Citation
Trade restrictiveness of Japanese agricultural import policies

Material Information

Title:
Trade restrictiveness of Japanese agricultural import policies
Creator:
Pantzios, Christos J., 1963-
Publication Date:
Language:
English
Physical Description:
vii, 165 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Food industry and trade -- Japan ( lcsh )
Restraint of trade -- Japan ( lcsh )
Trade regulation -- Japan ( lcsh )
Genre:
bibliography ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1993.
Bibliography:
Includes bibliographical references (leaves 159-164).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Christos J. Pantzios.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact the RDS coordinator (ufdissertations@uflib.ufl.edu) with any additional information they can provide.
Resource Identifier:
030445589 ( ALEPH )
31303752 ( OCLC )

Downloads

This item has the following downloads:


Full Text
TRADE RESTRICTIVENESS OF
JAPANESE AGRICULTURAL IMPORT POLICIES
By
CHRISTOS J. PANTZIOS
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 1993




ACKNOWLEDGMENTS
My appreciation is extended to a number of people for their support during my doctoral studies. I would first like to express my gratitude to Dr. T. G. Taylor and Dr. E. Dinopoulos. As chairman of my supervisory committee, Dr. Taylor not only contributed to my education, but he also influenced my way of thinking and my ideas as an economist. His continuous encouragement, support, and friendship were important factors for me during my studies at the University of Florida and are deeply appreciated. Although the external member of my supervisory committee, Dr. Dinopoulos took a very close interest in my work and was instrumental in the inception and theoretical justification of my research. I would like to express my appreciation for his encouragement throughout my research effort, and also for his friendship.
My deep appreciation also goes to Dr. L. Polopolus and Dr. W. G. Boggess for recommending my application at the Food and Resource Economics Department (F.R.E.) at the University of Florida and effectively initiating my Ph.D. effort. Special thanks are also owed to Dr. Boggess for his unequivocal support as a member of my supervisory committee and also his friendship throughout my Ph.D. studies at the University of Florida. My sincere appreciation is also extended to the rest of the members of my supervisory committee, Dr. J. L. Seale Jr. and Dr. G. F. Fairchild, for their contribution to my research.




I would also like to express my thanks to the staff of the System Support Center, and the secretarial personnel at the F.R.E. for their help and support during my studies. Last but not least, I would like to extend my thanks to my friends in Gainesville FL, for their concern and support during the last two years of my studies and especially to Ms Trinidad Reyes for her help, encouragement, and moral support.




TABLE OF CONTENTS
ACKNOWLEDGMENTS.....................................1ii
ABSTRACT............................................. vi
CHAPTER 1
PROBLEM STATEMENT AND OBJECTIVES OF THE STUDY.............
Introduction ........................................1I
Problem Statement and Objectives........................... 4
CHAPTER 2
MEASURES OF PROTECTION................................ 7
The Average Tariff Measuring the Height of the Tariff 'Wall'. ........7
The Nominal and Effective Rate of Protection, and the Domestic
Resource Cost.................................. 10
The Uniform Tariff Equivalent (UTE)........................ 17
The Producer (Consumer) Subsidy Equivalent (PSE and CSE)..........18 PSE and Trade Distortions............................... 22
The Trade Restrictiveness Index (TRI)........................ 24
Operationalization of the TRI............................. 26
The TRI in the Presence of Tariffs and Quotas................... 29
Comparisons Among the Measures of Protection.................. 37
CHAPTER 3
THE TRI APPROACH IN JAPAN'S AGRICULTURE.................. 41
Japan's Agricultural Trade Policies.......................... 41
The Beef Industry..................................... 43
The Pork and Poultry Industries............................ 45
The Rice Industry..................................... 47
The Wheat Industry................................... 49
The Fresh Citrus Industry................................ 51
Japan's Balance of Trade Function on Farm Imports............... 52
Demand Elasticities A System-wide Approach................... 63
Computing the Combined TRI for a Subset of Japan's Farm Imports,
1971-1987..................................... 76
iv




Comparing the TRI with the Respective PSE/CSEs ................ 85
CHAPTER 4
THE NEW GOODS PROBLEM AND THE TRI .................... 89
The TRI in the Presence of New Goods ..................... 91
Operationalizing the TRI in the Presence of New Goods ............ 94
An Application of the Generalized TRI the Case of Japan's Meat
Imports .... ................................100
Estimation of a C.E.S. Aggregator Function .................. 104
Calculating the TRI of Japan's Meat Imports .................. 111
CHAPTER 5
SUMMARY AND CONCLUSIONS ........................... 124
APPENDIX A
PROOF OF THE PROPOSITION ............................. 131
APPENDIX B
STOCHASTIC SPECIFICATION OF A C.E.S. AGGREGATOR FUNCTION 133
APPENDIX C
DESCRIPTION OF THE DATA AND THEIR LIMITATIONS ............ 142
REFERENCES ........................................ 159
BIOGRAPHICAL SKETCH ................................. 165
V




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 TRADE RESTRICTIVENESS OF JAPANESE AGRICULTURAL IMPORT POLICIES By
Christos J. Pantzios
December 1993
Chairperson: Timothy G. Taylor Major Department: Food and Resource Economics
This study is an attempt to re-evaluate existing measures of trade protection and empirically investigate new ones in search of measures that aggregate satisfactorily diverse sets of trade distortions and provide a valid estimation of trade restrictiveness across countries and over time. A review of the existing literature reveals that, from a methodological point of view, the Trade Restrictiveness Index (TRI), has some clear advantages over the rest of the trade protection measures.
The study implements the TRI by investigating the overall restrictiveness of a subset of Japan's agricultural policies. The level of trade protection is measured, ceteris paribus, by means of the TRI for the following farm products: (1) beef, (2) pork, (3) poultry, (4) wheat, (5) rice, and (6) fresh oranges. The computed index indicates a dramatic rise in trade protection in the early 1970s, followed by a slow process of easing
Vi




price and quantity restrictions so that a slight trade liberalization trend appears towards the end of the examined period.
This study also attempts a generalization of the standard TRI by deriving the index in the presence of different sets of traded goods over time. Under the maintained assumption of a C.E.S. welfare function, it is shown that when newly traded goods are introduced over time, the standard TRI is adjusted by a factor that depends on the relative expenditure value of the new goods, and the elasticity of substitution of the underlying C.E.S. welfare function. This generalized version of the TRI is empirically investigated by calculating a partial TRI pertaining to Japanese meat imports under the maintained hypothesis that beef imports are differentiated products, distinguished by origin (supplier country). The standard TRI is adjusted in 1969 to account for the introduction of U.S. beef in the Japanese meat market. This exercise shows that the TRI may be shaped by both policy changes and cross-commodity effects and as a result the adjustment in the generalized TRI may be either upwards or downwards.
Vii




