1)/T, where Tis the full sample size, and t the size of the sub-sample used for the first half of the regression. In the second half of the regression Yt is regressed on Yt+1 and leads of Yt-Yt-1 with weights [1(t1)] /T (Hall and Cummins, 1998). Phillips and Perron (1988) have suggested a variant to the Dickey-Fuller test by using a non-parametric General Methods of Moments (GMM) type

PAGE 106

90 method to compute a residual variance which is robust to serial correlation. The unit root tests indicate that for each variable tested, at least one test shows that the variable is non stationary in levels, but stationary in first differences. The conflicting test results reported in Table 4. 3 derive from the different power of these tests, which in turn depend upon the estimators used ( Pantula et al., 1994) This should not pose any serious estimation problem, since Hansen and Juselius (1995, p.l), argue that ... not all indi victual variables included in [a multivariate cointegration model] need be I(l), as is often incorrectly assumed. To find cointegrat i on between non-stationary variables, only two of the variables have to be I(l). Consequently, the estimation of crop supply response which follows is based on the conclusion that the variables included in the models are all integrated of order one, i.e., Y i ~I(l), which implies that, ~Yi~I(O). 4.4 Estimating Long-run Supply Response For the banana ECM, the vector of stochastic variables is: Xt = (lbanq, lbanpr, lsugpr, lwage, lferp)' where lbanq = quantity of banana; lbanpr = price of banana; sugpr = sugar price (the alternative crop to banana); lwage = average wage in the agricultural sector; lferp = average

PAGE 107

91 price of fertilizer. The 'l' that begins each variable name indicates that the variables are measured in logarithms. The diagnostic statistics for residual analysis for the unrestricted model for banana are shown in Table 4. 4 and suggest that the assumed model for the banana ECM is supported by the data. This is also supported by the graphs in Figures 4 .1-4. 5, where in particular, the standardized residuals all show a zero-reverting pattern. The next step Table 4.4: Diagnostic Statistics for Residual Tests for the Banana ECM. Tests Dfa x2 p-value For normality of residuals 10 10.322 0.21 For autocorrelation L-B ( 9) 124 102.050 0.06 LM (1) 25 23.629 0.54 LM(4) 25 20.661 0.71 For ARCH(2) Lbanq 2 1. 961 Lbanpr 2 1. 492 Lsuqpr 2 0.575 Lferp 2 0.638 Lwage 2 0.913 Notes: The critical value for X 2 (2) = 5.99 at the 95 percent significance level. a Df = degrees of freedom. estimates the number of cointegrating relationships, r, in the system. In Table 4.5 the likelihood ratio test statistics for the rank of Il (from equation 3.19) are presented along with

PAGE 108

Actual and Fitted for DLBANPR 0 6 -------~----~ 0 4 0 2 -0 0 -+-~++--+->---',.....--i-->-+-+--+-t-+-..-+--< -0 2 -0 4 -0 6 -+--rr-rrncrn"TTTTTTTTTTTTrrrrrnrrn"TTT-,,--' 1968 1974 1982 1990 Standardized Residuals 2 -1 1968 1974 1982 1990 Histogram of Standardized Residuals 0 5 ------------~ 0 4 0 3 0 2 Correlogram of INiduala 1 00 ------------~ 0 75 0 50 0 25 0 00 -0 25 -0 50 -0 75 -1 00 -+-----,-~,.......~--. -~~--. ~ 5 Lag Figure 4.1: Residuals for Banana Price in Banana ECM. Actual and Fitted for DLBANQ 0 15 ------------~ 0 10 0 05 -0 00 -t--+-,=-++-+-'-lrl---\----++-+......._---1 -0.05 -0 10 -0 15 -0 20 -0 25 -+-7T1TTTTTTTTTirrMTTTTTTrrrrrn"TTTTTTr-' 1968 1975 1984 1993 -1 -2 1968 1974 1982 1990 Histogram of Standardized Residuals 0 7 ---= =-----.-------~ ..__, 0 5 -1-----~ 0 5 0 4 0 3 0.2 0 1 0 0 -'___ .__.__.__ __ ,_,___ Correlogram of residuals 1 00 ------------~ 0 75 0 50 0.25 0 00 + --_,, ___ ,.... -0 25 -0 50 -0 75 1 00 ~--r-~~-----,--~~---.~ 5 Lag Figure 4.2: Residuals for Banana Quantity in Banana ECM. 92

PAGE 109

93 Actual and Fitted for DLSUGPR Histogram of Standardized 0 8 0.5 0 6 0.4 0 4 0 2 0 3 0 0 0 2 -0 2 0 1 -0 4 1966 1974 1962 1990 0 0 Standardized Residuals 3 1.00 Comtlogram ct 2 0 75 0 50 0 25 0 0 00 -0 25 -1 -0 50 -2 -0 75 -1.00 -3 5 1966 1974 1962 1990 Lag Figure 4.3: Residuals for Sugar Price in Banana ECM. Actual and Fitted for DLWAGE Histogram of Residuals 0 2 0 5 ....... 0 1 0 4 -0 0 -0 1 0 3 -0 2 0 2 -0 3 0 1 -0.4 1966 1974 1962 1990 0 0 Standardized Reliduala Comtlogram of 2 1 .00 0 75 0 50 0 0 25 -1 0 00 -0 25 -2 -0 50 -3 -0 75 -1 00 -4 5 Lag Figure 4.4: Residuals for Wage in Banana ECM.

PAGE 110

Crossand autocorrelograms of the residuals DI..SANQ DLBANPR DLSUGPR DLWAGE ~E:EB El B ~BBEB -EJEEElE ~E:J83 EJ Lags 1 to 9 Figure 4.5: Crossand Autocorrelograms in Banana ECM. The eigenvalues of the companion matrix 1 00 D D u u D D 0.75 0.50 co Dao 000 OD DO D 0 D D 0 D D 0 D 0 D D D D 0.25 l De:, p D D 0.00 ---~----+--------..---~~ -----I J D -0.25 ~a f D D D D -0.50 DD D D D D D D DD Op D D -0.75 D D ODO ODO 1 00 a n ,... _ ,... n 0 '-----~~-~'-=----------' -1.0 0.0 1.0 Figure 4.6: Plot of Eigenvalues for the Banana ECM. 94

PAGE 111

Table 4.5: Tests of Cointegration Rank for Banana ECM. Eigen Values Amax Trace Ho:r P-r Amax 90 Trace 90 0.649 37.66 80.99 0 5 21. 74 71.66 0.369 16.56 43.33 1 4 18.03 49.92 0.332 14.54 26.27 2 3 14.09 31. 88 0.022 8.96 12.23 3 2 10.29 17.79 0.087 3.27 3.27 4 1 7.50 7.50 95 the 90 percent quantiles of the appropriate limiting distributions. The estimated maximum eigenvalue and trace statistics are shown in the columns Amax and Trace, respectively. The corresponding critical values at the 90 percent level are in the columns Amax 90 and Trace 90. Testing is done sequentially (i.e., row by row), beginning with the first row in Table 4.5, until non-rejection of the null hypothesis is found. This sequential testing begins by testing the null hypothesis Ho:r = 0, (i.e., there is no cointegrating relationship in the system) in row one of Table 4.5. Testing is done by comparing the estimated Amax and Trace statistics with their theoretical (tabular) counterparts, Amax 90 and Trace 90, respectively. The null hypothesis Ho: r = 0 is rejected since the estimated test statistics, (Amax = 37. 66 and Trace = 80. 99), are greater than the corresponding critical values for these statistics (Amax 90 = 21. 74 and Trace 90 = 71. 66), respectively. Row

PAGE 112

96 two is then examined. In Table 4. 5, Ho: r = 1 cannot be rejected, since the estimated test statistics (Amax= 16.56 and Trace = 43. 33) are less than their critical values (Amax 90 = 18.03 and Trace 90 = 49.92), respectively. This suggests that there is one long-run or cointegrating relationship in the banana ECM. This is also supported by the graph in Figure 4. 6 which shows that the eigenvalues are all within the unit circle. The model is then re-estimated under the restriction that only one cointegrating relationship exists in the system. The resulting cointegrating relationship is: ( 4 5) -6.158lbanq 3.943lbanpr + 0.957lsugpr + 0.655lwage + 3.464lferp 23.365 = 0 After normalizing on the banana variable and transferring the banana quantity variable to the left hand side of the equal sign the estimates for the cointegrating vector, P, and the vector of weights, a, are shown in Table 4.6. Table 4 6: Estimated Long-run and Adjustment Coefficients, P's, a's--Banana EOf. lbanq lbanpr lsugpr lwage lferp constant ~s 1 0.640 -0.156 -0.567 -0 .106 -3.795 a.' s -0.141 -0.352 0. 511 0.169 0.097 (-1.909)c (-1.949)c (2.452)b (3.329)a (0.641) Note: Figures in parentheses are t-values. a b, c, indicate statistical significance at one, five and ten percent levels, respectively.

PAGE 113

97 The cointegrating parameters can be interpreted as long-run elasticities. Thus, a ten percent increase in the price of banana will lead to a 6. 40 percent increase in banana supply in the long run, other things remaining constant. Similarly, a ten percent increase in the average price of fertilizer will lead to a 1.06 percent decrease in banana supply. With respect to the substitute crop, a ten percent increase in sugar price will lead to a 1.60 percent decrease in banana quantity. Finally, a ten percent increase in agricultural wage rate will lead to a 5. 70 percent decrease in banana output. The coefficients in the a-vector are the weights with which the cointegrating vector(s) enter(s) each of the indi victual equations in the system. This can be seen from the estimated rr matrix (4.6): ( 4 6) rr = ai,, = lbanq lbanpr rr = lsugpr = lferpr lwage In terms of an coefficients can be 0 .141 0.352 0.511 [1 0.640 0.156 0.567 0.106 3 795] 0.097 0.169 0.14 1 0.091 0.022 0.080 0.015 0.536 0.352 0.225 0.055 0.198 0.037 1.334 0.511 0.327 0.079 0.287 0.054 1.938 0.097 0.062 0.015 0.055 0.010 0.369 0.169 0.108 0.026 0.095 0 018 0.642 economic interpretation, these a considered as the average speed of adjustment towards the estimated equilibrium. Thus, a small

PAGE 114

98 coefficient indicates a slow adjustment whereas a large coefficient indicates rapid adjustment. At the indicated levels of statistical significance, all variables in the banana ECM (except fertilizer price) adjust to a deviation from the long-run equilibrium relationship. In the equation that estimates the long-run supply response of banana ( lbanq in equation 4. 6) the average speed of adjustment is cx1banq = -0 .141, implying that about 15 percent of any deviation from the long-run equilibrium in quantity is corrected within the first period. When the speed of adjustment is low, this might be caused by government regulations, high cost of adjustment, lags between planting and harvesting, inadequate supportive infrastructure in the agricultural sector, and other factors that tend to slow down the process of adjustment to the equilibrium growth path. The estimation and diagnostic testing of the long-run cointegrating relationship for the other crops that are included in this study, follows the same format and procedure as for banana. First, the vector of stochastic variables for each crop's ECM was specified as follows: Sugar ECM: Xt = (lsugq, lsugpr, lbanpr, lwage, lferp)'; Coffee ECM: Xt = (lcofq, lcofpr, lbanpr, lwage, lferp)'; Pimento ECM: Xt = (lpimq, lpimpr, lbanpr, lwage, lferp)';

PAGE 115

99 Yam ECM: Xt = (lyamq, lyampr, lcaspr, lwage, lferp)'; Orange ECM: Xt = (lorq, lorgpr, lbanpr, lwage, lferp)'; Cocoa-bean ECM: Xt = (lcobq, lcobpr, lbanpr, lwage, lferp)'; Potato ECM: Xt = (lpotq, lpotpr, lcaspr, lwage, lferp)'; where lsugq = sugar quantity; lsugpr = price of sugar; lcofq = coffee quantity; lcofpr = price of coffee; lpimq = pimento quantity; lpimpr = price of pimento; lyamq = yam quantity; lyampr = price of yam; lorq = orange quantity; lorpr = price of orange; lcobq = cocoa-bean quantity; lcobpr = price of cocoa bean; lpotq = potato quantity; lpotpr = price of potato. The 'l' that begins each variable name indicates logarithms, and all other variables are as previously defined. Second, residual analysis was conducted to ensure that the models were appropriate for the data. Finally, the hypothesis of reduced rank is then tested on the matrix II= a~', which defines the cointegrating vectors and adjustment coefficients a. A normalization on the crop variable is taken, and the normalized~ and a vector(s) is (are) reported. The estimates of the~ and a vectors for each crop are presented in Table 4.7. The first column in Table 4.7 lists

PAGE 116

100 Table 4. 7: Estimated Long-run and Adjustment Coefficients, P's, 's for all Crops. Crops Q Po Pa w F C Banana P's 1 0.640 -0.156 -0.567 -0.106 -3.795 ct' s -0.141 -0.352 0. 511 0.169 0.097 (-1.909)b (-1.949)b (2.452)a (3.469)a (0.641) Sugar P's 1 0.155 -0.640 -0.106 -0.563 -3.290 ct' s -0 .141 0. 511 0.352 0.169 0.097 (-2.899)a (2.452)a (1.949)b (3.329)a (2.64l)a Coffee CE *1 p;-;1 0.969 0.291 -0.213 -0. 724 -5.348 ct' s -0.069 -0.202 -0.664 0.088 0.107 (-2.420) (-1.316) (-4.823)a (1.617) (1.326) CE #2 p;-;1 0.474 -0.005 -0.448 -2.938 6.503 ct' s 0.049 -0.091 -0.100 -0.002 -0.193 (2.717)b (-1.911)c (-1.930)c (-2.091)b (-5.697) 4 Pimento CE *1 P's 1 0.576 -0.849 -0.534 -1.526 1.683 ct' s -0.047 -0.318 0.031 0.004 0.070 (-2.763)a (-6.173)a (0.400) (2.156)b (2.455)a CE #2 -P's 1 -0.225 -0.026 0.542 0. 468 -10 579 ct' s -0.820 0.140 0.200 -0.106 -0.036 (-4.588) (0.932) (0.889) (-1.575) (-0.252) Yaa CE *1 1 -3.755 5.138 0.484 1.225 -16.819 ct' s -0.060 0.128 -0.124 -0.012 -0.098 (-2.184)b (2.813)b (-4.377)a (-0.539) (-2.125)b CE #2 -Jl's 1 0.418 -0.967 1.860 1. 941 -23.344 ct' s -0.186 -0.273 0.156 -0.045 -0.028 (-4.718)a (-4.147)a (3.808)a (-1. 730)c (-0.427) CE #3 P's 1 1.706 -2.958 -5.880 -1. 700 23.210 ct' s -0.033 -0.062 0.026 0.021 -0.042 (-2.851)a (-3.240)a (2.184)b (2.257)b (-2.189)b Orange J3' s 1 2.347 -1. 620 -1. 979 -1. 492 11.270 ct' s -0 .139 -0.366 0.021 0.062 0.174 (-1.472) (-1.858) (0.094) (2.134) (3.047) Cocoa Bean P's 1 0.359 -0.081 -1. 405 -0.983 -12.350 ct' s -0.797 0.676 -0.841 -0.284 -0.062 (-3.971) (2.103) (-3.449) (-1. 621) (-0.694)

PAGE 117

Table 4.7: Cont'd. Crops Q P o Pa w Potato CE #1 -Ws 1 1. 952 -1.030 -5.197 a's -0.053 -0.043 0.077 0.031 (-1.108) (-1.273) (4.035)a (3.052)a CE #2 1 -0.251 0.220 0.127 a's -1.008 0.071 -0.056 -0.018 (-3.569)a (2.364)a (-2.504)a (-1. 999)b Note: Figures in parentheses are t-values. CE indicates 'cointegration equation'. 101 F C 0.704 -38.815 -0.068 (-2.793)a 0.323 -5.860 -0 .117 (-2.815)a a, b, c, indicate statistical significance at one, five and ten percent levels, respectively. Q = quantity; Po= own-price; Pa= price of the alternative crop; W = average wage rate in agriculture sector; F = average price of fertilizer; C = constant (intercept). the crops, which are inc l uded in this study, followed by the~ and a vectors. When there are more than one long-run cointegrating relationships, these are numbered as "CE # (.)" with the associated and a vectors. Columns 2-6 indicate the variables and coefficients in the ECM for each crop, and column 7 records the intercepts. Thus, in the case of sugar, there is one long-run or cointegrating relationship, which, from Table 4.7, can be expressed as: Q = -3.290 + 0.957Po 0.640Pa 0.106W 0.563F where Q = quantity of sugar; P 0 = own-price (i.e., price of sugar); Pa= price of alternative crop; W = average wage in agriculture; F = average price of fertilizer. A number of inferences can be drawn from the information presented in Table 4.7. First, the long-run

PAGE 118

102 relationship between supply and own-price is inelastic for five of the eight crops. Crops that are price elastic are yam, orange and potato, with own-price elasticities of 1.7, 2.3, and 2. 0, respectively. The own-price elasticity coefficients reported in Table 4.7 are generally within the range estimated by other studies for Jamaica (see Table 3. 1) A second point to note with respect to the information in Table 4.7 is the relatively low elasticities with respect to the substitute crop price. For the banana ECM it is -0 .16, for CE #2 in the coffee ECM, -0. 01, and for sugar and cocoa-bean ECMs, -0. 64 and -0. 08 respectively. Relatively higher elasticities are recorded for the pimento -0.85, yam -3.0, orange -1.62 and potato -1.03. Third, the relationship between quantity and input prices is generally inelastic. The exception are for wages in the yam and potato ECMs, and for fertilizer price in coffee, pimento, orange, and yam. Thus, in the coin tegration equation #3 for yam, in the long-run a one percent increase in average agricultural wage will lead to a 6 percent fall in yam supply, other things remaining constant. A similar conclusion is arrived at for the first cointegrating relationship for potato. Similarly, the long run relationship between supply and fertilizer price is

PAGE 119

103 elastic for coffee (CE #2), pimento (CE #1) orange and yarn (CE #3) A fourth point to note with respect to Table 4. 7 is about the adjustment coefficients a. Generally, the adjust ment coefficients are significant. The exceptions include fertilizer price in the banana ECM, own-price, and wages in CE #1 for the coffee ECM, price of the alternative crop in CE #1 for the pimento ECM, and a few others. An insignificant adjustment coefficient, ai, is a sign of weak exogenei ty of the variable Xi in the vector of stochastic variables, Xt, in the ECM. Weak exogeneity means that although a long-run relationship exists between Xi and the other variables in Xt, Xi does not adjust to deviations from the long-run equilibrium. As such, any deviation from the long-run equilibrium in the system after a shock, is restored by adjustments made by variables other than Xi in the system (Johansen and Juselius, 1990, 1992). According to Johansen (1995), this does not preclude Xi from the cointegrating relationship in Xt, An issue of some importance is raised here regarding the results in Table 4. 7. This relates to the negative relationship observed between quantity and own-price for some of the crops, and what interpretation is possible in these cases. An examinat i on of the data on quantity and

PAGE 120

104 own-price reveals strong negative correlations for banana (-0. 737), and sugar (-0.760), and modest negative quantity-own-price correlations for coffee (-.420) and potato (-0.390). The quantity-own-price correlations for the other crops are positive. In Table 4. 7 negative relations between quantity and own-price show up for pimento (CE #2), yarn (CE #1) and potato (CE #2). Given the strong negative quantity-own price correlation coefficients for banana and sugar, it is surprising that negative relations between quantity and price do not show up in the estimated cointegrating relationships for these two crops. A number of possible explanations can be offered for these results. First, graphical examination of the quantity-own-price relationship reveals that the negative correlations are less pronounced than what the magnitude of the estimates would imply. Second, and as an extension of the previous point, the negative relationship is observed for specific and short period(s), rather than for the entire 1962-1999 period. In fact, in chapter 5 where supply functions are estimated for the eight crops over two subsample periods, 1962-1979 and 1980-1999, the negative quantity-own-price relations are more prevalent (see Tables 5.4, and 5.5, pp.157-160).

PAGE 121

105 Third, error correction models based on cointegration account for the effects of not one or two, but of all the cointegrated variables in the system. In particular, the ECM technique provides a way of separating the long-run relationship between the economic variables (Yt = ~Xt ) from the short-run responses (i.e., the llyt, AXt terms) In this regard, Engle and Granger (1987) have demonstrated that, asymptotically, these effects are dominated by the long-run relationships, and as the sample size becomes larger, the estimates of the cointegrating vector converges to the true Value at a rate equal to T 1 h T h 1 w ere 1.s t e samp e size. This rate of convergence is faster than the case where the variables are stationary, in which case the rate of convergence is T112 This is ref erred to as the superconsistency property of the estimates produced by ECMs. In effect, therefore, the estimates of the cointegrating vectors produced by ECMs based on cointegration, are not confined to, but rather, go beyond the simple correlations which may be observed between variables. Finally, cases where the cointegrating relationships reveal a negative relationship between quantity and own price, need to be commented on. It is tempting to interpret these cases as reflecting relationships on a demand function, i.e., a negative relationship between quantity

PAGE 122

106 and own-price. However, this might not be a plausible explanation, since variables which would usually be included in a demand function such as consumer income, an index for consumer tastes, and other demand forcing variables, are not included in the crops' ECM. For those cases observed in Table 4.7, the strongest statement which could be offered for the negative relationships between quantity and own-price is that there is a long-run relationship observed for some crops, with possible properties of an incompletely specified real demand relationship. The diagnostic statistics for model specification and tests for cointegration rank for each estimated ECM are shown in Appendix D. Graphs for residual analysis (not reported) similar to those for banana, were examined in conjunction with the diagnostic statistics. The residual analysis suggests that for all crops, appropriate for the data. the ECMs are 4.5 Analysis of Short-run Dynamics The previous section suggests that the series in the crop ECMs are cointegrated. Since the cointegrated variables are in equilibrium, exogenous shocks to one of the variables in time to, will result in time paths of the system which will eventually stabilize to a new equilibrium

PAGE 123

107 unless further shocks occur in the system. Lutkepohl and Reimers (1992), contend that these time paths provide insights into the short-run and long-run relationship between/among the variables in the system, especially in cases where there are more than one cointegrating relationships in an ECM. In this regard, innovation accounting (analysis of impulse response functions and variance decompositions) has become a useful tool in the analysis of cointegrated systems Lutkepohl and Reimers, 1992). (Enders, 1995, 1996; The impulse response functions, specified as VARs, generate impact or dynamic multipliers. These coefficients capture the effects of exogenous shocks of each variable on its own time path, and on those of the other variables in the system, by imposing a recursive structure on the moving average representation of the VAR model (Enders, 1995). The moving average representation facilitates analysis of the interaction between sequences in the series of a VAR. Impulse response functions can be plotted to provide a visual representation of the behavior of each series in response to shocks in the system. Identification of the imposing some necessitates restrictions) on the system. parameters in the VAR structure (via parameters One method of imposing

PAGE 124

108 restrictions uses the Choleski decomposition (Doan, 1996). In this method, the ordering of the variables in the VAR is important and is based on statistical properties of the data as well as theoretical insights (Enders, 1995). However, the method can be tedious since, for example, in a five variable system there are 5 (=120) possible orderings of the variables. Sims (1980) and Bernanke (1986) have proposed an alternative method of imposing restrictions for the identification of the VAR. In this scheme, exact identification of the structural model for an n-variable VAR necessitates (n 2 n)/2 restrictions but additional ( over identifying) restrictions can be imposed and tested statistically. The restrictions are based on economic theory. The VAR in standard form (Enders, 1996), which is used for the impulse response analysis is specified as: ( 4 7) where: Xt = (6Q, 6P 0 6Pa, 6F, 6W) ; Xti = (6Ot-i, 6Po,t-i, 6Pa,t-i, 6F,t-i, 6W,t-d', i = 1,2 ... Ai= parameters to be estimated; et= vector of error terms. All other variables as previously defined.

PAGE 125

109 Restrictions on ( 4. 7) are based on the relationships that are specified in Table 4 8 In effect, the specification suggests that Po, Pa, F and W are exogenous variables, and further, that innovations in any one of these variables will induce contemporaneous changes in itself and on Q, but on no other variable in the system. However, this specification, when combined with the Sims Bernanke error variance decomposition, does not rule out induced changes in the other variables in future periods. This is to be expected, given that the variables are cointegrated. Table 4.8: Endogenous and Exogenous Variables in Crops' Impulse Response Functions. The Contemporaneous Is Affected by the Value of: Contemporaneous value of: Q All variables in the system Po No other variable Pa No other variable F No other variable w No other variable Notes: Q = crop output; Po= (own-price), i.e., price of output; Pa= price of substitute crop; F = fertilizer price; W = average agriculture wage rate. Related to impulse response functions are forecast error variance decompositions, which provide information about the proportion of the movements in a variable due to its own shocks and to those from other variables in the

PAGE 126

110 system (Enders, 1996). A twelve-period forecasting horizon is specified, and the Sims-Bernanke forecast error variance decompositions are used based on the restrictions implied in Table 4.8. Impulses for 12 periods were examined for each crop's VAR and the first and second period results are reported in table format, while the plots for the entire forecast period are reported graphically. For all variables, the effects of the shocks after the fourth and fifth periods converge to zero or the equilibrium path, which is to be expected from a cointegrated system. The results for the banana VAR are reported in Table 4. 9. Thus, a one-standard-deviation shock in banana price (equal to 0.220 units) induces contemporaneous decrease of 0.165 units in banana quantity. By model specification, there is no contemporaneous change in the other variables. After one period, banana price is still O .112 uni ts above its mean, while banana quantity has increased by 0.076 units. In the second period, the shock in banana price has induced an increase in sugar price of 0.081 units, a decrease in fertilizer price of 0.007 units, and an increase in wage of O. 041 units. These effects are also shown graphically. As seen in Figures 4.7-4.14, equilibrium is achieved between the fourth and sixth periods.

PAGE 127

Table 4.9: Responses of Banana Quantity to Shocks in the Banana VAR. Responses to Shock in sugar price Period dlsugpr dlferp dlwage dlbanpr Dlbanq 1 0.210 0 0 0 -0.162 2 0.071 0.020 -0.051 0.171 0.092 Responses to Shock in fertilizer price Period dlsugpr dlferp dlwage dlbanpr dlbanq 1 0 0.150 0 0 -0.039 2 -0.007 0.023 -0.047 0.014 0.018 Responses to Shock in wage rate Period dlsugpr dlferp dlwage dlbanpr dlbanq 1 0 0 0.055 0 0.002 2 -0.014 0.020 0.031 -0.018 0 Responses to Shock in banana price Period dlsugpr dlferp dlwaqe dlbanpr dlbanq 1 0 0 0 0.220 -0.165 2 0.081 -0.007 0.041 0.112 0.076 Notes: The prefix 'd' to the variable names denotes change in the variable. All variables as previously defined. 111 To complement the results from the impulse response functions, the forecast error decompositions of the banana VAR are reported in Table 4.10. Enders (1996) argues that in applied research, it is typical for a variable to explain almost all of its forecast error variance at short horizons and smaller proportions at longer horizons. The Sims-Bernanke forecast error variance decompositions (based on restrictions specified in Table 4. 8) reported in Table 4 .10 show that each variable explains 100 percent of its forecast error variance in the first period. However, after the first period, the forecast error variance of a variable, which is explained by its own shock, falls. In

PAGE 128

1.0 --,-----\--,-/\-------.-----------, ,, ~ ,,N._ g 0 0 -+-+ : c:,rc......:....._ -s ~ v= ,:; P. ,:,. ,;:; :::~ ~ -""" "" ~-!.cc.:==== ~ =====I \f r -0 5 -1.0 -1 5 i 1 ... / I \ : \ I \ I '., ,' ! i -2 0 ~--,,,----r--., --,.---,----r---,--,,,----r--..--,--,---' 0 2 4 6 8 10 Figure 4. 7: Impulse Responses to Innovation in Sugar Price--Banana VAR. 1 0 --,---r-----------,----------~ \ ' 0 8 0.6 0 -4 0.2 ' ' '. ' \ \ ' ' ,,,, --.. -'::..... 0 0 -r--~::-=----~-:--: =-=--""'~=~-==-=-~-------, -0 2 --'---,,--r--..--,--,--~,----.,--r--,,----r--..-,--.---' 0 2 -4 6 8 10 an Figure 4. 9: Impulse Responses Innovation in Wage--Banana VAR. to an 1 00 -,------------,-----------, 0 7 0. 0 25 > :~ 0 00 -t------,-.:;-..;~ ---;; ~ :~~~:;;,-~ --~~------l \; ::--=---;. ;.------0.25 i;' / : \ / / \ ,; -0 50 \ ,' -0 75 \ ,, \ 1.00 ~--.---,,----r--.----r--,----,--,----,--.--., -...,..-~ 0 2 4 6 8 10 Figure 4. 8: Impulse Responses to an Innovation in Fertilizer Price--Banana VAR. 1.0 --,--~--------,------------, \ ., (-:, ,, \ ._ ........ ...... \ ... ...... 0 5 / ,,' ::--_ .. _ 0.0 +--.L_~~~:;=~;;.;.. ....... --!= ========j -/-~ -0.5 -1.0 -1.5 -2 0 --'-----,--,----,--.---,----r---,----,--,----,--.---,-~ 0 2 6 8 10 Figure 4.10: Impulse Responses to an Innovation in Banana Price--Banana VAR.

PAGE 129

3 2 cllugpr chi) 2.4 .-ge -.,, cbnq 1 6 0 8 0 0 -0 8 -1 6 0 2 4 6 6 10 Figure 4.11: Impulse Responses to an Innovation in Banana Quantity--Banana VAR. 1 00 \ ... "" \ chi) 0.75 \ .-ge \ clcofpr \ dcotq 0 50 \ I 0 25 0 00 -0 25 -0 50 0 2 4 6 8 10 Figure 4.13: Impulse Responses to an Innova t ion in Coffee Quan t i t y--Coffee VAR. 3.2 dlbanpr dlfllfl) 2.4 dlwage laugpr 1 6 dleugq 0.8 0 0 -0 8 -1 6 0 2 4 6 8 10 Figure 4.12: Impulse Responses to an Innovation in Sugar Quantity--Sugar VAR. 1 00 I \ 0 75 i I 0.50 \ 0 25 0 00 -0 25 -0 50 0 Figure 4.14: Innova t ion in VAR. 2 4 Impulse Pimen t o _,, .-ge dpfl1II" dlplmq 6 6 10 Responses to an Quan ti ty -Pimen t o

PAGE 130

114 the case of the banana VAR, in the third period, sugar price explains 75.7 percent of its own forecast error variance, fertilizer price 71.2 percent, wage 33.0 percent, banana price 42.9 percent, and banana quantity 57.3 percent. These percentage distribution of the forecast error variance among the variables remain constant after the second and third forecast periods. Table 4.10: Variance Decomposition Percentage of One-period and Three-period Forecast Error Variance--Banana VAR. Forecast error Percentage forecast error variance in: explained by shocks in: Sugar price dlsugpr dlferpr dlwage dlbanpr dlbanq 1 Yr 100 0 0 0 0 3 yrs 75.712 0.105 0.613 10.395 13.176 Fertilizer price 1 Yr 0 100 0 0 0 3 yrs 11.999 71.220 1.999 1.119 13.663 Wage 1 Yr 0 0 100 0 0 3 yrs 24.430 24.341 32.961 18.196 0.072 Banana price 1 Yr 0 0 0 100 0 3 yrs 21.262 0.149 0.340 42.880 35.368 Banana quantity 1 Yr 19.407 1.105 0.004 19.940 59.543 3 yrs 21. 416 1.147 0.006 20.094 57.337 The cross-variable effects are zero in the first period (by model specification), but change with the time horizon. In Table 4 .10, the cross-variable effects in the third period show that the shock in banana price can

PAGE 131

115 explain 43 percent of its own forecast error variance, shocks in banana quantity and sugar price account for 35 percent and 21 percent, respectively. Shocks in fertilizer price and the wage rate account for only 0 .14 9 and 0. 340 percent, respectively, of the forecast error variance in banana price. Similarly, in period three, the cross-price effect of a shock in fertilizer price can explain 71 percent of its own forecast error variance, while 14 percent and 12 percent are accounted for by shocks in banana quantity and sugar price, respectively. Finally, a shock in wage rate explains 33 percent of its own forecast error variance, while shocks in banana, fertilizer and sugar prices account for 18.2, 24.3 and 24.4 percent, respectively. The last two rows in Table 4 .10 show the forecast error variance decompositions when the exogenous variables (Po, Pa, W, F) are modeled to impact banana quantity in all time periods. Thus, the banana quantity variable explains 60 percent of its own 1-step ahead forecast error variance, and 57 percent in the 3-step ahead forecast error variance. Similarly, in the first period, banana and sugar prices account for 20 percent and 19 percent of the forecast error variance in banana quantity, respectively. These percent ages increase slightly in the 3-year ahead forecast.