CHAPTER I
PROBLEM STATEMENT AND OBJECTIVES OF THE STUDY
Introduction
The rapid growth of international trade during the last two decades has resulted in an increasing interdependence of the world's economies. The globalization of markets, however, has been often accompanied by heavy government intervention. This is especially true in agricultural trade where government intervention is the rule rather than the exception. In attempting to insulate their farm sectors from the variability of world markets, governments have devised national agricultural policies whose diversity and complexity obscures the global interdependence of agricultural markets.
Nowhere is this clearer than in the contrast between the agricultural policies of the European Community (EC) countries and the U.S. and the resulting declining export power of the US farm sector. To resolve this conflict, trade in agricultural products has been included for the first time in the Uruguay round of the General Agreement on Tariffs and Trade (GATr) negotiations. A major objective of the negotiations is to achieve a progressive reduction of government assistance to agriculture in developed countries.
Given the complexities and regulated environment of agricultural sectors, evaluation of the level of protection ( overall trade restrictiveness) poses a challenge for




2
both theorists and applied economists. In principle, a straightforward way to carry out such an evaluation is through an aggregate scalar measure (one that combines the various types of distortions and interventions). The resulting index would provide a useful and relatively simple means for international and intertemporal comparisons of trade protection.
However, the use of an such an aggregate measure may easily result in miscalculations of the degree of protection. In order to quantify the effects of diverse and often overlapping policies one must cope with a number of methodological and practical issues. Schwartz and Parker (1988) outline at least four criteria that an ideal aggregate measure of trade protection should satisfy. First, the measure should be a consistent aggregation across products, countries, and over time so that the resulting measurements can be meaningfully ranked and compared. Second, the measure should be relatively simple so that it is easily understood. Third, it should be flexible enough to capture the effects of diverse policies. Fourth, it should make the effects of trade distorting policies transparent, thus separating trade distortions from inefficiencies in resource allocation.
Measuring the magnitude of the distortions which government policies inflict on the trade flows of a country has long been a concern for economists. Over the years, researchers have developed various measures designed to quantify in a scalar measure the level of trade distortion. At the same time, there has been a continuous evolution of trade policies as global economic interdependence has progressed and the importance of the international marketplace was recognized.




3
In agriculture, one may observe two major evolutions related to trade policy tools. First, there has been a shift towards quantitative restrictions instead of tariffs, as the predominant means of protection. These range from quotas to various supply control schemes and marketing orders. Second, government assistance to the farm sector has taken numerous forms ranging from direct income transfers to structural policies such as research and extension expenditures. While some of these cause distortions to trade others cause distorting effects only to the domestic market. Thus any particular good may be affected by a number of different policies.
This evolution of trade policies and the inclusion of agricultural issues in the GATT negotiations have revitalized the interest of economists and trade negotiators in obtaining reliable indicators of trade protection. A comprehensive quantification of trade restrictiveness can provide a sound basis for trade negotiations and subsequently a consistent monitoring of the trade liberalization process. Thus an examination of the relevant existing measures on trade protection has been initiated in recent years, along with a search for new ones which can better cope with the new realities of agricultural protection.
Separately, the growth in international trade has revived the importance of the 'new goods-problem' as trade among the world's economies involves different sets of goods over time. Indeed, observed trade patterns suggest that new products play an increasingly important role in international trade. Product differentiation leads to the development of intra-industry trade and the importation of new varieties of goods across countries. It is also common practice for countries to change the trading status of goods




4
by liberalizing old goods and protecting newly developed ones; thus the list of goods subject to trade restrictions differs over time. In addition, regional or global trade agreements (such as the North American Free Trade Agreement (NAFTA) or a possible future GATT agreement on liberalizing trade in farm products) are expected to facilitate the importation of goods which were not traded earlier (or were traded in negligible quantities) due to trade impediments.
In the area of agricultural trade, the 'new goods problem' can be linked to the Armington assumption (1969) which is often used in empirical work. The Armington assumption states that imports and domestically produced goods are imperfect substitutes in consumption or production. Also exports are imperfect substitutes for domestically consumed goods. Such differentiation is rationalized on the grounds of packaging, taste, safety requirements etc. --something which is very relevant in the case of trade in certain agricultural products. As the liberalization process of a particular country develops, agricultural imports from countries trading for the first time in certain products with the country in question can be viewed as a special case of the 'new goods problem'.
Problem Statement and Objectives
The preceding discussion makes clear that the empirical measurement of the size of trade distortions remains a complicated, but timely task. The intent of this study is to contribute to the attempts being made to evaluate and generalize aggregate measures that satisfactorily combine diverse sets of trade distortions and provide valid estimation of the trade restrictiveness across countries and over time.




5
Among the available aggregate measures of protection, the Trade Restrictiveness Index (TRI) developed recently by Anderson and Neary (199 1) seems to be a particularly promising approach in evaluating trade distortions. This study reviews the existing measures of protection and subsequently focuses on the TRI approach in an attempt to examine empirically the performance of this newly developed protection index and furthermore investigate the effects of trading in new products on the measured level of trade protection.
The first objective of this study is an empirical implementation of the TRI approach. Given that the TRI is a recently developed concept in the measurement of trade protection, very few empirical constructions of the index exist (Anderson, 1991; Anderson and Bannister, 1991). In particular, this study applies the TRI approach in the agricultural sector of the Japanese economy in an attempt to
- subject the concept of the TRI to empirical testing and investigate its
applicability;
- compute a measurement of trade restrictiveness which has a robust theoretical derivation and explicitly incorporates considerations missing from other standard measures, notably both production and consumption side interventions, cross
commodity relationships, and quantity restrictions; and
- examine how various government interventions peculiar to agriculture can be
modelled and subsumed in one parameter-measures.




6
The second objective of this study is to generalize the TRI to account for the fact that different sets of goods may be traded over time, in an attempt to incorporate the 'new goods-problem' in the measurement of trade restrictiveness.
An overview of this study is as follows. Chapter 2 provides a review of the literature on trade protection measures with particular emphasis on the TRI, and then a comparison of the available measures. Chapter 3 presents an empirical application of the TRI in Japan's agricultural sector. In particular, Japan's government policies on beef, pork, poultry, wheat, rice, and fresh oranges during the period 1970-87 are considered, and the respective partial equilibrium TRI is constructed. The index aggregates the government programs in all -six commodities in a scalar measure and provides an estimation as how the overall level of protection on these commodities changed over the examined period.
Chapter 4 provides a discussion of the 'new goods' problem in index number theory and demonstrates how the standard TRI can be generalized to account for the presence of newly traded goods over time. Subsequently, it presents an empirical exercise on this generalization of the standard TRI. Specifically, a TRI pertaining only to Japan meat imports is constructed with the additional assumption that beef imports from the three major supplier countries (Australia, New Zealand, and the United States) are considered as different varieties of beef. Since U.S. beef imports were negligible in the 1960s, the meat TRI is computed for the period 1969-87, and adjusted in 1970 for the introduction of U.S. beef. Finally, conclusions and remarks are summarized in Chapter 5.