PAGE 132

Fertilizer price and wages account for 116 negligible percentage variations in banana forecast error variance in both of the periods, suggesting thereby, the relative importance of own-price and price of the substitute crop on the short-run variations of banana quantity. The estimates from the impulse response functions and forecast error decompositions for the other crops are reported in Tables 4 .13-4. 2 6. The associated plots of the impulse response functions are shown in Figures 4 .15-4. 4 6. In order to present the information in a manner that is succinct and easy to read, a major portion of the estimates that are reported in Tables 4 .13-4. 26 are summarized in Tables 4 .11 and 4 .12. Table 4 .11 shows the responses of quantity to shocks in the exogenous variables in the system in periods one and two. Table 4.12 reproduces the last two rows of the tables that record the forecast error variance decompositions for the crops' ECM. That table shows the percentage forecast error variance in quantity, which is explained by the variables in the system. The response of quantity to positive shocks in own price (Po} is expected to be positive. However, Table 4.11 shows that in period one, five of the eight crops (banana, sugar, yam, orange, and potato} show a decrease in output, which is induced by an own-price shock. Thus, in period one

PAGE 133

117 Table 4 .11: Responses of Quantity to Shocks in Exogenous Variables. Responses of Quantity to Shocks in: Variable Period Po Pa w F Banana 1 -0.165 -0.162 0.002 -0.039 Quantity 2 0.076 0.092 0 0.018 Sugar 1 -0.162 -0.165 0.002 -0.039 Quantity 2 0.092 0.076 0 0.018 Coffee 1 0.021 -0.006 -0.016 0.004 Quantity 2 -0.005 0.021 0.009 0.004 Pimento 1 0.022 -0.006 -0.023 -0.021 Quantity 2 0.033 0.078 0.019 -0.090 Yam 1 -0.087 -0.008 -0.046 -0.019 Quantity 2 0.100 0.018 0.014 0.026 Orange 1 -0.038 0.061 0.097 0.025 Quantity 2 0.210 -0.148 -0.026 0.023 Cocoa Bean 1 0.007 0.028 0.031 0 Quantity 2 -0.062 -0.005 -0.046 0.006 Potato 1 -0.194 -0.026 -0.050 0.049 Quantity 2 0.177 -0.028 0.021 -0.012 Source: Tables 4.13, 4.15, 4.17, 4.19, 4.21, 4.23 and 4.25. a one-standard-deviation-shock in banana price (equal to 0.220 units--see Table 4.13), induces a decrease in banana supply of 0.165 units. Other crops whose supply decrease in period one consequent upon a one-standard-deviation-shock in own-price are sugar (-0. 162 uni ts) and potato (-0 .194 uni ts) However, with the exception of coffee and cocoa bean, all crops increase supply in the second period following a shock in own-price. The response of quantity to positive shocks in the substitute crop price, Pa, wage, W, and fertilizer price, F, is expected to be negative. In period one this expectation is confirmed for shocks in Pa on all crop supply except

PAGE 134

118 orange and cocoa-bean. Thus, in the case of coffee supply response, a one-standard-deviation-shock in banana price (equal to 0.223 units--see Table 4 .15) induces a contemporaneous decrease in coffee supply of 0. 006 uni ts. Interestingly, in the second period, with the exception of orange, cocoa-bean and potato supply, shocks in the substitute crops' price induce positive response in crop supply response for the other crops. Wage rate shocks appear to have negligible contemporaneous effect and no effect in period two on banana and sugar responses. Wage shocks do have negative contemporaneous effect on coffee, pimento, yam, and potato supply responses. However, in period two, it appears that wage shocks generally have a positive impact on short-run supply response. The exceptions are orange and cocoa-bean. Finally, shocks in fertilizer price have negative contemporaneous effect on banana, sugar, pimento and yam supply response. However, with the exception of potato, and pimento, supply response of crops to a shock in fertilizer price are positive in period two. Table 4.12 shows the proportion of forecast error variances in quantity, which are attributable to the variables that are specified as exogenous in the VARs for each crop. As previously mentioned, it is common for a

PAGE 135

119 variable to explain forecast a larger proportion of its contemporaneous error variance and smaller proportions at later periods. This is generally the case for the crops shown in Table 4 .12. Thus, 92 percent of coffee's contemporaneous forecast error variance is explained by its own shock, five percent by own-price (Po), three percent by wage shocks and by negligible percentages in fertilizer and the substitute crop prices. However, in the third period, the percentage forecast error variance Table 4 .12 : Summary of Forecast Error Variance in Crop Quantities Explained by Variables in the VAR (Percentage). Forecast Percentage of Forecast Error Variance Error in Quantity Explained by Shocks in: Variance in: Period Pa F w Po Q Banana 1 19.407 1.105 0.004 19.940 59.543 3 21. 416 1.147 0.006 20.094 57.337 Sugar 1 19.940 1.105 0.004 19.407 59.543 3 20.094 1.147 0.006 21. 416 57.337 Coffee 1 0.394 0.150 2.727 4.564 92.165 3 4.732 0.805 2.857 3.950 87.655 Pimento 1 0.070 0.839 0.948 0.917 97.226 3 7.285 10.403 2.914 1. 959 77.439 Yarn 1 0.127 0.653 3.691 13.330 82.199 3 1. 869 2.100 3.244 25.605 67.183 Orange 1 6.461 1.110 16.195 2.465 73.769 3 21. 606 1.548 5.842 43.470 27.534 Cocoa Bean 1 2.026 0 2.451 0.141 95.383 3 1. 588 0.133 6.502 11.047 80.730 Potato 1 0.714 2. 626 2.711 40.583 53.367 3 1. 485 2.440 2.459 54.282 39.334 Source: Tables 4.14, 4.16, 4.18, 4.20, 4.22, 4.24 and4.26.

PAGE 136

120 explained by coffee quantity shock reduces to 88 percent, while the proportion explained by the substitute crop price shock increases to five percent. Enders (1995, 1996) argues that if X1 shocks do not explain any of the forecast error variance in X2 in a VAR at all time horizons, then the X 2 sequence is exogenous. This means that the X2 sequence evolves independently of shocks in X1. On the other hand, if X1 shocks explain all the forecast error variance in X 2 sequence at all forecast horizons, then X2 is said to be endogenous. Neither of these two extreme cases are observed in Table 4.12. With respect to the issues of endogeniety / exogeniety and the results reported in Table 4 .12, one point must be noted. With only a few exceptions, shocks in fertilizer price and wages appear to explain relatively small proportions of the forecast error variance in quantity. The exceptions are pimento, orange and to some extent cocoa bean. This is also true for the price of the substitute crop in the cases of coffee, cocoa-bean and potato.

PAGE 137

Table 4.13: Responses of Sugar Quantity to Shocks in the Sugar VAR. Responses to Shock in banana price Period dlbanpr dlferp dlwage dlsugpr dlsugq 1 0.220 0 0 0 -0.165 2 0.112 -0.007 0.041 0.081 0.076 Responses to Shock in fertilizer price Period dlbanpr dlferp dlwage dlsugpr dlsugq 1 0 0.150 0 0 -0.039 2 0.014 0.023 -0.047 -0.007 0.018 Responses to Shock in wage Period dlbanpr dlferp dlwage dlsugpr dlsugq 1 0 0 0.055 0 0.002 2 -0.018 0.020 0.031 -0.014 0 Responses to Shock in sugar price Period dlbanpr dlferp dlwage dlsugpr dlsugq 1 0 0 0 0.210 -0.162 2 0.171 0.020 -0.051 0.071 0.092 Notes: The prefix 'd' to the variable names denotes change in the variable. All variables previously defined. 121 Table 4.14: Variance Decomposition Percentage of One-period and Three-period Forecast Error Variance--Sugar VAR. Forecast error Percentage forecast error variance in: explained by shocks in: Banana price dlbanpr dlferpr dlwage dlsugpr dlsugq 1 Yr 100 0 0 0 0 3 yrs 42.880 0.149 0.340 21.262 35.368 Fertilizer price 1 Yr 0 100 0 0 0 3 yrs 1.119 71.220 1. 999 11.999 13.663 Wage 1 Yr 0 0 100 0 0 3 yrs 18.196 24.341 32.961 24.430 0.072 Sugar price 1 Yr 0 0 0 100 0 3 yrs 10.395 0.105 0.613 75.712 13.176 Sugar quantity 1 Yr 19.940 1.105 o 004 19.407 59.543 3 yrs 20.094 1.147 0.006 21.416 57.337

PAGE 138

Table 4.15: Responses of Coffee Quantity to Shocks in the Coffee VAR. Responses to Shock in banana price Period dlbanpr dlferp dlwage dlcofpr dlcofq 1 0.223 0 0 0 -0.006 2 0.004 -0.034 0.023 0.003 0.021 Responses to Shock in fertilizer price Period dlbanpr dlferp dlwage dlcofpr dlcofq 1 0 0.150 0 0 0.004 2 -0.006 0.014 -0.058 -0.082 0.007 Responses to Shock in wage Period dlbanpr dlferp dlwage dlcofpr dlcofq 1 0 0 0.073 0 -0.016 2 -0.014 0.032 0.034 -0.005 0.009 Responses to Shock in coffee price Period dlbanpr dlferp dlwage dlcofpr dlcofq 1 0 0 0 0.178 0.021 2 0.009 -0.015 0.014 -0.034 -0.005 Notes: The prefix 'd' to the variable names denotes change in the variable. All variables previously defined. 122 Table 4.16: Variance Decomposition Percentage of One-period and Three-period Forecast Error Variance--Coffee VAR. Forecast error Percentage forecast error variance in: explained by shocks in: Banana price dlbanpr dlferpr dlwage dlcofpr dlcofq 1 Yr 100 0 0 0 0 3 yrs 87.482 0.080 1.039 0.180 11.219 Fertilizer price 1 Yr 0 100 0 0 0 3 yrs 4.411 87.721 5.209 1.034 1.625 Wage 1 Yr 0 0 100 0 0 3 yrs 8.878 38.064 50.623 2.393 0.041 Coffee price 1 Yr 0 0 0 100 0 3 yrs 1. 898 16.105 0.419 76.390 5 .188Coffee quantity 1 Yr 0.394 0.150 2.727 4.564 92.165 3 yrs 4.732 0.805 2.857 3.950 87.655

PAGE 139

Table 4.17: Responses of Pimento Quantity to Shocks in the Pimento VAR. Responses to Shock in banana price Period dlbanpr dlferp dlwage dlpimpr dlpimq 1 0.234 0 0 0 -0.006 2 0.001 -0.037 0.025 0.019 0.078 Responses to Shock in fertilize price Period dlbanpr dlferp dlwage dlpimpr dlpimq 1 0 0.150 0 0 -0.021 2 -0.006 0.013 -0.056 -0.048 -0.090 Responses to Shock in wage Period dlbanpr dlferp dlwage dlpimpr dlpimq 1 0 0 0.071 0 -0.023 2 -0.008 0.030 0.035 -0.063 0.019 Responses to Shock in pimento price Period dlbanpr dlferp dlwage dlpimpr dlpimq 1 0 0 0 0.199 0.022 2 0.004 -0.010 0.015 -0.017 0.033 Notes: The prefix 'd' to the variable names denotes change in the variable. All variables previously defined. 123 Table 4.18: Variance Decomposition Percentage of One-period and Three-period Forecast Error Variance--Pimento VAR. Forecast error variance in: Percentage forecast error explained by shocks in: Banana price dlbanpr dlferpr dlwage dlpimpr dlpimq 1 Yr 100 0 0 0 0 3 yrs 97.005 0.622 0.354 0.108 1. 911 Fertilizer price 1 Yr 0 100 0 0 0 3 yrs 5.401 87.773 4.943 0.456 1.427 Waqe 1 Yr 0 0 100 0 0 3 yrs 11.946 35.878 48.819 2.842 0.516 Pimento price 1 Yr 0 0 0 100 0 3 yrs 0.913 8.551 10.265 79.974 0.298 Pimento quantity 1 Yr 0.070 0.839 0.948 0.917 97.226 3 yrs 7.285 10.403 2.914 1.959 77.439

PAGE 140

Table 4.19: Responses of Yam Quantity to Shocks in the Yam VAR. Responses to Shock in cassava price Period dlcaspr dlferp dlwage dlyampr dlyamq 1 0.119 0 0 0 -0.008 2 0.046 0.044 -0.033 0.051 0.018 Responses to Shock in fertilizer price Period dlcaspr dlferp dlwage dlyampr dlyamq 1 0 0.147 0 0 -0.019 2 -0.037 0.002 -0.047 -0.018 0.026 Responses to Shock in wage Period dlcaspr dlferp dlwage dlyampr dlyamq 1 0 0 0.064 0 -0.046 2 0.007 0.022 0.035 0.007 0.014 Responses to Shock in yam price Period dlcaspr dlferp dlwage dlyampr dlyamq 1 0 0 0 0.179 -0.087 2 -0.013 -0.031 0.038 -0.023 0.100 Notes: The prefix 'd' to the variable names denotes change in the variable. All variables previously defined. 1 24 Table 4.20: Variance Decomposition Percentage of One-period and Three-period Forecast Error Variance--Yam. VAR. Forecast error Percentage forecast error variance in: explained by shocks in: Cassava price dlcaspr dlferpr dlwage dlyampr dlyamq 1 Yr 100 0 0 0 0 3 yrs 84.304 8.617 0.466 1.210 5.404 Fertilizer price 1 Yr 0 100 0 0 0 3 yrs 7.353 83.702 2.931 4.949 1.064 Wage 1 Yr 0 0 100 0 0 3 yrs 17.798 20.621 43.241 17.782 0.558 Yam price 1 Yr 0 0 0 100 0 3 yrs 6.503 1.294 0.321 83.129 8.752 Yam quantity 1 Yr 0.127 0.653 3.691 13.330 82.199 3 yrs 1. 869 2.100 3.244 25.605 6 7 .183

PAGE 141

Table 4.21: Responses of Orange Quantity to Shocks in the Orange VAR. Responses to Shock in grapefruit price Period dlgrpr dlferp dlwage dlorpr dlorq 1 0.523 0 0 0 0.061 2 -0.103 0.027 -0.038 0.081 -0.148 Responses to Shock in fertilizer price Period dlgrpr dlferp dlwage dlorpr dlorq 1 0 0.148 0 0 0.025 2 -0.088 0.012 -0.045 -0.122 0.023 Responses to Shock in wage Period dlgrpr dlferp dlwage dlorpr dlorq 1 0 0 0.072 0 0.097 2 -0.052 0.042 0.039 -0.107 -0.026 Responses to Shock in orange price Period dlqrpr dlferp dlwage dlorpr dlorq 1 0 0 0 0.477 -0.038 2 0.124 0.023 0.042 -0.201 0.210 Notes: The prefix 'd' to the variable names denotes change in the variable. All variables previously defined. 125 Table 4.22: Variance Decomposition Percentage of One-period and 'l'hree-period Forecast Error Variance--Orange VAR. Forecast error Percentage forecast error variance in: explained by shocks in: Grapefruit price dlgrpr dlferpr dlwage dlorpr dlorq 1 Yr 100 0 0 0 0 3 yrs 84.559 2.263 1. 511 10.961 0.705 Fertilizer price 1 Yr 0 100 0 0 0 3 yrs 4.577 83.741 7.159 4.251 0.272 Wage 1 Yr 0 0 100 0 0 3 yrs 12.886 22.566 50.088 13.979 0.480 Orange price 1 Yr 0 0 0 100 0 3 yrs 2.306 5.298 3.512 83.017 5.867 Orange quantity 1 Yr 6.461 1.110 16.195 2.465 73.769 3 yrs 21. 606 1. 548 5.842 43.470 27.534

PAGE 142

Table 4.23: Responses of Cocoa-bean Quantity to Shocks in the Cocoa-bean VAR. Responses to Shock in banana price Period dlbanpr dlferp dlwage dlcobpr dlcobq 1 0.225 0 0 0 0.028 2 0.013 -0.048 0.022 -0.030 -0.005 Responses to Shock in fertilizer price Period dlbanpr dlferp dlwage dlcobpr dlcobq 1 0 0.146 0 0 0 2 -0.015 0.025 -0.056 0.093 0.006 Responses to Shock in wage Period dlbanpr dlferp dlwage dlcobpr dlcobq 1 0 0 0.072 0 0.031 2 -0.011 0.027 0.037 0.073 -0.046 Responses to Shock in cocoa-bean price Period dlbanpr dlferp dlwage dlcobpr dlcobq 1 0 0 0 0.275 0.007 2 0.032 -0.031 -0.004 -0.083 -0.062 Notes: The prefix 'd' to the variable names denotes change in the variable. All variables previously defined. 126 Table 4.24: Variance Decomposition Percentage of One-period and Three-period Forecast Error Variance--Cocoa-bean VAR. Forecast error Percentage forecast error variance in: explained by shocks in: Banana price dlbanpr dlferpr dlwage dlcobpr dlcobq 1 Yr 100 0 0 0 0 3 yrs 90.042 1.424 0.760 2.612 5.163 Fertilizer price 1 Yr 0 100 0 0 0 3 yrs 8.454 81.851 3.994 3.370 2.331 Wage 1 Yr 0 0 100 0 0 3 yrs 11.070 35.763 48.791 2.466 1.910 Cocoa bean price 1 Yr 0 0 0 100 0 3 yrs 0.857 11. 874 5.585 77.946 3.738 Cocoa-bean quantity 1 Yr 2.026 0 2.451 0.141 95.383 3 yrs 1. 588 0.133 6.502 11.047 80.730

PAGE 143

Table 4.25: Responses of Potato Quantity to Shocks in the Potato VAR. Responses to Shock in cassava price Period dlcaspr dlferp dlwage dlpotpr dlpotq 1 0.125 0 0 0 -0.026 2 0.045 0.049 -0.036 0.050 -0.028 Responses to Shock in fertilizer price Period dlcaspr dlferp dlwage dlpotpr dlpotq 1 0 0.148 0 0 0.049 2 -0.037 0.001 -0.046 -0.020 -0.012 Responses to Shock in wage Period dlcaspr dlferp dlwage dlpotpr dlpotq 1 0 0 0.062 0 -0.050 2 0.003 0.025 0.030 0.008 0.021 Responses to Shock in potato price Period dlcaspr dlferp dlwage dlpotpr dlpotq 1 0 0 0 0.187 -0.194 2 -0.017 -0.030 0.039 -0.029 0.177 Notes: The prefix 'd' to the variable names denotes change in the variable. All variables previously defined. 1 27 Table 4.26: Variance Decomposition Percentage of One-period and Three-period Forecast Error Variance--Potato VAR. Forecast error Percentage forecast error variance in: explained by shocks in: Cassava price dlcaspr dlferpr dlwage dlpotpr dlpotq 1 Yr 100 0 0 0 0 3 yrs 90.133 7.985 0.168 1.591 0.123 Fertilizer price 1 Yr 0 100 0 0 0 3 yrs 8.889 84.140 2.715 3.706 0.550 Wage 1 Yr 0 0 100 0 0 3 yrs 19.565 19.787 36.307 22.798 1.543 Potato price 1 Yr 0 0 0 100 0 3 yrs 6.058 1.034 0.269 89.234 3.405 Potato quantity 1 Yr 0.714 2.626 2.711 40.583 53.367 3 yrs 1. 485 2.440 2.459 54.282 39.334