CHAPTER 2
MEASURES OF PROTECTION
The methodologies that have been used over the years to quantify the degree of trade protection reveals that the major protection measures can be classified into four groups: (a) the Average Tariff; (b) the Nominal and Effective Rate of Protection (NRP and ERP), and the Domestic Resource Cost (DRC); (c) the Producer/Consumer Subsidy Equivalent (PSE/CSE); and (d) the Trade Restrictiveness Index (TRI). A discussion of the merits and shortcomings of these measures is the objective of this chapter.
The Average Tariff Measuring the Height of the Tariff 'Wall'
Early attempts to measure the effect of protection on trade flows focused on tariffs (League of Nations 1927, Crawford 1934). In particular, researchers devised various methods to compute the overall height of the individual tariff levels (in the sense of a tariff 'wall') for the purpose of international comparisons across countries or intertemporal comparisons of a particular country. The procedures which have been generally employed involve the calculation of the percentage of all import duties to a
7




8
certain basis (e.g. the value of all imports) or equivalently, the calculation of weighted and unweighted averages of import duties.'
It was early recognized, however (Loveday 1929, Haberler 1936), that such computations have grave theoretical pitfalls and thus they can be quite misleading. When the height of the (aggregate) tariff is computed as the percentage of all duties collected over the value of total imports, one gets the rather absurd result that a more protective tariff regime yields a smaller percentage (i.e. a lower tariff 'wall'). This is because as duties become more protective, the value of total imports declines. In fact, if all duties became prohibitive, imports would be zero and the tariff 'wall' would be also zero. The same fallacy applies to the measurement of the tariff 'wall' as the percentage of imports subject to tariffs over all imports.
Calculations of a weighted average of all import duties faces the same grave objections. The basic difficulty is associated with choosing the proper weights of an average tariff rate. It is clear that weighing the various duties by the value shares of their own imports produces the same sort of distorting results. Low duties correspond to high relative levels of imports, thus they are given large weights; high duties are given small weights and prohibitive duties are given zero weights.
To circumvent these difficulties, various alternatives have been suggested (Llaberler 1936, Lerdau 1957, Balassa 1965). A possible alternative would be the share
'Unweighted averages are obtained by computing a simple average of the import duties upon total imports. Weighted averages are obtained by assigning different weights on the duties of individual imports. For an empirical application see, Research and Policy Committee of the Committee for Economic Development, 1964.




9
of the imported good in total exports. Others include the share of the imported good in the volume of the world trade or its share in the volume of production (or consumption) in one or more countries. Nonetheless, even the weights based on these shares can be biased given the volume of world trade, and domestic production and consumption are influenced by the import duties already in place. On the other hand, the calculation of unweighted averages of import duties fails to take into account the relative importance of individual imports.
Besides the aforementioned difficulties in computing a reliable average of all import duties, there is a great deal of ambiguity surrounding the very concept of 'the height of the tariff'. In other words, even if all calculation problems are overpassed, it is still not clear what is to be established or indicated by such a measurement. It is only clear that an estimate of the average tariff alone by no means provides an assessment about the degree of a country's total protection.
Several reasons are cited in the literature (Bieda. 1963, Towle 1956). First, there is a variety of other equally important measures upon which the protection of a country may rest (e.g. quantitative restrictions) that are not incorporated in the average tariff. Second, even if all trade restrictions were expressed in terms of a tariff, the true degree of a country's protection would still not be revealed. Different countries are expected to have different elasticity of demand for an imported good and different elasticity of domestic supply. Thus, any particular level of tariff will yield a less protective effect when the elasticity of domestic supply is low rather than high. In turn, elasticity of demand for imported goods depends on tastes, availability of close substitutes etc., while




10
elasticities of domestic supply depend on a country's resources and technological constraints.
Third, the degree of protection of final goods may well be affected by protectionist measures imposed on the respective inputs (raw materials and intermediate goods). Consequently, there is a clear difference between nominal and effective rates of tariffs wherein the latter include all duties levied on inputs. This last consideration led researchers to develop the concept of the Effective Rate of Protection (ERP) which examines net or effective rather than nominal tariff rates.
The Nominal and Effective Rate of Protection, and the Domestic Resource Cost
The idea that a distinction should be drawn between the nominal and the effective protection of an economic activity was first developed by Barber (1955) and further elaborated by Johnson (1960), Humphrey (1962) and Corden (1963, 1966, 1971, 1985) among several others. Uarge-scale empirical contributions of the concept of effective protection are given in Balassa (1965) and Basevi (1966).
The effective rate of protection deals with the true or net rate of protection associated with an economic activity which produces a final or value-added' product by using purchased material inputs that are themselves traded, and thus may be subject to distorting government policies. In this case the question becomes: Does the nominal protection on the final product indicate the true protection rate of the associated economic
2Value added is the value of the final output less the cost of purchased intermediate inputs.




11
activity, and if not, how is the protection of the final product affected by government policies on its intermediate inputs?
Formally, the ERP is defined as the percentage change in the value-added per unit (or effective price) of an economic activity with and without the existing government intervention. Thus, algebraically the ERP of the jth economic activity is given by the ratio
ERPJ = Vjo VjW (2.1)
where
V = the value-added per unit of the final product under the existing protective
structure,
vw1 the value-added per unit of the final product in the absence of any
distortions (i.e. under free trade).
The ERP so defined is based on the following assumptions: (i) the ratios of intermediate inputs to output (i.e. what is often called the physical input-output coefficients) are fixed' and identical for all firms, (ii) a small country-framework is adopted implying the elasticities of demand for all exports and supply for all imports are
I The fixed input-output coefficients assumption needs to be interpreted with caution as it does not mean that there is a fixed coefficient production process. The basic idea is that the various value-added production functions are functions of purchased intermediate inputs, as well as, primary factors such as, capital labor and so on. Then assumption (i) still permits substitution among primary factors along isoquants but rules out substitution among intermediate inputs or intermediate inputs and primary factors. It is shown that calculation of ERPs always tends to overstate the true effective rates, if there are indeed substitution relationships among the intermediate inputs or intermediate inputs and primary factors (Corden 1971, 1985).