PAGE 144

1.0 -,---r--------,----------, 0 5 --dlbanpr dlferp dlwage l1ugpr -0 0 -r-~~:-::-"5::-~ ~ ..,,. --::,,, --.;. ;..__ _____ dl ugq ___ --l -0 5 -1 0 -1 5 2 0 ~----r--,---.---,-----.---,-----,---,--~-.---,-----.---' 0 2 6 8 10 Figure 4.15: Impulse Responses to Innovation in Banana Price--Sugar VAR. 1 0 dlbanpr I ', dlferp 0 8 I I dlwage '. leugpr I I dl1ugq 0 6 I I ', '. I 0 4 ' ' 0 2 ', ' ,, ,----.. ~0 0 -' :::::... -_.... ----..... -0 2 I I I I I I 0 2 .. 6 8 10 an Figure 4.17: Impulse Responses Innovation in Wage--Sugar VAR. to an 1.00 -,-----------,----------~ 0.75 0 50 0 25 ;,\ o oo -+-......-= ;::::;-----.,.t:. ::;; s --= .:;::::-~:..:... ...., __ __ ____ -J ; :-; -;., _.,... -0 25 -0 50 -0 75 \ 'I i;' / \ I '. I I '. \ I _, I I I I I I I I -1.00 --'---~ ---,--~-~ -.--I I 0 2 .. I I I 6 8 10 Figure 4.16: Impulse Responses to an Innovation in Fertilizer Price--Sugar VAR. 1.0 0 5 0 0 I I I ) -0.5 1 ,., '. -1 0 ! i -1 5 i -2.0 0 2 .. 6 dlbanpr dlfllfJ) dlwage l1ugpr dlaugq 8 10 Figure 4.18: Impulse Responses to Innova t ion in Sugar Price -Suga r VAR. an t-' l'v CX)

PAGE 145

1 00 -,----r----------,-----------, 0 75 0 50 0 25 0 2 6 8 10 Figure 4.19: Impulse Responses to an Innova t ion in Banana Price--Cof f ee VAR. 1 0 -,----,---------,----------~ 0 8 0 6 0 4 0 2 I I \ \ I \, \ I I \. .......... \ dlbanpr dlle,p dtwage dk:olpr dk:olq ...... ,.,.. ~.. \~ ........... 0 0 -+---....-f----'>'"7'r-,.....,.. -'-~-=---~-~""' ,.. -.::-==~ ----, ------0 2 ~--.---,---,---,---,---,---,---,---,---,---,---,-~ 0 2 4 6 8 10 Figure 4.21: Impulse Responses to an I nnova t ion i n Wage C o ff ee V AR. 1 0 0 8 0 6 0 4 0 2 -0 0 -,. -0 2 ,, .. \ '. -0.4 \ \ / -0 6 \,' -0 8 0 2 6 8 dlbenpr dlfe,p dtwage dk:olpr dk:olq 10 Figure 4.20: Impulse Responses to an Innovation in Fertilizer Price--Coffee VAR. 1 0 -,---,----------.-----------, 0 8 0.6 0 4 0 2 I '. i ..... -. > I~' .. dlbanpr dlle,p dlwage dk:olpr dk:olq >;t---.... -:..-:.,--":.. ....., ___ 0 0 -+-~_:::J_-:-~-~;, '? e:,:=:;;, '3"' ,.,;; ~~,,,..,.. -------j "' I I -0 2 0 2 4 6 8 10 Figure 4.22: Impulse Responses to an I nno v a t ion in Coffee P r iceCof f ee VAR. I-' N \.0

PAGE 146

1 00 --.---r--------~--------~ 0 75 0 50 0 25 0 2 4 6 8 10 Figure 4.23: Impulse Responses to an Innovation in Banana Price--Pimento VAR. 1.0 --.---.----------r----------, 0.8 0.6 0.4 0 2 ,"' ... _,,. ''. 0.0 -t---1 / ~ ., ~:::.;: __ ::::;;,,,. _~ 7~-""--~--~-~-::':; ,?. __ !',' _'_ ;;,...;;:~ ------i '/ '.------0 2 / \' --0 4 ~---.---,---.---,---.----,---.---,---.---,---.---,-~ 0 2 4 6 8 10 Figure 4.25: Impulse Responses to an Innovation in Wage--Pimento VAR. 1 0 --.----------~--------~ 0.8 0 6 0.4 0 2 -~ ..,,,;:: -_-:--~. --0 0 --+ -,\""", /+_-_ ----==, .. == --~ ;"'-_ ~~~~-......,=-------i --0.2 / / ', I ,' -0 . \ ,.. ,' \ / \ .,/ --0 6 \ ,' --0 8 ~---.---,'-,, --,---.---,---.---,,----.--,----,,----,-~ 0 2 6 8 10 Figure 4.24: Impulse Responses to an Fertilizer Price--Pimento Innovation in VAR. 1.0 --.-~:-------~---cbnpr ---~ I cS'9IJ) 0.8 0 6 0.4 0 2 I .-ge I clplffll' I I I I ~ -, ', _/ _i_ ,. --'-' ... ..::.--0 0 --+--<-:: ~_ ,:1._,_-::.:<.: __ :=_===-_--~ri~ ~ :!'7!1 --------j --0 2 ......l...-~1----,1-----.,---,---.--,---.--,-......--~ 0 2 4 6 8 10 Figure 4.26: Impulse Responses to an Innovation in Pimento P ri ce--Pimento VAR. I-' w 0

PAGE 147

1.0 -.---.---------,----------~ 0.8 0.6 0 -4 0 2 _,,, chi) .-ge dlyamq -0 0 +---b,--~---'k_~ --,,,,.,::::;;::.-::;1_.,.-----------l -0.2 -0.-4 _1 / "'----:;;..... ,,,,,,' -0.6 ~----r-----.----r-----.-~----.-~----.----r-----.----r-----.--' 0 2 6 8 10 Figure 4.27: Impulse Responses to an Innovation in Cassava Price--Yam VAR. 1 0 \ 0.8 \ \ 0.6 \ ' I I 0.-4 ' ' 0.2 \ , ... ...... ', 0.0 ~ -="' I ....... --~ I -0 2 ,' I I I 0 2 -4 6 8 Figure 4.29: Impulse Responses Innovation in Wage -Yam VAR. to an 1.00 dcaapr 0.75 chi) .-ge ... 0 50 dyamq 0 25 0.00 -0 25 -0 50 -0.75 0 2 6 8 10 Figure 4.28: Impulse Responses to an Innovation in Fertilizer Price--Yam VAR. 1.0 ~----r----------r-----------, 0 8 0 6 \ / ', 0 -4 0.2 ' './ 1\ \ i. i \ ', / y \ \\ / !\ \ ,.: ---' :::;:,,.,.. 0.0 -+--~~f, ~ "':--7 :::;:, ~ ~~ -------j ; .. .._ .. -0 2 I i -0.-4 -'---~,-~----. -~----.,-~----.,-~--,,~~ .--,-r-~ 0 2 -4 6 8 10 Figure 4.30: Impulse Responses Innovation in Yam Pr i ce -Yam VAR. to an

PAGE 148

1.0 \ .-.p 0 8 \ ..._ \ 0.6 clyWnq \ 0.4 \ I 0 2 \ i 0 0 .... -\ ........... -0.2 -0 -1 0 2 .. 6 8 10 Figure 4.31: Impulse Responses to an Innovation in Yam Quantity--Yam VAR. 1.0 -,----,---------~-------~ 0 8 0 6 0.-1 0 Figure 4.33: Innovation in bean VAR. 2 .. 6 8 10 Impulse Responses to an Cocoa bean Quan t ity -Cocoa 1.0 dlD'P' .-.p 0 8 ..._ clo
PAGE 149

1.00 <111,J)r 0 75 cl0fpr 0 50 clorq 0.25 0.00 ---0.25 -0 50 -0 75 0 2 6 8 10 Figure 4.35: Impulse Responses to an Innovation in Grapefruit Price--Orange VAR. 1.00 ~-~--------~--------~ 0 75 0 50 0 25 ' ' \, \ ' \ I \ \ \ . ..... :\ \ ' : ', ... .... \ \. .. .... : , 0.00 -+-...,,~-'1-------i:;,...~ ........ "" ~-------, \ I -0 25 ~-,, --.------, ,..--.------,..---.---,..----,-----..--,--,-----..--,--,----~ 0 2 4 6 8 10 Figure 4.37: Impulse Responses Innovation in Wage -Orange VAR. to an 1 00 <111,J)r 0 75 cl0fpr 0 50 clorq 0 25 0 00 ,,__ -0 25 -0.50 -0 75 0 2 6 8 10 Figure 4.36: Impulse Responses to an Innovation in Fertilizer Price--Orange VAR. 1.0 d91lr 0 8 chi> 0 6 ctorpr clorq 0 4 '\ I 0 2 -0.0 ,,._ -0 2 -0 4 -0.6 -0 8 0 2 6 8 10 Figure 4.38: Impulse Responses to an Innovation in O r ange Price--Orange VA R I-' w w

PAGE 150

1 0 -,----,---------~-----------, 0 8 0 6 0.-4 0.2 0 2 6 8 10 Figure 4.39: Impulse Responses to an Innovation in Banana Price--Cocoa-bean VAR. 1.00 -,-----,----------,------------, 0 75 0.50 0.25 ' ' '. ' \ '. ' \ \\ \ . \ \ ~---.. \ ... \ ): : '':.~~--:: .. 0 00 -+-----'-<------'T::--=:-~ .__,,, =:'!!_~-~~=-------i I / \ /. -0.25 ~---.---,--,----,---,,.-----,--,----,--.----,--r-,--r--~ b 2 e a 10 Figure 4.41: Impulse Responses to Innovation in WageCocoa-bean VAR. an 1 0 0.8 0.6 0.-4 \., 0 2 1; I .... .,.._ -0.0 ' -0 2 '. -0 '4 \ -0 6 \ \ ,,, -0.8 \/ 0 2 6 8 10 Figure 4.40: Impulse Responses to an Innovation in Fertilizer Price--Cocoa-bean VAR. 1 0 0 8 I I 0.6 0 -4 I I 0.2 I 0 0 -0 2 -0 '4 0 2 6 8 10 Figure 4.42: Impulse Responses to an Innovation in Cocoa-bean Price--Co c oa-bean VAR.

PAGE 151

1.0 -,---,---------~--------~ 0 8 0 6 0 4 0.2 dcaapr ..._ .,_,, ::: -+---1.:._L\1-~ _,, ---..<'.1=,,,;,_""' /: _. _:~ ~ ; 7: ::: ,:;; .,;.=!,= =-----------l -0 4 ' \_ _,, -0.6 ~--.---,--,---,--,---,--,---,--,---,--.---,-~ 0 2 4 6 8 10 Figure 4.43: Impulse Responses to an Innovation in Cassava Pr i ce--Potato VAR. 1 0 dlcaapr ' 0 8 ..._ I .,_,, \ 0 6 \ '. ' I I 0 4 ' ' 0 2 ' ' 0 0 ,, .. ~--:..::~~--~ I I -0 2 I I I I I I 0 2 4 6 8 10 Figure 4.45: Impulse Responses t o an I nnovation in Wage -Pota t o VAR. 1 0 dlcaapr 0 8 ..._ 0 6 ._,, 0.4 0 2 -0 0 -0 2 -0 4 -0 6 -0 8 0 2 4 6 8 10 Figure 4.44: Impulse Responses to an Innovation in Fertilizer Price--Potato VAR. 1 00 0 75 I I \ {--\. 0 50 li '. \ \ . . J \ 0 25 ,.' \ ,' '. \ . / !\ \, .. --~--, .. __ 0 00 / f ',. ,,:. -0 25 ! -0 50 i i -0.75 I I I I I 0 2 4 6 8 Figure 4.46: Impulse Responses to an I nno v a t ion i n Pota t o P r i ce -Po t a to VAR

PAGE 152

136 4.6 Summary This chapter provides empirical evidence of long-run equilibrium or cointegrating relationships between crop output and price incentives for eight crops grown in Jamaica. It was found that the data series relevant to the estimation of the long-run relationships are nonstationary. This statistical quality of the data has not been taken into consideration in previous studies on crop supply responses in Jamaica's agricultural sector. To deal with this data problem, an error correction modeling approach based on cointegration theory was adopted, within an estimation framework developed by Johansen (1988). The results from the estimation found low own-price elasticities as well as low adjustment coefficients. Additionally, impulse functions and forecast error variances were estimated in order to analyze the short-run dynamics of the output-price incentive relationships. It was found that overall, it appears that short-run crop responses are low, and respond more to crop price variations than on input-price changes. Slow adjustments in the short-run may be a sign of constraining government regulations, inadequate supportive infrastructure, lags between planting and harvesting, and other reasons.

PAGE 153

CHAPTER 5 SUPPLY RESPONSE AND COUNTERFACTUAL ANALYSIS In this chapter the before-after approach and counter factual analysis are used to assess the impact of the economic reforms on crop supply response in Jamaica. The before-after analysis is done in Section 5.1, and the counterfactual analysis in Section 5.2. 5.1 Before-After Analysis The before-after approach compares the performances of target economic variables during a particular program regime with their performances prior to the program. The approach assumes that any changes in the variables in the reform period are attributed solely to the reforms in that period. Two before-after estimators, which have been used extensively in the literature to evaluate Fund/Bank economic reform programs, are the mean change in the target variable and its variance (Khan, 1990). The mean estimator can be expressed as fj,y = Y2 Yi, where Y 2 and Y 1 are the means of the variable in the reform and pre-reform periods, respectively. If fj,y > 0, this is interpreted to mean that the reforms impacted positively on the variables. 137

PAGE 154

138 Price-variability is measured by the coefficient of variation, standard expressed as CV= s =, X deviation and mean where s, and X are the respectively. A brief justification for using these estimators is warranted here. One of the primary aims of the economic reforms in agriculture is to provide farmers with prices that are remunerative enough to induce output growth. Indeed, the observation that government interventions have been historically biased against agriculture (Schiff and Valdes, 1992a), led naturally to the belief by the IMF and World Bank that the economic reforms would reduce government distortion of prices, and improve price and non-price incentives to agriculture (World Bank, 1996). In this regard, it seems logical to investigate whether the reforms have ushered-in stimulative prices in agriculture. However, a related but antecedent question is what constitutes a stimulative price shift? Given the biological lag between planting and harvesting in agriculture, and the observation that in developing countries few farmers can fully mitigate against temporal price risk because of lack of resources, modern microeconomic theory provides little predictive confidence on agricultural production response to a given shift in the price distribution. However, micro

PAGE 155

139 economic theory of the firm under price uncertainty does postulate a positive relationship between output and the mean of output price, and a negative relationship between output and the variance of output price (Baron, 1970; Sandmo, 1971; Ishii, 1977). The issues raised above therefore justify an empirical examination of the first two moments of the price and quantity series used in the crop supply response models. In order to facilitate comparative analysis of variability among the different crops, the coefficient of variation is used instead of the variance. As mentioned in chapter 1, the period 1962-1979 is considered as the pre-reform period, and 1980-1999 the reform period. Two sets of estimates of the means and coefficients of variation for all eight crops' output and prices are computed. The first set covers the entire pre-reform and the reform periods, and the second set uses observations from the last five years in the pre-reform, and the first five years in the reform periods, respectively. These estimates are reported in Tables 5 .1 and 5. 2, and are robust with respect to the time periods used. With respect to quantity, Table 5 .1 shows that the mean output was higher in the reform period for four of the eight crops analyzed. These are coffee, pimento, yam and

PAGE 156

Table 5.1: Descriptive Price and Quantity Statistics --1962-1979 and 1980-1999. Pre-Reform Period Reform Period 1962-1979 1980-1999 Variable Mean CV Mean CV banq 134.771 0.189 98.274 0.086 banpr 2.978 0.398 8.436 0.242 sugq 175.214 0.204 93.300 0.088 suqpr 0.419 0.169 5.542 0.577 cofq 70.603 0.218 92.263 0.309 cofpr 41.021 0.219 21.485 0.229 pimq 189.586 0.197 339.261 0.148 pimpr 63.750 0.429 21. 555 0.238 yamq 74.936 0.357 135.259 0.226 yampr 7.297 0.141 9.227 0.593 orq 142.996 0.331 135.266 0.315 orpr 1. 575 0.364 11. 000 0.967 cobq 104.918 0.133 120.861 0.240 cobpr 19.774 0. 464 16.600 0.494 potq 162.886 0.237 152.036 0.307 potpr 8.654 0.212 12.390 0.382 ferp 9.432 0.219 8.215 0.214 wage 283.077 0.191 246.531 0.190 Note: CV= six, is the coefficient of variation, where, sand x are the standard deviation and mean respectively. All other variables as previously defined. 140 cocoa-bean. These increases in output are likely due to the government expansion programs in these crop activities, such as coffee and cocoa-bean rehabilitation schemes by the government, and to the stimulus of increased prices and export opportunities following the liberalization of prices in the reform period. With regards to the crops whose mean output did not increase over the reform period (banana, sugar, orange and potato), this is likely due to the

PAGE 157

Table 5.2: Descriptive Price and Quantity Statistics --1975-1979 and 1980-1984. Last 5 years of First 5 years of Pre-Reform Period Reform Period 1975-1979 1980-1984 Variable Mean CV Mean CV banq 105.833 0.118 109.286 0.055 banpr 4.428 0.186 7.771 0.393 sugq 141. 758 0.130 89.588 0.094 sugpr 0.482 0.161 1. 525 0.592 cofq 62.004 0.245 76.606 0.189 cofpr 48.084 0.165 21.103 0.152 pimq 193.388 0.241 214.163 0.425 pimpr 40.569 0.257 20.414 0.035 yamq 102.466 0.107 100.227 0.087 yampr 8.513 0.147 10.572 0.123 orq 94.350 0.145 78.325 0.017 orpr 1.116 0.591 1.061 0.089 cobq 94.169 0.095 125.868 0.207 cobpr 25.660 0.735 24.911 0.071 potq 162.018 0.266 133.925 0.321 potpr 10.746 0.112 11.568 0.139 ferp 10.020 0.246 9.175 0.219 wage 328.513 0.119 233.430 0.036 Notes: CV= six, is the coefficient of variation, where, sand x are the standard deviation and mean respectively. All other variables as previously defined. 141 persistence of structural factors, which slow down the adjustment process of output to policy changes. Table 5 .1 also shows that mean real prices for five crops (banana, sugar, yam, orange, and potato), increased over the reform period. The exceptions are coffee, pimento and cocoa-bean. However, real price-variability appears to have increased for most crop prices over the reform period. The exceptions are banana and pimento. Thus, the