12
infinite, (iii) all tradeable goods remain traded even after tariffs and other distortions have been imposed, so that the domestic market price of each importable is simply given by its international price plus tariff."
The algebraic formula for the ERP of the jth economic activity can be further elaborated. Consider first the simple case of a traded final product j, which has a single intermediate input i, subject to a tariff, in its value-added production function. Let:
aj= the physical input-output coefficient of the intermediate input i,
under free trade,
=the nominal tariff rate on the final product,
=the nominal tariff rate on the intermediate input,
pj= the world price of the final product.
Then by definition, it holds
Df=p( a1) (2.2)
vJ = Pj (1 + y~ -i ( + ti)] (2.3)
Combining equations (2.1), (2.2) and (2.3) yields
SRemoving this assumption implies that part of a tariff may be redundant. Since a redundant tariff has no effect of any kind, all calculations should then be based on the utilized part of the tariff-this could require detailed price data not always available.




13
.~~ alit, .4
ERPj = 0 ac ( 1a(a (2.4)
1-aij 1-aij 9 1-ai9)
Thus the ERP is a combination of two effects. The first, given by the term t/(la,) is the proportional increase in effective price (value-added per unit) v',, resulting from the nominal tariff on the final product. The second, given by [ag/(1-ag)]t is the proportional fall in the value-added per unit, resulting from the tariff on the intermediate input i. Clearly, an increase in the tariff rate imposed on input i reduces the rate of protection on the final product and vice versa. Furthermore, let
a'i = the physical input-output coefficient of intermediate input i, under
the existing distortions. Then
;1+tI
aj = a.j 1 + t (2.5)
and
1 asy
ERPj = 1 =
I + tj) (+ 3
tj t1a
(1 + tj) (1 + t) (2.6)
1, (2.6)
1 a
(1 + C,) (1 + te)




14
Equation (2.6) is a 'deflated' version of equation (2.4) expressing the ERPj in terms of the distorted input-output coefficient a'i instead of the undistorted (free trade) a~j. Its importance lies in the fact that it may be implemented more easily since inputoutput data are typically available in the presence of distortions.
Equation (2.4) can be readily extended to any number of inputs n, in which case n n
-j E aijF1 Ej a1 t1
ERP, 1 (2.7)
n n
that is, the tariff of the single intermediate input i is replaced by the weighted average of the tariffs on the individual inputs.
It must be noted that the ERP of a final product is not affected by any tariffs or other distortions imposed on the inputs of its intermediate inputs. In other words, one needs go only one step downward in the input-output structure. This is because for the producers of the jth final product only the cost of the inputs they themselves use, matters.
The ERP can be readily related to the much simpler concept of nominal rate of protection (or NRP). The NRP measures how nominal domestic prices for traded goods change in response to government policies and can be defined for either the producers (NRPp) or the consumers (NRP,).5
5 The NRP that applies to producers may differ from that of the consumers since trade policies often affect producer prices differently than consumer prices.




15
The NRP, measures the percentage difference (i.e. the 'wedge') between the domestic consumer price and the world (free-trade) price. Algebraically it is defined as the ratio
_PCI = pc Pi (2.8a)
pw
where pci is the domestic consumer price.
Correspondingly, the nominal rate of protection to producers (NRPP) measures the percentage difference between the domestic price received by the producers and the world (free-trade) price. Thus it is given by the ratio
D W
_ Pi Pi (2.8b)
Assuming ad valorem tariffs on final and intermediate goods, the ERPi in equation (2.4) can also be expressed as
n
NRPP- Eaij NRPi
ERPi = i=1 (2.9)
1 ai1
Thus the ERPi can be equivalently interpreted as a weighted average of the relevant nominal protection rates in the production of the final good j and the consumption of its intermediate inputs i =1... n.
Finally, the ERP bears also similarities with the concept of Domestic Resource Cost (or DRC). The DRC was developed by government planners in Israel during the 1950s as a means of project evaluation under conditions involving distorted official




16
exchange rates and distorted prices of tradeable goods (Bruno, 1972). In a general sense, the Domestic Resource Cost per unit of the ith economic activity, DRCi, is defined as the ratio
DRC' DCi (2. 10a)
NVAi
where the numerator, DCi, is the total value added of the domestic resources per unit of output employed in the ith activity, measured at opportunity cost. The denominator NVAi is the value added per unit of output of the ith activity, measured at world prices (i.e. its international value added).
Originally, the DRC method was used as a normative ex ante criterion of social comparative advantage, in ranking development projects. That is, the DRC criterion was used as an indicator of ranking future investments according to the real cost of net foreign exchange earned or saved. However, the DRC can also be viewed as a measure of the opportunity cost that a country incurs in order to sustain its existing import substitutes. In this sense, the DRC can be interpreted as an index of the social cost of trade protection and a means for evaluating the magnitude of trade distortions. From the definition of the DRC, it is clear that both the ERP and DRC involve valueadding activities in their measurement. A comparison between the two reveals that the ERP and DRC are in theory identical under only exceptional circumstances. In particular, it is shown (Krueger, 1972) that the following relationship is true DRCi = ERPi +1 (2. lOb)




17
under these stringent conditions: (1) all goods are traded (or tradable), (2) there are no transportation costs, (3) resources are perfectly mobile within the domestic economy but perfectly immobile internationally, and (4) all domestic output/input markets are perfectly competitive.
It follows that the ERP may be a sufficient measure for economies where tariffs are the predominant impediment to trade and factor markets are fairly competitive. For economies where, in addition to tariffs, there are quantitative restrictions, institutional constraints (e.g. government involvement in trade-related sectors), and market imperfections ( resulting in prices which do not reflect the true opportunity cost), the DRC has a conceptual advantage over the ERP. In these cases, evaluation of domestic resources at imputed shadow prices rather than market prices is a superior method to encompass as much as possible non-price distortions in measuring the cost of protection. This significant advantage comes, of course, at a higher information cost. Implementation of the DRC method requires besides input-output coefficients, shadow values for the domestic inputs which are usually estimated from programming models.
The Uniform Tariff Equivalent C=T).
Based on the concept of ERP, Corden (1966) defines a single aggregate measure of the various individual tariffs, which he calls the Uniform Tariff Equivalent (UTE). The UTE is defined as the uniform tariff which (if applied) would keep the value of imports at the same level as the existing (nonuniform) tariffs. In other words, the UTE




18
is the uniform tariff which is equivalent to the existing tariffs in its effect on the total value of imports.
By considering imports of two final goods and linear demand curves, Corden shows that the UTE is a weighted average of the ERPs on these two final goods; the respective weights are combinations of their demand elasticities, import values under free trade and input-output coefficients.
The UTE is then compared with average tariffs, calculated as weighted averages of all individual tariffs by using as weights either actual imports in the presence of distortions or domestic output. The diversion between the UTE and the average tariff is examined under various conditions on demand elasticities and input-output coefficients. These exercises provide a useful insight on the important factors ignored by the average tariff-approach mainly, demand elasticities, traded inputs and the reference point of the comparison. An additional empirical application of the concept of the UTE appears in Balassa (1965).
The Producer (Consumer) Subsidy Equivalent (PSE and CSE)
Developed by Josling (FAO, 1973; FAO, 1975), the notions of the producer (consumer) subsidy equivalent (or PSE/CSE approach) are designed to provide an aggregate measure of government support policies in a particular sector of the economy.
The concept of the PSE is straightforward. It is the subsidy that would be necessary to replace all current government policies applied to the agricultural sector of a particular country in order to leave the producer's income unchanged. Thus the PSE




19
measures total income transfer resulting from any policy that can be linked to incomes and it can be computed at any level of government--local, regional or national. The consumer subsidy equivalent (CSE) is defined in a symmetric fashion.
Several features of the PSE deserve particular attention. Unlike the measures mentioned above, the PSE combines both price and nonprice policies (ranging from import quotas to direct payments to farmers, disaster payments and so on). Like the previous measures, however, the PSE is commodity-specific, evaluated as the absolute sum of money received as support by the producers of that commodity. Then it can be expressed in relation to several bases:
(i) PSE per unit of output (i.e. PSE/volume of quantity produced)
(ii) PSE as a percentage of domestic production valued at domestic prices
(iii) PSE as a percentage of domestic production valued at world prices (iv) PSE as a percentage of actual net farm income (thus serving as an
indicator of income dependency).
The definition of PSEs and CSEs is inherently flexible. One may decide to include or exclude various government programs. Thus, while the first calculations of PSEs and CSEs included only commodity specific policies, a recent OECD study (OECD, 1987) broadened the policies covered in the PSE to include structural support programs, which are not necessarily commodity specific (e.g. research and extension). Additionally, the Economic Research Service (ERS) of the U.S. Department of Agriculture (USDA) has extended the OECD measurement of the PSE to include the effects of exchange rate distortions in the case of developing countries (USDA ERS, 1988).




20
Concerning the estimation of PSEs (and CSEs), it must be emphasized that different countries have quite different sets of policies in their agricultural sectors. Thus a standard framework is needed for comparisons across countries and across commodities6.
Several important points must be noted on the use and the interpretation of PSEs and CSEs. By aggregating a variety of government policies into a single measure, the PSE and CSE allow comparisons of government support across countries, commodity markets and types of policies. Additionally, they can indicate which forms of government support are most important in different countries and, when examined over time, they can indicate the intertemporal changes of government support.
As mentioned earlier, international comparisons of the PSE and CSE require a standard procedure and a common set of the policies covered. To the extent that policies excluded from this common framework provide a significant amount of support, the resulting comparisons may well be biased. Additionally, in intertemporal comparisons a country may change its government policy profile towards (or away from) the set of policies covered by the PSE (CSE). Therefore interpreting comparisons of PSEs (CSEs) requires consideration of these issues.
It is erroneous to conclude that, if all governmental programs were removed, incomes would decline by the value of the transfers estimated by the PSE. Income in
6 In addition to estimating PSEs and CSEs for each commodity, 'pooled' PSEs and CSEs are also calculated and compared across countries (USDA ERS, 1987). Theses are weighted averages of the individual PSEs weighted by each commodity's share with the total production value of the covered commodities.




21
the absence of the existing government programs would depend on the new levels of prices, production, consumption and trade. PSEs only measure the total transfer to producers under current policy and market conditions.
The PSE and CSE measures do not capture the large-country effects on the world prices (i.e. they are implicitly based on a small-country framework). Additionally, PSEs and CSEs do not reveal the distributional effect of government programs within a particular sector (e.g. PSEs cannot show whether total transfers in the dairy industry are received equally by all dairy producers).
Changes in the PSEs and CSEs can be merely due to changes in the world reference prices or controlled exchange rates--both of which may be quite variable over time. Almost all traded goods are priced in U.S. dollars; thus when the dollar appreciates, the world reference price observed by countries other than the U.S. rises (and vice versa). Then for countries supporting the producer price above the world price, the price 'wedge' is now narrower than it would have been under constant U.S. exchange rates (unless their agricultural policies are responsive to world prices). Therefore changes in the magnitude of the PSE (CSE) can be markedly affected by controlled fluctuations in exchange rates.
The PSE does not directly measure the effect of supply control policies (e.g. uncompensated acreage reduction programs). An effective supply-control program




22
reduces production from what it would have been in the absence of such program; therefore total government transfers (and thus the PSE) are lower under the program.7
Finally and most importantly, government policies which yield the same PSE/CSE across countries do not necessarily imply that these countries have the same degree of trade distortions. Different types of government policies produce different trade effects,' while the effect of government programs on trade depends on the country's share in international trade. Additionally, producers and consumers across countries respond differently to the same type of government intervention due to political and social factors, market characteristics, and resource and technology constraints.
PSE and Trade Distortions
Given the fact that the link between government support policies and the impact of that support on trade can be very weak,9 the use of the PSE/CSE as a yardstick of trade liberalization has come under criticism. It is argued that if the PSE/CSE is used in trade negotiations, countries may try to achieve lower levels of PSEs (CSEs) by
7 However, supply control programs may be implicitly taken into account in the PSE/CSE measures. For example, if a country is large enough to affect world prices, its supply-control programs will raise world prices. The price-effect of the supplycontrol program is then captured in the PSE/CSE calculations if the corresponding price data are used. Tangermann et al (1987) propose practical ways for countries employing supply-control programs to receive negotiating 'credits' on the PSEs for the commodities concerned.
' For example, deficiency payments stimulate domestic production but leave consumption unaffected; import quotas also raise domestic production but, at the same time, reduce demand.
9 An illustrating diagrammatic exposition of the differences between support measures and measures of trade distortion appears in Roningen and Dixit (1991).




23
merely eliminating government programs which bear little relationship to trade, thus leaving their trade barriers intact.
To cope with this deficiency, economists have suggested versions of the PSE including only policies which are likely to affect trade. Thus Schwartz and Parker (1988) proposed the modified PSE as a more appropriate measure of trade distortion. The modified PSE includes only policies with well-defined price effects which are more likely to reflect on trade; it excludes policies with ambiguous price effects such as structural programs.
Rondingen and Dixit (1991) proposed a more direct measure of trade distortion, termed the Trade Distorted by Support-Index (or TDS). The TDS index is a volume measure of distorted trade. In particular, it measures the change in the volume of the existing net trade, if a country eliminates completely all support to a commodity. The TDS in volume terms is generally expressed as
TDS1 = [qse~s, qdedsml + [qae~sp qdeds] + qsesi sso =
= [domestic market support] + [direct payments to farmers/consumers] + [other
farmer support] [offsets to support] (2.11)
where
e., ed = own-price supply and demand elasticities, respectively,
10 However, the TDS can be readily expressed in percentage form (i.e. TDS/[volume of production (consumption)]) or in value form (VTDS) by multiplying the TDS by the world (border) price.