PAGE 158

142 coefficient of variation for sugar price was 17 percent in the pre-reform period, but increased to 58 percent in the reform period. Similar high coefficients of variations are recorded for other crop prices. The observations made above in connection to the information in Table 5.1 are in general agreement with the information in Table 5.2. In addition to the information in Tables 5.1 and 5.2, Figures 5.1-5.8 show changes in nominal and real crop prices over the period 1962-1999. For most crops, nominal prices shifted upwards from the late 1970s and early 1980s, but the most dramatic increases are observed from the late 1980s onwards. While nominal prices have been increasing in the reform period, real prices show mixed changes as shown in Figures 5. 5-5. 8. Real prices for coffee and pimento have declined steadily over the 1962-1999 period. Some price stability and windfall gains have been observed for cocoa bean and potato prices but they seem to be on the descent, since 1987, below levels observed in the earlier years. The real prices for yam and orange increased briefly after the announced price liberalization policies by the government in 1985. By 1994 both prices were on the down turn but trending above the pre-liberalized price regime. Finally, real prices for banana and coffee showed dramatic increases

PAGE 159

25000 -----.--------~-------~ 20000 15000 10000 5000 1965 MNP SUGP 1975 1985 1995 Figure 5.1: Nominal Price Changes--Banana (BANP) and Sugar (SUGP). 40000 30000 20000 10000 1965 1975 1985 1995 Figure 5.3: Nominal Price Changes--Yam (YAMP) and Orange (ORP). 56000 ----r--------~---------, COFP 48000 --+.-----PIM P __ __, 40000 32000 24000 16000 8000 0 _J_-f 'r=i=F?i-'FF'#'i:::;::.,r-rr-.-.-,r,-rm-.,-rrrrn-rrm___J 1965 1975 1985 1995 Figure 5.2: Nominal Price Changes--Coffee (COFP) and Pimento (PIMP). 25000 COBP POTP 20000 15000 ,\: 10000 5000 0 1965 1975 1985 1995 Figure 5.4: Nominal Price Changes--Cocoa bean (COBP) and Potato (POTP).

PAGE 160

10.0 7.5 5.0 2.5 1964 BANPR SUGPR 1973 1982 1991 Figure 5.5: Real Price Changes--Banana and Sugar (1980=100). 35 ~-------~-------YR.FR /:, ORPR : 30 -i-------------' : \ 25 / \ : ... 20 i \ i \ 15 f ',, 10 5 .. .. 0 ---'-~~~~~~~..........:;..:.:.....~~~~~~~~-_J 1964 1973 1982 1991 Figure 5. 7: Real Price Changes--Yam and Orange (1980 = 100). 140 COFPR 120 PIMPR 100 80 60 40 20 0 1964 1973 1982 1991 Figure 5.6: Real Price Changes--Coffee and Pimento (1980=100). 60 ~---------~---------, COBPR POTPR 50 _J...-------~-,--J 40 30 20 10 ... __ __ __ ..... 1964 1973 1982 1991 Figure 5.8: Real Price Changes--Cocoa-bean and Potato (1980 = 100).

PAGE 161

145 in the early 1980s, and while sugar price continued upwards until 1991, banana price peaked in 1985, fluctuated somewhat, and along with sugar, have been on a decline, with some evidence of an upturn by 1998. In effect, therefore, these observations suggest that the pro-competitive effects that were expected to accompany the reforms as well as from the demise of state monopolies, may have outweighed the stability impulses of controlled prices of the pre-reform era. 5.2 Counterfactual Analysis One of the premises upon which the long-run supply elasticities reported in chapter 4 is based, is that the recent economic reforms in Jamaica influenced the variation of the economic time series which are included in the various crops' error correction models. On the basis of this premise, the long-run crop supply elasticities reported in chapter 4 could be interpreted as what has happened to supply response under the economic reforms. However, the true effect of the economic reforms on crop supply response must be based upon a comparison of the supply response estimates in chapter 4 with estimates derived under the assumption that the economic reforms, which were observed over the period 1980-1999 did not occur. Crop supply response under this assumption would

PAGE 162

146 constitute a counterfactual, against which the supply response estimates in chapter 4 should be compared. A relevant issue, in this regard, is what would have been the policies pursued by the government if the reforms were not undertaken? Answers to this question would provide the counterfactual situations. The most manageable way of constructing a counter factual using time series data, is to assume that the policies, which the government was pursuing over the 19621979 period, continued over the 1980-1999 period. Under this assumption, the time series modeling technique would forecast the 1980-1999 period by extrapolating techniques, which incorporate no information other than that available in the past. A central issue, however, is whether the forecasts should be made on individual series or as a system. Since in each crop supply response ECM the variables are shown to be in long-run equilibrium, it seems justifiable to use a systems approach to forecasting the economic time series. As such, a VAR approach is utilized and in each VAR the stochastic variables are the same as those that were included in the supply response ECM in chapter 4. The forecasts of the economic time series are shown in Figures 5.9-5.28. Although the graphs show clear distinct

PAGE 163

147 ions between the actual and forecasted variables, for some crops the two sets of series show mixed results. For example, Figure 5.9 shows that the forecast of banana quantity is above the actual series over the 1980-1999 period (the exception is in 1983 when actual and forecast quantities are equal). Similarly, Figure 5.11 shows that actual price of sugar has been above the forecast for the entire 1980-1999 period. Given the assumption under which the forecasts were made, this suggests that, in the case of Figure 5.9, if past (1962-1979) policies had continued over the 1980-1999 period, banana output would have been higher than what actually prevailed during the period of economic reforms. This would have been the case despite the fact that the forecasted banana price is lower than actual price. The one-step ahead forecasts shown in Figures 5.9-5.28 suffer from two short-comings. First, they reflect the data trend of the last few years before the end of 1980. Further, the forecasts are smooth and do not reflect the fluctuations in the actual series. Finally, the forecasts were plagued with large standard errors, and therefore provide only a possible indication rather than a precise picture of the counterfactuals.

PAGE 164

5.2 LBANQ 5 1 FOLBANQ 5.0 4.9 4.8 4.7 4.6 4 5 4 4 1964 1973 1982 1991 Figure 5.9: Banana Quantity--Actual (LBANQ) and Forecasted (FOLBANQ). 3.0 -..,...----------------~ 2.5 2.0 1 5 1.0 0.5 0.0 ----------------------< -0.5 -1 0 -1 5 ---'-~--r-r-m--r--r-m-,-,-m-r-r ............. m~m~~~~,-----J 1964 1973 1982 1991 Figure 5.11: Sugar Price--Actual (LSUGPR) and Forecasted (FOLSUGPR). 2.50 ---r-----------,-----,------, 2.25 2 00 1.75 1.50 1.25 1 00 0.75 0.50 0 25 ---'-~.,......-,-r-r-.----.----r-T""T"--r-r-m--r-r-r-r-1---r--r-m--r-r-m--r-r-m""T""T'~ 1964 1973 1982 1991 Figure 5.10: Banana Price--Actual (LBANPR) and Forecasted (FOLBANPR). 5.9 -.-------=-------------, 5 8 5 7 5 6 5.5 5.4 5.3 5.2 5 1 ---'-~--r-r-m-r-r--r-r-m-,-,-......-m--r+-m-r-r.--.-m-r-r-r,-rr-~ 1964 1973 1982 Figure 5.12: Wage--Actual Forecasted (FOLWAGE). 1991 (LWAGE) and

PAGE 165

2.6 ---,------------------2.5 2.4 2.3 2 2 2.1 2 0 1.9 1.8 1, 7 ~---.--r-r-m---,-,--.--,-m---rr-m---,-,-,--r'-m--r-r-..---,-,m---rr-m~-' 1964 1973 1982 1991 Figure 5.13: Fertilizer Price--Actual (LFERP) and Forecasted (FOLFERP). 5.25 ---,---------~-------~ 5 00 1964 1973 1982 LCOFQ FOI.COFQ 1991 Figure 5.15: Coffee Quantity--Actual (LCOFQ) and Forecasted (FOLCOFQ). 5 50 ---,----------~--L SUOQ --~ FOLSUOQ 5.25 5.00 4.75 4 50 1964 1973 1982 1991 Figure 5.14: Sugar Quantity--Actual (LSUGQ) and Forecasted (FOLSUGQ). 4 2 ---,------------------~ 4 0 3 8 3.6 3.4 3.2 3 0 2.8 2 6 ....J....---,-.,.....-,m---rr-..---,-,--,.--r---rr-m---,-l-,-,--r-m--r-r-...--r-,m---rr-..---,-,-,--' 1964 1973 1982 1991 Figure 5.16: Coffee Price--Actual (LCOFPR) and Forecasted (FOLCOFPR).

PAGE 166

6.50 ---.-----------.---------~ 6.25 6.00 5 75 5.50 5 25 5.00 4.75 LPIIIQ FOLPIIIQ 4,50 __,__-,--.--,--,,--,-,--.--,--,,--,-,--.--,--,,--,-,-~m...,...,....~~~~~~--' 1964 1973 1982 1991 Figure 5.17: Pimento Quantity--Actual (LPIMQ) and Forecasted (FOLPIMQ). 5.4 LYAMQ 5.2 FOLYAMQ 5.0 4.8 4.6 4.4 4 2 4.0 3.8 1964 1973 1982 1991 Figure 5.19: Yarn Quantity--Actual (LYAMQ) and Forecasted (FOLYAMQ). 11 0 ~------------------, 10.5 10 0 9 5 9.0 8.5 8 0 7.5 7 0 6 5 ----'--,--,---r,,--,-,--.--,--,,--,-,-...,...,....,-r-r...,...,....,..-'r-,-...,...,....m...,...,....m...,...,....~~---' 1964 1973 1982 1991 Figure 5.18: Pimento Price--Actual (LPIMPR) and Forecasted (FOLPIMPR). 3.00 ----.------------------, 2.75 2.50 2.25 2 00 1.75 1.50 1 25 __,__-r~m~m...,...,....m...,...,....r+-....,...,....m...,...,....~~~,--,-,---' 1964 1973 1982 1991 Figure 5.20: Yarn Price--Actual (LYAMPR) and Forecasted (FOLYAMPR). I-' (.Tl 0

PAGE 167

1964 1973 1982 1991 Figure 5.21: Cassava Price--Actual (LCASPR) and Forecasted (FOLCASPR). 3.6 ~----------------~ 3 0 2.4 1.8 1.2 0.6 0.0 --+-----~r-+-+--.,.._~-------0.6 2 ~---,--r-,--r-r-,--,-,-m-,-,-m-,----,-,....,,_,~-r-r--m-,-,-m-,----,-~--' 1964 1973 1982 1991 Figure 5.23: Orange Price--Actual (LORPR) and Forecasted (FOLORPR). 5 40 LORQ 5.22 FOI.ORQ 5 04 4.86 4.68 4 50 4 32 4.14 3 96 1964 1973 1982 1991 Figure 5.22: Orange Quantity--Actual (LORQ) and Forecasted (FOLORQ). 3 5 ~----------------~ 2.8 2 1 1.4 0.7 0.0 -+-----~----........-=--------,---~--0.7 -1.4 1 ~---.---r-r-m-,-,-m-,----,--r,--,c-r-r-r',---m-,-,-m""T""T""-r,--,c-r-r--r-r-.--' 1964 1973 1982 1991 Figure 5.24: Grapefruit Price--Actual (LGRPR) and Forecasted (FOLGRPR).

PAGE 168

5.25 ~--------~-------~ 5 00 4.75 4.50 1964 1973 1982 L008Q FOL008Q 1991 Figure 5.25: Cocoa-bean Quantity--Actual (LCOBQ) and Forecasted (FOLCOBQ). 5.60 LPOTQ 5.44 FOi.POTO 5.28 5 12 4.96 4.80 4.64 4.48 4.32 1964 1973 1982 1991 Figure 5.27: Orange Price--Actual (LORPR) and Forecasted (FOLORPR). 1964 1973 1982 1991 Figure 5.26: Cocoa-bean Price--Actual (LCOBPR) and Forecasted (FOLCOBPR). 1964 1973 1982 1991 Figure 5.28: Grapefruit Price--Actual (LGRPR) and Forecasted (FOLGRPR). I-' (J1 N

PAGE 169

153 A second approach was used to construct a counterfactual in order to assess the impact of the economic reforms in Jamaica. In this approach, the actual observations were divided into two separate sub-samples, 1962-1979 and 1980-1999, corresponding to the pre-reform and reform periods, respectively. Supply functions for each crop were estimated for each of the two sub-samples. The estimation procedure followed the Johansen method, which was outlined in chapter 4. Fitted values from the 1962-1979 period were then forecasted over the 1980-1999 period and these forecasts were used as the counterfactuals against which the fitted values from the 1980-1999 supply functions were compared graphically. With respect to the forecasting exercise, the BOXJENK, CORRELATE, and FORECAST instructions in the RATS computer software were used to identify, estimate and forecast the fitted series which were used as the counterfactual series. After some experimentation, the forecasts were modeled as mixed autoregressive-moving average processes, denoted as ARMA(p, q), where p and q are the autoregressive and moving average orders, respectively (Pindyck and Rubinfeld, 1991). Following the Box-Jenkins methodology (Enders, 1995; Doan, 1996), the orders of p and q were determined by examining the sample autocorrelation functions (ACF)

PAGE 170

154 partial auto-correlations functions ( PACF) and graphs of the fitted series. For each series, several models were fitted to the data and the final choice was determined on the basis of parsimony, and various diagnostic statistics. The test criteria used are the root-mean-square error (RMSE), Theil' s Inequality Coefficient (Theil U), and the Ljung-Box Q-statistics. Low RMSE is a desirable quality from a forecasting model. The Theil U statistic is independent of the units of measurement. It is the ratio of the RMSE for the forecast model to the RMSE for a "naive" forecast of no change in the dependent variable from the previous value (Doan, 1996) Values of Theil U that are less than one indicate that a model performs better (i.e., has a lower RMSE) than the naive model. Doan (1996) claims that a Theil U of 0.8 or less is reasonable for a univariate forecast model. The values of (p, q) were determined by the ACF and PACF. For an ARMA(p, q) process, the ACF begins to decay at lag q, and the PACF begins to decay at lag p. In addition to the ACF and PACF, the Ljung-Box Q-statistic was used in model selection. The Q-statistic is a chi-square (X 2 ) test used to test the null hypothesis of no significant autocorrelations. The diagnostic statistics for the fore casts of fitted values are reported 1n Table 5.3.

PAGE 171

Table 5.3: Diagnostic Statistics for Forecasts of Fitted Values. Crop RMSE Theil u Q (2) a Q ( 4) a Banana 0.167 0.353 3.397 3.963 Sugar 4.799 0.524 0.062 0.266 Coffee 0.378 0.880 0.061 0.368 Pimento 0.094 0.838 2.304 3.559 Yam 0.190 0.538 0.177 3.596 Orange 0.710 0.598 2.324 5.878 Cocoa-bean 5.769 1.188 0.902 4.558 Potato 0.325 0. 262 1.957 3.095 a Ljung-Box Q-statistics. x~ and x~ critical values at 95% level are 5.99 and 9.49 respectively. RMSE = Root Mean Square Error. 155 The estimated a and 13 vectors for each crop for the two sample periods are reported in Tables 5. 4 and 5. 5, respectively. From Table 5.4 the most economically plausible equations (i.e., equations that meet a priori (sign) expectations) were selected to generate fitted values for the dependent variable in each crops' ECM for the 1962-79 period. For example, the fitted values for banana for the 1962-79 period were estimated as follows: BANFIT = .300 + 0.781*P 0 0.242*Pa 0.758*W 0.854*F where all coefficients are taken from the CE #1 relation for 13 for banana in Table 5. 4. These fitted values were then forecasted, using the methodology described above, over the 1980-99 period, and used as the counterfactual for banana.