24
q,, qd = observed production and consumption quantities, respectively,
s. = market support ratio (i.e. the support level per unit of the
commodity compared with domestic prices),
so, s, = direct support rate to farmers and consumers, respectively,
si = support ratio for all other types of assistance to producers, and
sso = set-aside offset policies, usually policies that require production (consumption) discipline for farmers (consumers) to be eligible for direct
payments.
The primary contribution of the modified PSE and the TDS is that they are more transparent measures of the trade distorting effects of government policies than the PSE. Methodologically however they share a lot in common with the PSE/CSE approach. Hence they may be viewed as 'trade-oriented' versions of the PSE".
The Trade Restrictiveness Index (TRI)
Unlike the previous protection measures which lack a clear theoretical basis, the TRI (Anderson and Neary, 1991a; 1991b; 1992) is explicitly based on the underlying economic structure of a trading economy.
Conceptually, the TRI is a general equilibrium distance function measure (Debreu, 1951; Deaton, 1979) for a trading economy. Conventionally the distance function is defined with some reference utility level u and an arbitrary quantity bundle
" The modified PSE is literally a trade version of the PSE; the TDS can be viewed as a volume-(trade) version of the PSE.




25
q as the factor by which the bundle q must be deflated (or inflated) in order to attain exactly utility u. Thus the distance function is a measure of the inefficiency of q relative to U.
The TRI utilizes the same notion in the space of trade distortions; in the case of only tariffs this space is the space of the domestic prices of the traded goods, while in the case of only quotas, it is the quantity space of the traded goods. Moreover, instead of a utility function, the starting point in the case of the TRI is the (general equilibrium) budget constraint of a trading economy. Such a function equates aggregate expenditure on traded goods to a gross revenue or GNP function and trade revenues (tariffs and quota rents). This permits the aggregate utility (welfare) level and the instruments of trade distortion (i.e. tariffs,quotas) to be subsumed conveniently in a single function called the balance-of trade function.
Based on the economy's balance-of-trade function, the TRI is defined as a compensating variation measure of the trade distortions towards a reference utility (welfare) level. Specifically, given a final period 1 and an initial period 0 the TRI is defined as the uniform rate by which the instruments of trade distortion (tariff rates, quota levels) must be deflated (or inflated) in order to take the economy back to the initial welfare level. Similar to the distance function therefore, the TRI can be viewed as a measure of the trade inefficiency associated with a set of trade distortions, relative to a reference welfare level.




26
Operationalization of the TRI
The TRI stems from the same theoretical reasoning as the cost-of-living-index or Consumer Price Index (CPI). To demonstrate this similarity a short digression on the concept of the CPI will first be presented. The CPI is the uniform rate of change in all consumer prices, which produces an equivalent rise in the expenditure required to maintain the same level of utility for a representative consumer. Formally the CPI is based on the consumer's expenditure function e(p,u) where p is the vector of consumer prices, u is the consumer's utility level and e(-) is the minimum level of income required to achieve the utility level u when the consumer faces prices p. Since e(p,u) is an optimal value-function, its gradient with respect to p gives the vector of the consumer's demands, X.
Given arbitrary changes in prices p, the level of income required to support utility u is X'Vp, where Vp is vector of first order partial derivatives (i.e. the gradient vector) of the price vector, p. If the price vector p were proportional to its initial level by some factor at (i.e. p = poct where p0 is the initial price vector) then the level of income required to support utility u, given a change in the prices p, is X'p~da. Then the uniform rate of change in all consumer prices which creates the same rise in the required expenditure as the arbitrary change Vp is
X-p-da X'Vp = X/P __ (2.12)
where a hat (^) over a variable denotes a proportionate change, i.e. Vp/p. For an initial value of ot = 1, dci is also a percentage change and represents the CPI.




27
At this point it must be stressed that the preceding formula of the CPI (bearing the opposite algebraic sign) can be obtained as a solution to the following problem: what is the (hypothetical) uniform reduction in consumer prices (instead of income transfer) which compensates the consumer (i.e. preserves initial utility u) given some arbitrary change in prices dp. As earlier
X/'pd0'-'Vp (2.13)
where the minus sign denotes the direction of the compensatory change in prices p.
The notion of the TRI is rooted in the same reasoning as the compensatory price change dot'. This is however where the similarities between a consumer and a trading economy end. The budget constraint of a price taking consumer links his expenditure to his monetary income. By contrast the budget constraint of a trading economy links the aggregate expenditure on all goods (domestic and imported) to the value of the national product (GDP) and the government revenues resulting from trade intervention (i.e. tariff revenues and quota rents). Thus a change in the tariff vector causes not only a change in aggregate expenditure but also a change in government tariff revenue.
Consequently, the starting point in the derivation of the TRI is the economy's budget constraint. The specification of this budget constraint can be conveniently carried out using standard dual techniques (e.g. Dixit and Norman, 1980). In particular, using dual functions the consumption and the production sectors of a trading economy can be modelled in terms of an aggregate expenditure function e(T,u) and a revenue (or GNP) function g(ir,v), respectively (where ir denotes the price vector of traded goods, u




28
aggregate utility or welfare, and v the vector of resource endowments). Given the fundamental requirement that imports must be paid for (i.e. trade must balance), the budget constraint of a trading economy becomes e Or, u) g(r, v) = b (2.14)
where b denotes the value of net imports; a positive b implies a trade surplus while a negative b implies trade deficit; for the purpose of this analysis, b is considered to be exogenous. Moreover, the above equilibrium condition can be re-written in terms of the function
B (nr,u, v) = e (nt,u) g~n, v) 3=0 (2.15)
which is termed the balance-of-trade function. Essentially, this function represents the overall equilibrium condition of the economy in each time period and can be expressed as
B(n, u; z) = 0 (2.16)
where z represents all exogenous variables, such as factor endowments, international prices of traded goods, tastes, etc.
The balance of trade function BO) can be specified to take into account the fact that trading economies impose price restrictions, or quantity constraints, or both on their imports (exports). Below, the balance of trade function (and subsequently the TRI) is defined with respect to an economy imposing both tariffs and quotas (the cases of only




29
imports or only quotas may be considered as special cases and can be found in Anderson (1991)).
The TRI in the Presence of Tariffs and Ouotas
Consider a competitive trading economy imposing both tariff and quota restrictions, wherein all goods are tradeable (non-traded goods can be considered in the background as quota-constrained goods whose quota levels are set equal to zero). The vector of goods subject to tariffs is denoted by m and the respective price vector by r. Quota levels are denoted by Q and the price vector of quota-constrained goods is p. Finally, the world prices for the tariff-ridden goods and quota-constrained goods are p* and r*, respectively.
The economy under examination is assumed to be small (price-taking) so that p* and r* are exogenous to the analysis. To facilitate the analysis, one can define
E(p,nU M= mino,m [p1Q + n'm I U(Q,m) = U] (2.17)
as the trade expenditure function E(p, w,U). In words, E() is the expenditure of the aggregate consumer in the traded quantities (net imports) of all goods and has the standard properties of an expenditure function: it is linearly homogeneous and concave in (p, w) and its derivatives with respect to prices are the economy's excess demand functions i.e.




30
E.(n,p,u) = m(n,p,u) E= Q(,p,u) (2.17a)
However, assuming that the quotas are binding the aggregate consumer can be regarded as minimizing expenditure over the tariff-ridden goods only. Thus optimizing behavior can be described equally well by the distorted trade expenditure function, defined as
E(Q,nM= minm [n'm I U(Q,m) = u] (2.18)
Alternatively one may write
fE(Q, n, M = maxp [E(p, n, u) plQ] (2.19)
and its first derivatives yield
EQ, n,u) = E. [p(Q,n, u) n,U] = = m[P (e,7rU) ,n, U(2.20a) E0(Q,7C,u) = -p(Q,,r,u) (2.20b)
The derivative (2.20a) follows from Shephard's Lemma since the distorted expenditure function is the minimum-value function for tariff-ridden goods. The derivative (2.20b) implies that relaxing the quota by one unit reduces the expenditure of




31
tariff-ridden goods by -p. Thus it gives the consumers' marginal willingness to pay for quota-restrained goods.
Turning to the quota rent(s) (p p*)'Q, it may be recognized that usually these rents are shared between the home country and its trading partners. 12 Let CO denote the fraction of the quota rents accruing to foreigners (supplying countries); then the quota rent retained in the home country is (1 w)(p -p*)'Q.
The budget constraint of such an economy (assuming away any trade balancesurplus or deficit) in domestic prices, is given by ROW,p,Uin + p'Q = C'M + (1-(a) (p -p*) 'Q (2.21)
In words, the left-hand side equals the expenditure of the aggregate consumer in all goods. In equilibrium this must equal the revenues from the trade restrictions retained at home. These are given by the right-hand side.
More conveniently this equilibrium condition can summarized in terms of the balance-of-trade function, defined as
12 The fraction wo may be interpreted as the fraction of quota licenses awarded to foreign importers. A common practice in international trade is to impose tariffs on goods already constrained by quotas. In this case wa is also equal to the tariff rate imposed on quota controlled goods, under the assumption that all imports are handled by foreign importers.




32
B(Q, n,u;y) : (Q, n,u) + p'Q
- tm (1-() (p-p*)IQ = 0 (2.21a)
As noted earlier B(') summarizes the equilibrium condition of the home economy. It may also be interpreted as the net foreign exchange requirement (in the sense of total income), necessary to support the utility level u for the aggregate consumer facing given levels of tariffs or quotas. Consider now a relaxation in both trade restrictions of this economy (i.e. the tariffs rates and the quota levels). Then the compensating income transfer required to move the aggregate consumer back to the initial utility u, is
-Bxdn + Bodo (2.22)
Assume now that the prices of tariff-ridden goods ir, and the quota levels Q are proportional to their initial levels at the same proportionality factor 0 (i.e. Q = Q'P, w = ', where i-, Q0 are the initial levels of prices and quotas respectively). Then the compensating income transfer, described in (2.22) is written + B QOdP (2.23)
By setting (2.22) equal to (2.23), one can solve for the uniform proportional change in tariffs and quotas which offsets the arbitrary change in (-,Q) in the sense that it returns the aggregate utility (welfare) to the initial level u. Thus:




33
B dQ B_____-Bd (2.24)
Note that for the initial value ( = 1, do is also a percentage change (dfl/3). The above expression is a scalar measure of the change in the overall trade restrictiveness of the economy as a result of any arbitrary change in trade policies. In particular, the percentage change do has the interpretation of an index number of trade distortion: it is the unifonn proportional change in trade distortions which returns the economy back to its initial welfare level, given some arbitrary change in these trade distortions. This index is due to Anderson and Neary (1991) and it is termed 'Trade Restrictiveness Index' (henceforward TRI).
It may be noted that the derivatives -B, and -BQ can be interpreted as the marginal cost of tariff-increases and the shadow price of quotas, respectively. Then the welfare costs associated with tariffs and quotas are -B.QQ and B',r, respectively Accordingly, the term (B'QQ B'1 r) equals the negative of the total welfare cost of the initial trade structure, and it is termed the 'shadow value of distorted trade' (Anderson and Neary, 1991).
As seen in (2.24), the TRI equals the sum of the (percentage) changes in all trade restrictions (r,Q) where each component (iri and Qj) is weighted by its contribution to the total welfare cost of the initial trade structure. As mentioned above, the balance of trade function can be interpreted as giving the net foreign exchange required to support a certain welfare level u. Thus the TRI can also be thought of as measuring the uniform




34
change in trade distortions which yields an offsetting change in the foreign exchange required to maintain the initial welfare.
A more formal definition of the TRI is given below:
P(n, Q, u) a[13: B71Pi, uO; z) =0] (2.25)
In words, the TRI is the uniform rate # by which the domestic prices of the tariffridden goods must be raised, and the levels of the quota-controlled goods must be reduced (tightened) so that the economy returns to the initial protective regime consistent with welfare level u'. If the trade policies (-,Q) do not change between periods 0 and 1, 0 = 1. As the prices of tariff-ridden goods get lower and quota levels expand, the rate of change d(/3 of the TRI (and subsequently its magnitude) rises. Thus, an increase in the size of the TRI indicates that the economy moves towards free trade.
Totally differentiating (2.25) for the reference welfare level u0, yields:
B.,(Vn'0+n-do) + BQ(VQ"3-Qdo) =0 (2.26)
Converting to proportional changes yields
M N
E> (B Q) Q j (BI 7t ) ft j (2.27)
P O Q BI 7 B Q
where Bi = aB/a71, Bi = aB/aQj, and a hat (A) over a variable denotes a proportionate change, i.e. dQj/Qj, drih/-r.
If distortions (ir,Q) decrease in period 1, then returning the economy to the initial protective regime associated with welfare uO implies lower quotas (so that dQ/Q is




35
positive) but higher tariffs and consequently higher domestic prices (so that dr/r is negative). In that case, (2.27) may be also written as
M N
S(B Qj) Q., Bn)I (2.28)
dp ___ I
P BO -Bx BO -B B.
Turning to the derivatives B,, BQ differentiation of B(*) in (2.21a) with respect to r, Q (and recalling that m, p are themselves functions of (r,Q,u) from (2.20) yields B,' = t 'm, + w Q'p, (2.29)
(2.30)
B1 = t'm + WQ'PO (1-W) (p-p)' Moreover, Anderson and Neary (1992) show that the derivatives of the import demand functions m(r,Q,u) and the inverse demand functions p(r,Q,u) can be expressed in terms of either the distorted or the standard trade expenditure function. Taking into account equations (2.17a) and (2.20) one may show that the following equations hold ~ -1E (2.31)
MO = E40 = E'PEP-P M. = f ., = E., E, p' EP, (.1
po-g =o=Epp' p =-2o =-E' E. (2.32)
The derivative matrix E-' is generally expected to be negative definite.
Finally, it may be noted that the TRI can be also extended to incorporate the effect on trade protection from policies applied to: (i) non traded goods and (ii) distortions in factor markets (Anderson and Bannister, 1991; Anderson and Neary 1992b). In