PAGE 172

156 In a manner similar to that described above, the fitted values for each crops' ECM were estimated for the 1980-99 period. For example, the fitted values for banana for the 1980-99 period were estimated as follows: BANFITA = -4.140 + 0.172*P 0 0.103*Pa 0.040*W 0.003*F where all coefficients are taken from the CE #2 relation for for banana in Table 5.5. The 'A' appended to 'BANFITA' is used to denote that the actual observations, Po, Pa, Wand F, were used to estimate the fitted series. The two sets of fitted values, the (forecasted) 'counterfactual' and 'actual', for each crop, were then plotted on the same graph. These graphs are shown in Figures 5.29-5.36. From a visual inspection of those graphs the following observations are noted. First, banana is the only crop in the sample for which the fitted values from the reform period are clearly above the fitted values for the counterfactual over the entire 1980-1999 period. Excluding the years 1987-1988, this observation is also applicable for pimento. The interpretation of this observation, based upon the assumptions that have been used to generate the fitted and forecasted series, is that if there were no change in the policy regime in Jamaican over

PAGE 173

157 Table S. 4: Estimated Long-run and Adjustment Coefficients, P's, a's for all Crops, 1962-1979. Crops Q Po Pa w F C Banana CE #1 [}' s 1 0.781 -0.242 -0.758 -0.854 0.300
PAGE 174

Table 5.4: Cont'd. Crops Q Po Pa w Cocoabean CE *1 -P's 1 -0.313 0.035 0.603
PAGE 175

159 Table 5.5: Estimated Long-run and Adjustment Coefficients, P's, u's for all Crops, 1980-1999. Crops Q Po Pa w F C Banana CE #1 1 -0.249 0.255 -0.006 0.921 -6.383 a's -0 .135 0.093 -1. 942 0.344 -0.805 (-1.534) (0.230) (-3.814)a (2.532)b (-3.89l)a CE #2 P's 1 0 .172 -0.103 -0.040 -0.003 -4.140 a's -1.047 -2.308 1.043 0.310 0.559 (-4.456)a (-2.150)b (0. 769) (0.858) (1.015) CE #3 P's 1 0.306 0.069 0.029 0.916 -7.344 a's -0.064 -1.064 -0.176 -0.162 -1.122 (-2.788)b (-2.880)a (-0.376) (-1. 301) (-5.918)a Sugar CE #1 1 0.255 -0.249 -0.006 -0.921 5.879 a's -0.135 -1. 942 0.093 0.344 -0.805 (-2.534)b (-3.814)a (0.230) (2.532)b (-3.89l)a CE #2 1 0.103 -0 .172 -0.040 -0.003 -3.636 a's -1.047 -1.043 2.308 0.310 0.559 (-4.456)a (-0.769) (2.150)b (0.858) ( 1. 015) CE #3 1 0.069 0.306 0.029 0.916 -6.839 a' s -0.064 -0 .176 -1. 064 -0.162 -1.122 (-0.788) (-0.376) (-2.880)a (-1.301) (-5.918)a Coffee CE #1 -~'s 1 -1.140 2.259 0.956 6.905 -25.382 a's -0.127 0.087 -0.143 -0.018 -0.193 (-1.792)c ( 1. 4 67) (-2.749)b (-0.600) (-7.023)a CE #2 -P's 1 1.108 -0.231 -0.440 -0 .116 -6.146 a's -0.016 -0. 711 o. 717 0.031 0.223 (-0.066) (-3.460)a (3.972)a (0.301) (2.339)b Pimento CE #1 p;-;1 -0.190 1.410 0.907 1.680 -15.165 a's -0.048 0.286 -0.380 -0.015 -0.127 (-0.446) (4.779)a (-3. 760)a (-0.281) (-2.lll)b CE #2 1 0.282 -0.406 -1.181 -0.388 -9.764 a.' s -0.769 -0.096 0.087 0.021 0.010 (-5.208)a (-l.186)c (2.629)b (0.293) (0.125)

PAGE 176

Table 5.5: Cont'd. Crops Q P o Pa w Yam CE *1 P's 1 2.543 -3.168 -0.052 s -0 .131 -0.391 0.030 0.075 (-2.183)b (-3.944)a (2.464)b (1.301) CE #2 -P's 1 0.356 -0.248 1.805 s -0.253 -0.744 0.384 -0 .179 (-3.344)a (-5.960)a (4.703)a (-2.470)a CE #3 -P's 1 -0.301 0.469 0.122 a's -0.336 0.564 -0.099 -0.024 (-3.850)a (3.928)a (-1.049) (-0.286) Orang CE U pr-;1 -5.652 4.301 -4.485 a's -0.047 0.089 -0.306 0.059 (-1.001) (0.816) (-5.225) a (2.889)a CE #2 pr-;1 0.224 -0.477 -0 .115 a's -0.271 -1.347 1.369 0.017 (-1.789) c (-3.846)a (7 .267) a (0. 260) CE #3 -P's 1 0.050 -0.307 1.185 a's -0.197 -0.576 0.453 0.049 (-1.147) (-1. 446) (2.114)b (0.655) Cocoabean CE U 6's 1 0.016 -1.159 -0.505 a's -0.153 -0.384 0.432 0.098 (-0.884) (-2.326)b (2.107)b ( 1. 027) CE #2 pr-;1 -0.302 -0.362 1.106 s -1. 831 0.558 0.532 -0.179 (-5.336) 8 (l.708)c (1.312) (-0.952) Potato CE U -P's 1 0. 971 -1.274 -2.240 s -0.081 -0.083 0.136 0.032 (-0.942) (-1.295) (5.318) 8 (l.928)c CE #2 -P's 1 -0.008 0.612 0.030 s -0.917 0. 472 -0.225 -0.179 (-1.239) (0. 855) (-1.016) (-1. 263) Note: Figures in parentheses are t-values. CE indicates 'cointegration equation'. 160 F C -0.543 -7.369 0.012 (0.100) 0.503 -18.315 -0.193 (-1.246) 0.980 -11.136 -0.453 (-2.533)b -4.097 32.444 0.049 (1.496) -1. 609 -0.501 0.237 (2.240)b 0.468 1.084 -0.527 (-4.348) 8 -2.212 -0.353 0.603 (4.688) 8 0. 471 -10.230 -0.216 (-0.851) -0.702 17.404 0.049 (1.019) 0.303 -6.541 -0.138 (-0.331) a, b, c, indicate statistical significance at one, five and ten percent levels, respectively.

PAGE 177

6.2 --,---------------------, 6.0 5 8 5 6 5 4 5 2 5 0 --------------' ....... 4 8 ............. .,// '.,/ IANF I TA 4.6 -'------ir--r--.-.--r-r--.---r-r-r---,----.-l,---,-----,--~T""""""T IA NF ~ ITC -.-~_J 1980 1985 1990 1995 Figure 5.29: Banana--Fitted Output for Reform Period (BANFITA) and Counterfactual (BANFITC) 150 --,----------.....-----------, 125 100 75 50 25 COFFITA COFFITC ,-. : ," .:' -.,; 0 ~---,--.---.--r-r--.-----.--.----r----,---.--.----r----,---.--.----r----,---.--.----r~ 1980 1985 1990 1995 Figure 5.31: Coffee--Fitted Output for Reform Period (COFFITA) and Counterfactual (COFFITC) 160 -,.------------,.-----8 UG FIT A _-, SUGFITC 155 150 145 140 135 130 : : / \ : : ,. 125 -'---r--r~~r--r~~~-~~-.-~~-.-~--' 1980 1985 1990 1995 Figure 5.30: Sugar--Fitted Output for Reform Period (SUGFITA) and Counterfactual (SUGFITC). 12800 -,.-----~ w ,r ,.. --.....---------, ~MFITC 11200 ---1--------~ 9600 8000 6400 4800 3200 .... ,./ \ __ ........ 1600 --'-------,-----.-~~-,---,----,--~~---,-~~-.--,---,----,--~ 1981 1986 1991 1996 Figure 5.32: Pimento--Fitted Output for Reform Period (PIMFITA) and Counterfactual (PIMFITC).

PAGE 178

8.40 ---.----------------~ 8.00 7.60 7 20 6.80 6.40 6.00 --t-----------, YAMFITA S : SO ~-.------r-r--r-~ Y~ ITC -~~'-,--~~~-,......,..---' 1981 1986 1991 1996 Figure 5.33: Yam--Fitted Output for Reform Period (YAMFITA) and Counterfactual (YAMFITC) 14 -,-----------,-----------, 12 10 8 6 4 2 COBFITA COBFITC 0 --'--~~--r-~~--r--~-~-,......,..~~-_J 1980 1985 1990 1995 Figure 5.35: Cocoa-bean--Fitted Output for Reform Period (COBFITA) and Counterfactual (COBFITC) 70 60 50 40 30 20 10 .. 0 1980 Figure 5.34: Reform Period (ORFITC) ORFITA ORFITC 1985 1990 1995 Orange--Fitted Output for (ORFITA) and Counterfactual 8 ~---------------------, 7 6 5 4 POTFITA POTFITC ' . ' . 3 -L--~~~-~ ~~~~-r--~~~~-.,...-~~ 1981 1986 1991 1996 Figure 5.36: Potato--Fitted Output for Reform Period (POTFITA) and Counterfactual (POTFITC).

PAGE 179

163 the 1980-1999 period, then the output of banana and pimento, with given price incentives, would have been lower than that which were actually observed under the reform policy regime. In other words, it appears that the reforms impacted positively on these two crops over the 1980-1999 period. Second, for some crops there are two distinct periods over 1980-1999 in which the fitted values from the reform period are either above or below the counterfactual. For example, over the 1980-1992 period the fitted values for sugar for the reform period were below those for the counterfactual. The reverse is observed for the series since 1992. Similarly, fitted values for coffee for the reform period have been above the counterfactual over the 1980-1995 period but the situation reversed after 1995. Similar observations can be made for orange and cocoa-bean. Finally, the cases of yarn and potato are different from the two patterns observed for the other crops. Yarn shows an oscillating pattern. Fitted values for the counterfactual were below fitted values for the reform period over 1980-1985. Potato showed similar oscillations but with relatively wider gaps within each oscillation. In effect, therefore, based on the counterfactual analysis, the data seem to suggest that the impact of the

PAGE 180

164 reforms are crop and time specific. For some crops it would appear that the reforms impacted negatively on output trends in the 1990s. This is the case of coffee, orange, cocoa-bean and potato. For banana and pimento the reforms appear to impact positively on output responses. For yam and potato, similar straightforward conclusions cannot be made since the reforms appear to impact positively on output response in some periods and negatively in others. 5.3 Summary Comparative analysis of crop output and prices between the pre-reform and reform periods suggests a mixed record regarding the impacts of economic reforms in Jamaican agriculture. For some crops, mean output increased under the reform periods, while for others, output contracted. Clear evidence of greater price variability is observed in the reform period compared to the pre-reform era. The counterfactual analysis suggests that the effects of the reforms are crop and time dependent.

PAGE 181

CHAPTER 6 CONCLUSIONS AND POLICY IMPLICATIONS Jamaica embarked upon a process of economic policy re orientation in the late 1970s, at a time when the economy was showing signs of acute economic crisis. Mainly the IMF and World Bank sponsored the economic reforms in consultation with the Jamaican government. Although, at various points in time, there were clear signs of opposition by the government towards the economic reforms emanating from the Bretton Woods Institutions, by 1990, Jamaica had, to a large degree, embraced most of the reform policies. Essentially pragmatic, the reforms highlighted the need for a more outward orientation of the Jamaican economy. Much emphasis has been placed on being competitive in world markets and also upon relying on market forces to determine prices and to allocate increasingly scarce capital resources. The reforms also entailed a critical re assessment of the role of the state in terms of its size, overall efficiency, and its relationships with the private sector. As a result of these considerations, topics such as 165

PAGE 182

166 deregulation, privatization and economic liberalization acquired increased attention in the policy debates over the past two decades. The reforms that began in the late 1980s engendered an entirely new economic environment for agriculture. However, it posed numerous challenges to economic agents such as small farmers and other agricultural producers generally, who had hitherto operated under various subsidies and guaranteed marketing and pricing schemes. In this regard, one government agency reports that Since the liberalization of the economy and the sub sequent restructuring of government support for essential services in the agricultural sector, farmers have had some problems remaining competitive. (Jamaica, Planning Institute of Jamaica, 1996, p.7.12) The quotation above is instructive insofar as it may have identified an important aspect of economic reality in Jamaica, namely, the difficulty of farm producers to make a successful transition between policy regimes, which in many respects make differential demands on economic actors. This conclusion achieves some degree of plausibility in light of the observation that the transformation or re-ordering of economic policy priorities constitutes a fundamental departure from the basic policy directives that had been adopted in Jamaica from the 1960s through to the late 1970s. These policies encouraged inward looking industrial

PAGE 183

167 ization; state-sponsored import substitution and nationalization of foreign enterprises; extensive use of price controls and subsidies; and, generally, a relatively strong populist approach to the solution of poverty, low income, unemployment, and other socio-economic problems. One of the principal tasks of this study has been to investigate the impact of the economic reforms on crop supply response in Jamaica. The estimates on crop supply responses presented in this study suggest that long-run equilibrating relationships exist which link crop output and price incentives for the eight crops studied. This result has important policy implications insofar as it indicates a significant relationship between agricultural supply and price incentives. In particular, and on the basis of the analytical method used in this study, the link between output and price incentives means that these variables move together over time and more importantly, respond to the same shocks in the system, albeit with varying degrees. Further, as markets become more competitive, this link becomes stronger since market signals are transmitted more effectively and efficiently. Although economic theory is generally silent as to what constitutes a stimulative price shift for agricultural producers, the econometric evidence presented in this study

PAGE 184

168 indicates a positive relationship between output supply and own-price. Crops such as banana and sugar, whose prices are still determined by preferential agreements, enjoyed increased real prices over the reform period. Jamaican sugar and banana have guaranteed access to the European Union (EU) market via the Lome Conventions. The prices paid for Jamaican exports in this market are significantly higher than world market prices. However, the EU' s Common Agricultural Policy (CAP) requires progressive reductions of subsidies to African Caribbean and Pacific (ACP) countries. This is expected to reduce the prices paid for ACP exports. In addition, with respect to bananas, EU's arrangements with the ACP countries are constantly being challenged by Central and Latin American (Dollar) banana countries. Proposals are for the indi victual ACP quotas to be replaced by a single global ACP group quota. While the tariff preference is expected to continue in the future, it is under constant review. The main implication is that Jamaican farmers must reduce cost of production in order to compete in the EU market. The twin issues of reducing cost of production and achieving international competitiveness appear to be major goals in the government's National Industrial Policy which

PAGE 185

169 was tabled in 1996. In this regard, J$750 million in international funding was sourced and allocated to various projects in response to the frequent debates in the EU to reduce preferential treatments of ACP exports (Jamaica, Planning Institute of Jamaica, 1998). In the case of banana, various projects have been implemented since the mid-1990s. These include programs for improved management of water systems and the banana disease, Sigatoka. For small and medium size farms, high cost and unavailability of inputs are major constraints to improving productivity. For these farmers the government has invested J$38 million to set up a supply store and has established a revolving loan scheme for improvements in irrigation and other infrastructural works. In addition, other projects have been implemented which aim at improving banana quality, such as experiments in de-handling, deflowering, hardening, and packaging and storage. bunch There is strong evidence to suggest that there are considerable constraints in the economic system, which slow down the adjustment process for economic variables. This is evidenced by the prevalence of low adjustment coefficients, which were estimated along with the long-run relationships. Own-price and quantity adjustments appear to exercise the major weight in the adjustment of the short-run to the

PAGE 186

long-run equilibrium process. 170 While significant elasticities were observed for the input prices, generally these were low, but fall within the range of estimates reported in other studies on supply response in Jamaica. The low input price elasticities are not surprising since inputs such as wages and fertilizer constitute important components in the production process. However, if these inputs are not to appear as constraints to production, policy initiatives have to be undertaken to make them appealing to producers. Given the biological lag between planting and harvesting seasons in agriculture, adjustments in the sector are necessarily slow processes. Additional factors that contribute to slow adjustments are the institutional and structural framework within which agricultural producers must operate. Slow adjustments in the short-run are signs of constraining government regulations, inadequate supportive infrastructure, and lack of credit to farmers. Although not explicitly incorporated in the modeling framework used in this study, these factors have been listed by farmers in coffee, cocoa-bean, citrus and yam production as major constraints to increasing output (Jamaica, Planning Institute of Jamaica, 1994, 1996, 1998).

PAGE 187

171 In an earlier section of this study, some of these factors were identified as the conduits through which policies are transmitted to target variables. It appears that the government has been aware of these problem as can be gleaned from the following quotation: For positive growth in the sector to be sustained, there is urgent need for infrastructural development and a strengthening of research and extension services ... This need has assumed an increasing sense of urgency in light of pending changes on the global economic landscape with the new General Agreement on Tariffs and Trade (GATT). (Jamaica, Planning Institute of Jamaica, 1994, p.7.1) To the extent that this is true, further improvements in crop supply response would be conditional upon the intensification and broadening of the reform process in agriculture. Indeed, the general decline in agricultural production since the mid-1990s, prompted the government to outline policy guidelines aimed at correcting some of the problems impacting the sector. These guidelines emphasized improving the competitiveness of traditional and non traditional exports in international markets; creating an enabling environment for farmers through additional capital funding over the long, medium and short-run period; implementation of projects/programs aimed at improving production and productivity for local and export markets; and improving support services (Jamaica, Planning Institute of Jamaica, 1998).

PAGE 188

172 While it may appear that government policies are being implemented to facilitate the farmer to operate in the new policy environment, budgetary constraints may not always permit an effective solution. For example, in light of increasing demands by farmers for agricultural credit at concessionary rates, the government offered agricultural loans at subsidized rate (13 percent). However, loans allocated to farmers have dropped almost 50 percent over the past five years because of budgetary constraints. The empirical evidence provides only partial support for the hypothesis that the reforms impacted significantly on agricultural supply responses. In particular, even when controlling for policy regime changes by constructing counterfactuals against which the reforms could be compared, it was found that for the eight crops included in this study, the effects are crop and time dependent. Although nominal crop prices have increased dramatically during the reform period, real crop prices have remained stable or have declined generally. Movements in the exchange rate and the general price level seem to be the immediate causes for these phenomena. It appears also that price variability increased significantly during the reform period. Given that farmers in Jamaica, as in most developing countries, can rarely mitigate against temporal

PAGE 189

173 price risk, the evidence seems to suggest that the procompetitive effects of the reforms may have acted more as a depressive factor on supply response, compared to the stability of the pre-reform period characterized by stable and guaranteed markets and prices by government commodity boards. A relevant question which may be posed here is how representative is the sample of crops which was chosen for the study. The eight crops were not randomly chosen but were selected the same way a researcher might have chosen to study just one of the crops. However, the crops included in the study include the most important traditional export crops of Jamaica. It is unlikely that any of these crops would have been insulated from the economic reforms over the past two decades. However, whether the conclusions from this study can be generalized for the other crops that are grown in Jamaica, is an empirical question. Further insights into the impact of the reforms on crop supply responses in Jamaica would be obtained by more intensive study of single crops. This will allow more in depth study, and guard against using the same modeling framework to accommodate crops whose supply dynamics may be different.

PAGE 190

APPENDIX A PRICE DATA USED IN CROP'S ECM ON JAMAICA, 1962-1999 (J$) Year banana 1962 28 1963 29 1964 29 1965 30 1966 30 1967 34 1968 34 1969 29 1970 30 1971 51 1972 49 1973 73 1974 159 1975 188 1976 196 1977 196 1978 175 1979 347 1980 520 1981 640 1982 700 1983 1,190 1984 1,998 1985 2,065 1986 2,249 1987 2,866 1988 2,668 1989 2,337 1990 3,748 1991 7,046 1992 8,818 1993 11,000 1994 17,637 1995 20,073 1996 17,813 1997 17,680 1998 18,510 1999 22,522 Source: ( 1) Institute of Food and Agricultural Cocoasugar coffee pimento yam orange bean potato 5 542 1050 80 18 210 99 5 549 1052 86 21 219 99 5 560 1065 88 25 228 102 6 570 1075 90 27 232 102 6 570 1,102 91 33 236 104 6 641 2,121 96 38 263 107 6 425 1,047 101 31 276 114 6 572 1,102 107 29 265 120 6 569 1,165 118 31 272 133 7 613 1,102 140 31 344 141 8 674 1,146 141 25 350 132 9 839 1,005 176 37 369 242 11 1,272 1,397 273 57 451 309 19 1,784 1,947 342 26 559 350 19 1,966 1,994 399 64 566 466 21 2,483 1,837 424 78 616 573 32 3,560 1,900 424 21 3,590 651 45 3,000 2,010 551 42 3,000 816 71 2,205 2,100 1,168 104 2,208 1,230 102 2,021 2,205 1,224 108 2,991 1,100 130 2,572 2,425 1,168 122 2,991 1,320 239 2,940 2,425 1,345 122 2,991 1,587 483 3,123 3,417 1,499 208 4,228 1,852 762 5,144 5,071 1,631 380 5,104 2,601 908 8,157 7,716 1,764 454 6,022 4,189 1,249 7,716 8,047 2,557 381 6,482 5,754 1,354 7,716 8,818 3,527 536 6,941 5,225 1,849 7,800 9,149 5,842 3,960 6,556 5,688 4,943 8,000 9,590 4,410 2,860 6,756 4,670 5,324 11,684 11,464 7,980 8,800 7,706 10,780 9,743 27,925 23,148 12,050 30,800 8,000 16,950 10,957 27,355 27,550 14,700 44,660 9,000 19,880 14,825 30,433 33,460 15,000 45,000 10,000 20,000 18,120 43,333 34,171 11,400 45,522 21,000 15,200 20,456 43,333 36,376 11,562 45,521 21,987 16,982 17,170 43,333 45,194 11,256 45,987 21,558 16,989 16,752 46,065 50,706 11,789 45,887 22,154 17,586 17,635 48,221 52,602 12,156 45,978 22,546 17,892 1962-65, 1995-1999: Jamaica, Statistical Jamaica, (STATIN), (Data Files) (2) 1966-1995: Agricultural Organization (FAO). FAOSTAT Data. http://fao.org. August, 1998. 174

PAGE 191

APPENDIX B QUANTITY DATA USED IN CROPS' ECM, 1962-1999 (METRIC TONNES) Year banana sugar coffee pimento yarn 1962 221741 523215 1721 3012 61245 1963 210523 521000 1785 3012 62546 1964 226331 512663 1798 3111 62946 1965 231046 505621 1852 3125 63012 1966 240,400 508,247 1,861 3,200 63,486 1967 234,200 455,815 1,991 4,500 74,580 1968 198,000 451,895 1,972 4,800 66,306 1969 196,000 389,124 1,589 3,800 62,556 1970 195,000 376,057 1,805 2,671 80,825 1971 187,000 385,029 1,358 3,510 122,864 1972 198,000 379,214 1,040 3,290 127,489 1973 169,000 331,181 1,146 3,410 127,650 1974 132,000 372,391 1,486 3,570 131,323 1975 127,000 360,578 1,186 3,320 134,254 1976 140,000 368,737 1,929 4,970 119,566 1977 160,000 295,237 1,208 3,320 124,941 1978 160,000 291,541 1,477 3,830 149,767 1979 170,000 283,084 958 2,340 157,169 1980 140,000 231,800 2,216 1,840 132,893 1981 150,000 225,000 1,379 4,266 136,410 1982 160,000 195,531 1,516 2,759 116,978 1983 160,000 198,202 1,632 4,729 130,633 1984 155,000 187,791 1,745 6,109 149,060 1985 150,000 225,492 1,274 5,390 163,763 1986 145,000 206,136 1,697 4,162 165,633 1987 140,000 188,985 1,658 4,412 175,628 1988 135,000 221,706 2,231 3,834 166,864 1989 130,000 204,915 1,260 5,444 133,281 1990 127,660 208,592 1,560 3,770 161,462 1991 134,000 235,904 2,280 3,750 186,104 1992 130,000 228,024 1,920 5,730 214,386 1993 125,000 219,046 1,500 7,120 221,928 1994 120,000 223,041 2,460 9,370 233,912 1995 130,000 248,558 2,580 9,630 240,371 1996 130,000 237,943 2,580 10,370 253,371 1997 130,000 236,510 2,887 10,400 212,567 1998 130,000 179,251 3,104 10,542 198,402 1999 130,000 223,000 3,412 11,221 201,354 Source: (1) 1962-65, 1995-1999: Institute of Jamaica, (STATIN), (Data Food and Agricultural Organization cultural Data. http://fao.org. August, 175 oranqe Cocoa-bean potato 73125 1542 10257 74552 1945 11450 75489 1854 12453 78562 1945 10254 79,772 2,174 13,608 82,236 1,642 13,297 85,085 1,864 7,454 73,304 1,857 8,717 65, 142 2,163 8,440 72,726 1,872 13,207 83,160 2,378 16,132 23,408 1,952 8,941 39,309 1,618 14,606 40,849 1,780 13,617 41,234 1,646 7,643 42,042 1,692 7,579 49,665 1,797 11,794 31,224 1,366 11,331 43,159 1,752 6,848 33,000 1,752 12,428 32,879 2,408 6,667 31,686 2,302 7,603 28,298 2,812 12,310 42,119 2,380 7,075 41,850 2,406 5,439 75,460 3,186 9,443 50,820 1,549 9,893 57,750 1,386 10,818 82,236 2,104 14,296 61,908 1,750 7,548 70,000 2,478 6,935 72,000 2,522 9,134 72,546 2,576 12,188 72,892 2,538 17,036 73,452 1,407 13,775 74,561 1,653 12,661 74,987 1,687 12,875 75,984 1,702 13,256 Jamaica, Statistical Files) (2) 1966-1995: (FAO). FAOSTAT Agri1998.

PAGE 192

APPENDIX C SELECTED ECONOMIC STATISTICS ON JAMAICA Year Adef ferp 1962 13.50 144 1963 13.80 138 1964 14 .10 139 1965 14.50 138 1966 15.90 124 1967 16.70 137 1968 17.40 132 1969 19.70 142 1970 20.50 146 1971 20.40 144 1972 20.10 149 1973 23.90 193 1974 31.37 229 1975 38.13 298 1976 44.75 339 1977 52.40 492 1978 72.09 778 1979 86.54 1,007 1980 100.00 1,158 1981 103.67 1,228 1982 110. 58 983 1983 133.06 966 1984 198.67 1,149 1985 249.27 1,582 1986 316.83 1,820 1987 354.07 2,013 1988 413.28 1,913 1989 442.73 2,697 1990 546.16 4,636 1991 835.26 6,776 1992 869.95 7,704 1993 893.32 7,725 1994 804.54 13,904 1995 841. 98 19,017 1996 879.42 18,136 1997 886.67 21,176 1998 897.56 20,254 1999 909.78 21,924 Sources: (1) Data files, Jamaica (STATIN); (2) Jamaica (various issues). rwage 208.74 209.58 210.28 212.34 216.42 255.69 289 .11 300.95 289.13 299.33 325.47 307.26 302.97 362.47 361.25 354.16 326.04 264.19 234.02 221. 24 231.40 244.66 235.83 262. 21 277.54 316.75 340.99 314.55 268.05 182.63 175.35 179.30 206.60 218.66 213.69 238.27 270.68 298.20 Jamaica, Jamaica, Cpi80 gdef ER 11. 60 0.40 0. 71 11. 95 0. 40 0. 71 12.20 0.40 0. 71 12.71 0.40 0.71 14 .60 0.50 0. 71 15.16 0.50 0.83 16.43 0.60 0.84 17.78 0.80 0.83 19.01 0.90 0.84 19.58 0.90 0.78 21.03 0.90 0.85 24.81 1.10 0.91 31.53 1. 40 0.91 37.04 1. 80 0.91 40.64 1. 90 0.91 45.02 2.20 0.91 60.97 2.70 1. 70 78.79 3.20 1. 78 100.00 3.80 1. 78 112. 75 4.10 1. 78 119. 92 4.50 1. 78 113. 86 5.20 3.28 170.91 7.00 4.93 215.14 9.20 5.48 247.41 10.80 5.48 263.75 12.00 5.50 285.66 13.60 5.48 326.27 15.30 6.48 398.45 18.90 8.04 601.99 27.90 21.49 1,066.93 44.40 22.19 1,302.79 59.70 32.48 1,759.36 78.40 33.20 2,109.56 100.00 39. 62 2,666.53 121.10 34.87 2,934.30 134.30 41.02 2,942.14 154.00 42.40 3,023.26 159.00 42.84 Statistical Institute of Planning Institute of Notes: Adef = Agricultural deflator (1980=100); ferp = Average price of fertilizer; rwage = index of average agricultural real wage (1980=100); CPI80 = consumer price index (1980=100); gdef = GDP deflater (1980=100); ER = Jamaican exchange rate ($J/US). 176

PAGE 193

APPENDIX D DIAGNOSTIC STATISTICS FOR MODEL SPECIFICATION AND TEST STATISTICS FOR COINTEGRATION Table D-1: Tests of Cointegration Rank for Sugar Ea. Eigen values Amax Trace Ho:r P-r Amax 90 Trace 90 0.649 37.66 80.99 0 5 21.73 71. 66 0.369 16.56 43.33 1 4 18.03 49.92 0.332 14.54 26.77 2 3 14.09 31.88 0.220 8.96 12.23 3 2 10.29 17.79 0.087 3.27 3.27 4 1 7.50 7.50 Table D-2: Diagnostic Statistics for Residual Tests for the Sugar Ea. Tests De x2 p-value For normality of residuals 10 15.108 0.10 For autocorrelation L-B(9) 195 124.208 0.21 LM (1) 25 23.084 0.57 LM ( 4) 25 21.299 0.68 For ARCH(2) lsugq 2 1. 825 lsugpr 2 0.441 lbanpr 2 1. 555 lwage 2 0.919 lferp 2 4.600 Notes: The critical value for 'X, 2 (2) = 5.99 At the 95 percent significance level. a Df = degrees of freedom. 177

PAGE 194

Table D-3: Tests of Cointegration Rank for Coffee ECM. Eigen value lMax Trace Ho:r P-r lMax 90 Trace 90 0.675 40.44 88.74 0 5 21. 74 71. 66 0.497 24.72 48.30 1 4 18.03 49.92 0.289 12.30 23.58 2 3 14.09 31. 88 0.189 7.56 11.28 3 2 10.29 17.79 0.098 3.72 3.72 4 1 7.50 7.50 Table D-4: Diagnostic Statistics for Residual Tests for the Coffee ECM. Tests oe x2 p-value For normality of residuals 10 11.349 0.33 For autocorrelation L-B ( 9) 190 238.190 0.20 LM (1) 25 38.821 0.06 LM ( 4) 25 26.303 0.39 For ARCH(2) Lcofq 2 2.763 Lcofpr 2 4.747 Lbanpr 2 0.086 Lferp 2 2.478 Lwage 2 1.345 Notes: The critical value for X 2 (2) = 5.99 at the 95 percent significance level. a Df = degrees of freedom. Table D-5: Tests of Cointegration Rank for Pimento ECM. Eigen value lMax Trace Ho:r P-r lMax 90 Trace 90 0.608 33.75 81.34 0 5 21.74 71. 66 0.438 20.74 47.59 1 4 18.03 49.92 0.284 12.00 26.85 2 3 14.09 31. 88 0.263 10.99 14.85 3 2 10.29 17. 79 0.102 3.86 3.86 4 1 7.50 7.50 178

PAGE 195

Table D-6: Diagnostic Statistics for Residual Tests for the Pimento ECM. Tests oe 'X.2 p-value For normality of residuals 10 19.394 0.09 For autocorrelation L-B ( 9) 190 212.972 0.06 LM (1) 25 23.861 0.53 LM (4) 25 36.915 0.06 For ARCH(2) Lpimq 2 3.255 Lpimpr 2 1.131 Lbanpr 2 2.209 Lferp 2 9.196 Lwage 2 4.575 Notes: The critical value for 'X, 2 (2) = 5.99 at the 95 percent significance level. a Df = degrees of freedom. Table D-7: Diagnostic Statistics for Residual Tests for the Yam ECM. Tests Dfa 'X.2 p-value For normality of residuals 10 7.472 0.68 For autocorrelation L-B ( 9) 185 214.194 0.07 LM (1) 25 28.044 0.31 LM ( 4) 25 22.418 0.61 For ARCH(2) Lyamq 2 0.337 Lyampr 2 2.800 Lcaspr 2 3.434 Lferp 2 5.101 Lwage 2 0.065 Notes: The critical value for 'X, 2 (2) = 5.99 at the 95 percent significance level. a Df = degrees of freedom. 179

PAGE 196

Table D-8: Tests of Cointegration Rank for the Yam EOI. Eigen value 1Max Trace Ho:r P-r 1..Max 90 Trace 90 0.682 41.23 110 0 5 21. 74 71. 66 0.549 28.66 68.67 1 4 18.03 49.92 0.462 22.31 40.01 2 3 14.09 31. 88 0.246 10.17 17.70 3 2 10.29 17.79 0.189 7.53 7.53 4 1 7.50 7.50 Table D-9: Diagnostic Statistics for Residual Tests for the Orange EOI. Tests oe 'X.2 p-value For normality of residuals 10 5.691 0.10 For autocorrelation L-B(9) 195 217.078 0.12 LM (1) 25 26.658 0.37 LM(4) 25 30.412 0.21 For ARCH(2) Lorq 2 0.088 Lorpr 2 0.405 Lgrpr 2 6.827 Lferp 2 0.797 Lwage 2 6.931 Notes: The critical value for x_ 2 (2) = 5.99 at the 95 percent significance level. a Df = degrees of freedom. Table D-10: Tests of Cointegration Rank for Orange EOI. Eigen value 1Max Trace Ho:r P-r 1..Max 90 Trace 90 0.679 40.90 75.90 0 5 21. 74 71.66 0.341 15.01 34.99 1 4 18.03 49.92 0.315 13.63 19.99 2 3 14.09 31.88 0.142 5.53 6.36 3 2 10.29 17.79 0.021 0.82 0.82 4 1 7.50 7.50 180

PAGE 197

Table D-11: Diagnostic Statistics for Residual Tests for the Cocoa-bean ECM. Tests oe x2 p-value For normality of residuals 10 15.772 0.09 For autocorrelation L-B ( 9) 195 171.387 0.10 LM (1) 25 18.188 0.83 LM(4) 25 35.577 0.08 For ARCH(2) Lcobq 2 0.910 Lcobpr 2 0.438 Lbanpr 2 1.278 Lferp 2 6.781 Lwage 2 3.322 Notes: The critical value for X 2 (2) = 5.99 at the 95 percent significance level. a Df = degrees of freedom. Table 0-12: Tests of Cointegration Rank for Cocoa-bean ECM. Eigen value A.Max Trace Ho:r P-r A.Max 90 Trace 90 0.665 39.41 84.17 0 5 23.72 82.68 0.404 18.65 44.77 1 4 19.88 58.96 0.298 12.74 26.12 2 3 16.13 39.08 0.263 10.98 13.38 3 2 12.39 22.95 0.064 2.39 2.39 4 1 10.56 10.56 Table D-13: Tests for Cointegration Rank for Potato ECM. Eigen value AMax Trace Ho:r P-r 1Max 90 Trace 90 0.713 44.92 93.40 0 5 21.74 71. 66 0.415 19.30 48.47 1 4 18.03 49.92 0.302 12.92 29.17 2 3 14.09 31. 88 0.254 10.55 16.25 3 2 10.29 17.79 0.146 5.70 5.70 4 1 7.50 7.50 181

PAGE 198

Table D-14: Diagnostic Statistics for Residual Tests for the Potato E04. Tests oe t p-value For normality of residuals 10 14.295 0.16 For autocorrelation L-B ( 9) 190 218.901 0.07 LM(l) 25 24.219 0.51 LM(4) 25 22.342 0.62 For ARCH(2) Lpotq 2 1.816 Lpotpr 2 3.973 Lcaspr 2 1.283 Lferp 2 1.237 Lwage 2 1.401 Notes: The critical value for X 2 (2) = 5.99 at the 95 percent significance level. a Df = degrees of freedom. 1 82

PAGE 199

REFERENCES Alogoskoufis, G., and R. Smith. "On Error Correction Models: Specification, Interpretation, Estimation. Journal of Economic Survey. 5(1991) :97-128. Anderson, A., and M. Witter. "Crisis, Adjustment and Social Change: A Case Study of Jamaica." Consequences of Structural Adjustment: A Review of the Jamaican Experience. Elsie Le Franc, ed., pp.1-55. U.W.I., Jamaica: Canoe Press, 1994. Askari, Hossein, and John Thomas Cummings. Agricultural Supply Response: A Survey of the Econometric Evidence. New York: Praeger Publishers, 1976. Baron, David P. "Price Uncertainty, Utility, Equilibrium in Pure Competition." Economic Review. 11(1970) :463-480. and Industry International Bates, Robert. "The Reality of Structural Adjustment: A skeptical Appraisal." Structural Adjustment and Agriculture: Theory and Practice in Africa and Latin America. Simon Commander, ed., pp. 221-227. London: Overseas Development Institute, 1989. Behrman, Jere R. "Macroeconomic Policies and Rural Poverty: Issues and Research Strategies." Research Paper Series RP-145, Williams College, Williamstown, Massachusetts, 1990. Bernanke, Ben. "Alternative Explanation of Money-Income Correlation." Carnegie-Rochester Conference Series on Public Policy 25. 1986:49-100. Binswanger, H. "The Policy Response of Agriculture." Proceedings of the World Bank Annual Conference on Development Economics. Supplement to the World Bank Economic Review and the World Bank Research Observer, pp. 231-271. Washington D.C., 1989. 183

PAGE 200

184 Binswanger, H., and K Deininger. and Agrarian Policies in Working Paper Series, World 1997. "Explaining Agricultural Developing Countries." Bank, Washington D.C., Bonnick, G. "Liberalization to Centralization and Back." World Economic Growth. Arnold C. Harberger, ed., pp. 265-291. San Francisco: Institute for Contemporary Studies Press, 1984. Boyd, Derick. Economic Management, Income, Distribution, and Poverty in Jamaica. New York and London: Praeger, 1988. Brown, L. "Crisis, Adjustment and Social Change: The Middle Class under Adjustment." Consequences of Structural Adjustment: A Review of the Jamaican Experience. Elsie Le Franc, ed., pp. 56-117. U.W.I., Jamaica: Canoe Press, 1994. Campbell, H., and Howard Stein, eds. The IMF and Tanzania. Harare, Zimbabwe: Southern Africa Political Economy Series Trust, 1991. Chadha, R., S. Pohit, A. Deardorff, and R. Stern, eds. The Impact of Trade and Domestic Policy Reforms in India: A CGE Modeling Approach. Ann Arbor, Michigan: The University of Michigan Press, 1998. Chhibber, A. "The Aggregate Supply Response: A Survey." Structural Adjustment and Agriculture: Theory and Practice in Africa and Latin America. Simon Commander, ed., pp.55-68. London: Overseas Development Institute, 1989. Commander, Simon, ed. Structural Agriculture: Theory and Practice in America. London: Overseas Development Adj us tmen t and Africa and Latin Institute, 1989. Connors, Thomas A. "The Apparent Effects of Recent IMF Stabilization Programs." International Finance Discussion Paper 135. Washington: Board of Governors of the Federal Reserve System, April 1979. Davis, Carlton, Ballayram, Edward Evans, and C. Y. Thomas. "Trade Liberalization and Small Economies in the Americas: The Case of the Caribbean Community (CARICOM)." International Working Papers IW 99-1, Food

PAGE 201

and Resource Economics Department, Florida, Gainesville, February 1999. 185 University of Doan, T.A. RATS: User's Manual, Version 4.2. Estima, Illinois. 1996. Donovan, Donal J. "Macroeconomic Performance and Adjustment Under Fund-Supported Programs: The Experience of the Seventies." IMF Staff Papers. 29(1982) :171-203. "Real Responses Associated with Exchange Rate Action in Selected Upper Credit Tranche Stabilization Programs." IMF Staff Papers. 28(1981) :698-727. Doornik, J.A., and H. Hansen. "An Omnibus Test for Uni variate and Multi variate Normality." Working Paper, Nuffield College, Oxford, 1994. Enders, W. Applied Econometric Time Series. John Wiley and Sons: New York. 1995. RATS Handbook for Econometric Time Series. John Wiley and Sons: New York, 1996. Engle, Robert F., and Clive Granger. "Cointegration and Error Correction: Representation, Estimation, and Testing." Econometrica. 55(1987) :251-276. Food and Agriculture Organization (FAO). FAO Production Yearbook. Rome: FAO, various years. FAOSTAT Agricultural Data. http://fao.org. August, 1998. Fuller, W. A. Introduction to Statistical Time Series. New York: Wiley, 1976. Gafar, John. "Price Responsiveness of Tropical Agr~cultural Exports: A Case Study of Jamaica, 1954-1972." The Developing Economies. 18(1980) :288-297. "The Supply Response of Aggregate Agricultural Output in Jamaica." Agricultural Economics. 16(1997) :205-217. Gardner, B. L. The Economics of Agricultural Policies. New York: Macmillian Publishing Company, 1987.

PAGE 202

186 Garrity, Michele. "Jamaica: Business Associations and the Public Policy Process." Policy Reforms for Sustainable Development in the Caribbean. M. Garrity, and L.A. Picard, eds., pp. 39-57. Amsterdam: ISO Press, 1996. and Peter J. Montiel. "Evaluating Fund Programs with Multi-country Data: Some Goldstein, Morris, Stabilization Methodological 33(1986) :304-344. Pitfalls." IMF Staff Papers. Green, Reginald H. "Articulating Stabilization Programmes and Structural Adjustment: Sub-Saharan Africa." Structural Adjustment and Agriculture: Theory and Practice in Africa and Latin America. Simon Commander, ed. pp. 35-54. London: Overseas Development Institute, 1989. Greene, William H. Econometric Analysis, 2 nd ed. New York: Macmillian Publishing Company, 1993. Guitian, Manuel. Fund Conditionality: Evolution of Principles and No.38(1981). Practices. IMF Pamphlet Series. Gujarati, Damodar N. Basic Econometrics, 3 rd ed. New York: McGraw-Hill, Inc., 1995. Gylfason, Thorvaldur. "Credit Policy and Economic Activity in Developing Countries with IMF Stabilization Programs." Princeton Studies in International Finance, No.60, Princeton University Press, 1987. Hall, Bronwyn H., and Clint Cummins. Time Series Processor, Version 4.4: Reference Manual. TSP International: Palo Alto, California, 1998. Hansen, Henrik, and Katarina Juselius. CATS Cointegration Analysis of Time Series. Evanston, Illinois, 1995. in RATS: Estima: Hendry, D.F., A.R. Pagan, and J.D. Sargan. "Dynamic Specification." Handbook of Econometrics, Vol. II. Z. Griliches, and M. D. Intriligator, eds., pp .1023-1100. Amsterdam: Elsevier, 1984. Hylleberg, S., and G.E. Mizon. "Cointegration and Error Correction Mechanisms." Economic Journal (Supplement) 99 (1989): 113-125.

PAGE 203

187 Inter-American Development Bank (IDB). Latin America After a Decade of Reform. Washington, D.C., 1997. International Monetary Fund (IMF). International Financial Statistics. Washington, D.C., 1998. Ishii, Yasunori. "On the Theory of the Competitive Firm Under Price Uncertainty: Note." American Economic Review. 67(1977) :768-769 Jaeger, W. "The Impact of Policy in African Agriculture: An Empirical Investigation." Working Papers, WPS-640, Policy, Research and External Affairs, World Bank, Washington, D.C., 1991. Jamaica, Agency for Public Information (API), Michael Manley. "The Need for a new Economic Path." Broadcast to the Nation, Sunday 3, February, 1980. Jamaica, Bank of Jamaica. The Balance of Payments. Kingston, Jamaica: Bank of Jamaica, 1985. Statistical Digest. Kingston, Jamaica: Bank of Jamaica, various issues. Jamaica, Planning Institute of Jamaica. Economic and Social Survey: Jamaica. Kingston, Jamaica: Planning Institute of Jamaica, various years. Jamaica, Statistical Institute of Jamaica (STATIN). Census of Agriculture. Kingston, Jamaica: Statistical Institute of Jamaica, various years. National Income and Product. Kingston, Jamaica: Statistical Institute of Jamaica, 1970. Production Statistics. Kingston, Jamaica: Statistical Institute of Jamaica, various years. Statistical Yearbook of Jamaica. Kingston, Jamaica: Statistical Institute of Jamaica, various years. Jefferson, O. The Post-war Economic Development of Jamaica. Mona, Jamaica: University of the West Indies, Institute of Social and Economic Research, 1972

PAGE 204

1 88 Johansen, S0ren. Likelihood-based Inference in Cointegrated Vector Autoregressive Models. Oxford University Press: New York, 1995. "Statistical Analysis of Cointegration Journal of Economic Dynamics and Control. September 1988) :231-254. Vectors." 12(JuneJohansen, S0ren, and Katarina Juselius. "Maximum Likelihood Estimation and Inference on Cointegration with Applications to the Demand for Money." Oxford Bulletin of Economics and Statistics. 52(1990) :169-210. "Testing Structural Hypotheses in a Multi variate Cointegration Analysis of the PPP and the UIP for UK." Journal of Econometrics. 53(1992) :211-244. Kaufman, M. Jamaica under Manley: Dilemmas of Socialism and Democracy. London, U.K.: Zed Books Ltd., 1985. Khan, Mohsin S. "The Macroeconomic Supported Adjustment Programs." 37:2(1990) :195-231. effects of Fund IMF Staff Papers Khan, Mohsin, and Malcolm Knight. "Fund-Supported Adjustment Programs and Economic Growth." Occassional Paper 41, International Monetary Fund, 1985. "Stabilization Programs Formal Framework." IMF 1981) :1-53. in Developing Countries: A Staff Papers. 29 (January Krueger, A.O. The Political Economy of Agricultural Policy, Volume 5: A Synthesis of the Political in Developing Countries. Baltimore: Johns University Press, 1992. Pricing Economy Hopkins La Guerre, John, ed. Structural Adjustment, Public Policy and Administration in the Caribbean. St. Augustine Campus, Trinidad and Tobago: University of the West Indies, School of Continuing Studies, 1994. Le Franc, E., ed. Consequences of Structural Adjustment: A Review of the Jamaican Experience. U.W.I., Jamaica: Canoe Press, 1994.

PAGE 205

189 Lora, Edwardo. "A Decade of Structural Reforms in Latin America: What Has Been Reformed and How to Measure it." OCE Working Paper #348, IDB, Washington, D.C., 1997. Loxley, John. The IMF and the Poorest Countries. Ottawa, Canada: North South Institute, 1984. Lutkepohl, H., and H.E. Reimers. "Impulse Response Analysis of Cointegrated Systems". Journal of Economic Dynamics and Control. 16:1(1992) :53-78. Mamingi, N. "The Impact of Policies on Agricultural Available Results." 16(1997) :17-34. Prices and Supply: A Agricultural Macroeconomic Synthesis of Economics. Nallari, R. "Jamaica: The Search for Stabilization and Adjustment." World Bank Staff Paper, Washington, D.C., World Bank, 1992. Nelson, C.R., and C.I. Plosser. "Trends and Random Walks in Macroeconomics Time Series." Journal of Monetary Economics. 10(1982) :140-162. Newman, M., and Elsie Le Franc. "The Small-Farm Sub-Sector: Is there Life after Structural Adjustment?" Consequences of Structural Adjustment: A Review of the Jamaican Experience. Elsie Le Franc, ed. pp.118-181. U.W.I., Jamaica: Canoe Press, 1994. Nickell, S. "Error Correction Partial Adjustment and all That: An Expositionary Note." Oxford Bulletin of Economics and Statistics. 47(1985) :119-129. Osterwald-Lenum, Michael. "Practitioner's Corner: A Note with Quantiles for the Asymptotic Distribution of the Maximum Likelihood Cointegration Rank Test Statistic." Oxford Bulletin of Economics and Statistics. 54(1992) :461-471. Pantula, Sastry G., Graciela Gonzalez-Farias, and Wayne A. Fuller. "A Comparison of Unit-Root Test Criteria." Journal of Business and Economic Statistics. 12(October 1994) :449-459. Pastor, Manuel. "The Effects of IMF Programs in the Third World: Debate and Evidence for Latin America." World Development. 15(1987) :249-262.

PAGE 206

190 Pindyck, Robert S., and Daniel L. Rubinfeld. Econometric Models and Economic Forecasts 3 rd ed. New York: McGraw Hill, Inc., 1991. Phillips, P.C.B., and Root in Time 75(1988) :335-346. Pierre Perron. "Testing for a Unit Series Regression." Biometrika. Pollard, S.K., and D.H. Graham. Agricultural Export Performance Development. 13 ( 1985) : 1067-107 5. "Price Policy and in Jamaica." World Reichmann, Thomas M., and Richard T. Stillson. "Experience with Programs of Balance of Payments Adjustment: Stand by Arrangements in the Higher Credit Tranches, 196372." IMF Staff Papers. 25(1978) :293-309. Robinson, S., and C. Gehlhar. "Impacts of Macroeconomics and Trade Policies on a Market-Oriented Agriculture." Egypt's Agriculture in a Reform Era. Lehman B. Fletcher, ed., pp. 271-293. Iowa: Iowa State University Press, 1996. Sandrno, Agnar. "On the Theory of the Competitive Firm Under Price Uncertainty." American Economic Review. 61 (1971): 65-73. Schiff, Maurice, and Alberto Valdes. "A Synthesis of the Economics in Developing Countries." The Political Economy of Agricultural Pricing Policy (Vol.4): A World Bank Comparative Study. Baltimore: Johns Hopkins University Press, 1992a. The Plundering of Agriculture in Developing Countries. Washington, D.C.: The World Bank, 1992b. Serieux, John. "Democratization and Structural Adjustment in Sub-Saharan Africa: An Empirical Investigation." Canadian Journal of Development Studies. 20(1999):521544. Sharpley, Jennifer. "Jamaica, 1972-80." The IMF and Stabilization: Developing Country Experiences. Tony Killick, ed., pp.115-163. New York: St. Martin's Press Inc., 1984. Sims, Christopher. "Macroeconomics and Reality." Econometrica. 48(1980) :1-49.

PAGE 207

191 Singh, R. "The Impact of Structural Adjustment Policies on the Performance of Agriculture: The Case of Jamaica." Structural Adjustment and the Agricultural Sector in Latin America and the Caribbean. John Weeks, ed., pp. 229-257. New York: St. Martin's Press Inc., 1995. Stephens, E.H., and J.D. Stephens. Democratic Socialism in Jamaica: The Political Movement and Social Transformation in Dependent Capitalism. Princeton, New Jersey: Princeton University Press, 1986. Thomas, Desmond. "Anatomy of a Stabilization Process: The Case of Jamaica, 1984 to the Present. Canadian Journal of Development Studies. 20(1999) :159-179. Tsakok, Isabelle. Agricultural Price Policy: A Practioner's Guide to Partial-Equilibrium Analysis. Ithaca, New York: Cornell University Press, 1990. Tshibaka, T.B., ed. Structural Adjustment and Agriculture in West Africa. Dakar, Senegal: Council for the Development of Social Science Research in Africa (CODESRIA), 1998. Valdes, Alberto, and Kay Muir-Leresche, eds. Agricultural Policy Reforms and Regional Market Integration in Malawi, Zambia, and Zimbabwe. Washington, D.C.: International Food Policy Research Institute ( IFPRI) 1993. Weeks, John. Structural Adjustment and the Agricultural Sector in Latin America and the Caribbean. New York: St. Martin's Press, 1995. World Bank. Adjustment in Africa: Reforms, results, and the road ahead. New York: Oxford University Press, 1994. Trends in Developing Economies. Washington, D.C.: World Bank, 1996.

PAGE 208

BIOGRAPHICAL SKETCH Ballayram of Essequibo Coast, Guyana, worked in the primary educational and civil service systems in Guyana before pursuing higher education. He completed his B.S.Soc. degree with distinction in economics at the University of Guyana in 1980. After three years as an Assistant Lecturer in the Department of Economics at the University of Guyana, he proceeded to pursue a Master of Arts degree in development economics, under a Fulbright Scholarship, at the University of Notre Dame, Indiana. On completion of his master's degree he lectured first in the Economics Department at the University of Guyana, and then subsequently, in the Economics Department at the University of the West Indies, Mona Campus, Jamaica. Ballayram was awarded a Graduate Research Assistantship, in 1994, to pursue a Ph.D. in the Food and Resource Economics Department at the University of Florida. His areas of specialty are environmental economics, and international trade and economic development. 192

PAGE 209

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 Doctor of ~ iloSORhy . OM" G. Davis, Chair 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, and a i ty, as a dissertation for the degree of ---~ OB y. Dr 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 Ph'losophy. I certify that Professor of Food and Resource Economics opinion it conforms presentation and is fully ad as a dissertation for the d Robert D. Emerson Professor of Food and Resource Economics

PAGE 210

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. a: Dr. David A. Denslow, J Distinguished Service Professor of Food and Resource Economics This dissertation was submitted to the Graduate Faculty of the College of Agricultural and Life Sciences and to the Graduate School and was accepted as partial fulfillment of the requirement for the degree of Doctor of Philosophy. May 2001 Dean, College of Agri and Life Sciences Dean, Graduate School