36
particular, consider the case where some goods are not traded and their price vector is denoted by s. The TRI in this case is defined as
(A, SQ, u) Bz)= (2.33)
Note that the index is still defined over (r,Q) only; however the price vector s is an additional variable in the analysis as it reflects the policies which may be applied on the non traded goods. The TRI can now be derived by total differentiation of (2.33) with respect to i-, Q, and s
M N
P(B',Q B.') (BQQ B~n)
Blds
- S(2.34)
It may be noted that a new complication arises here as the level of the TRI appears in the third denominator of the right-hand side. This difficulty may be overcome by multiplying (2.34) by f and transforming it into a first-order differential equation in /, i.e.
= D + D2 (2.35)
dt
where D, represents the first two terms on the right-hand side of (2.34) and D2 represents the third term. Solving for 0 yields




37
1t = [pt_' LeD~t- D (2.36)
L~1 D2jT
However in empirical applications of the index with discrete time-data, (2.35) can be approximated as simple difference-equation and be solved as
0 = [(1 + D) t-1 + D'2] (2.37)
This formula can be applied to each period t2,t3,... with the normalization that i3 =1 in the initial period t1. The introduction of distortions in factor markets can be treated in an analogous fashion (Anderson and Neary, 1992).
Comparisons Among the Measures of Protection
In attempting comparisons among these protection measures, one should bear in mind that they do not include the same set of policies in their definitions. Additionally, some of these policies cause trade distortions while others do not.
The focus of the PSE/CSE approach on income-transfers makes it the least transparent measure of trade distortion. The ERP, by including government policies on intermediate input-prices, is a superior indicator of government intervention over the NRPp which captures only border measures and measures taken at the level of the final products; quantity restrictions nonetheless, are considered only implicitly in the ERP. The NRPC on the other hand, provides a good analytical measure of trade distortions for consumers.
The PSE/CSE measure is the most flexible as it covers a wide range of government policies. However its flexibility opens the possibility for policy changes




38
which reduce the size of the PSE (CSE) without affecting the existing trade distortions. Similarly, the NRPP, by focusing only on policies applied on the final goods, opens up loopholes though which trade distortions are manipulated while the level of NRPp remains unaffected (e.g. by applying government intervention on inputs). The PSE does not include the effects of government policies on intermediate product prices. By contrast, the ERP by explicitly including policies affecting intermediate inputs, is a more reliable indicator in the sense that certain targets of ERP can be negotiated and attained via either input or output (final good) policies. Moreover, the DRC is superior to the ERP in capturing, at least implicitly, nonprice trade distortions.
In terms of calculation, the PSE (CSE) and the modified PSE are most easily calculated by adding up budget expenditures on income transfers. The ERP, on the other hand, requires explicit estimates of input-output coefficients. The DRC can be a more accurate measure than the ERP; it requires however estimates on both input-output coefficients and shadow prices for the domestic factors. The TDS and modified PSE are most appropriate analytical measures of trade distortion in production; however calculation of the TDS requires data on supply, demand elasticities.
Apart from the above mentioned differences, the ERP (NRP) approach, the DRC and the CSE/PSE approach share a number of common methodological traits. First, they all are partial equilibrium measures, meaning that the prices of non-traded goods and goods in other (e.g. non-agricultural) sectors are held constant. Cross commodity substitution effects for both producers and consumers are ignored. The small-country framework is maintained thus ignoring the (potentially) important effect government




39
policies may cause on world prices. Supply-control programs and income stabilization programs are not modelled explicitly in any of the measures (the TDS is an exception by including set-aside policies, while the DRC may implicitly capture nonprice distortions via imputed shadow prices).
Last but not least, none of the aforementioned measures has any theoretical foundation. From a methodological point of view, this may be their most serious drawback. Aggregate measures should summarize the changes of their individual components in the sense of an index number. The change in value of an index number is generally a weighted average of changes in the components of the index while the respective weights are explicitly derived from the economic structure associated with the index. This makes the weights consistent with the underlying economic theory and the resulting index meaningful.
The TRI approach remedies some (but not all) of the above shortcomings. First, the TRI is defined in theory as a general equilibrium measure. Second, it is explicitly derived from the underlying economic structure gaining theoretical consistency. Third, the TRI incorporates explicitly both the consumption and production sector of the economy and also allows cross-commodity relationships (cross price elasticities are explicitly included in the formula that computes the TRI); this permits a theoretically consistent aggregation over commodities and (or) sectors. Fourth, by specifically incorporating expenditure and revenue functions in its definition, the TRI can potentially allow the explicit modelling of supply control policies (e.g uncompensated acreage




40
reduction or production quotas). Fifth, the effect of purely domestic policies can be included in the definition of the TRI thus enhancing its comprehensiveness.
Nonetheless, the TRI approach still maintains the small country framework and has rather high information requirements (price elasticities are needed in its computation). Additionally, in its practical implementations, the TRI is still likely to be specified at a partial-equilibrium context unless the researcher has access to a complete general equilibrium model. Furthermore, the use of elasticities has been criticized as a practical means of trade negotiations. With respect to that, it must be stressed that elasticities can be viewed as weights that differentiate the importance of the various commodities included in the analysis and thus they can easily be interpreted even by non experts (Roningen and Dixit, 1991). Most importantly, the price elasticities in the formula of the TRI can be viewed as a theory-based means of aggregating the degree of trade protection across the commodities under examination.




CHAPTER 3
THE TRI APPROACH IN JAPAN'S AGRICULTURE
This chapter presents an empirical implementation of the TRI. The focus of this application is the farm sector of Japan and the respective policies affecting the trade flows of agricultural products. Japan provides a typical example of a country that protects the domestic agricultural sector predominantly by tight border measures along with a wide array of support programs within the farm sector itself. Indeed in recent years Japan, along with the European Economic Community (EC), have been charged with being two major world markets fairly closed to international trade in farm goods. In the subsequent sections the TRI pertaining to a subset of Japan's farm sector is first specified, the respective information requirements are then empirically estimated, and the index is computed for the period 1971-87.
Japan's Agricultural Trade Policies
With relatively poor endowments of natural resources and arable land, Japan is one of the world's largest net importers of farm products (OECD, 1987). It is also the largest and one of the most stable overseas markets for U.S. agricultural products. The U.S. consistently accounts for a significant share of the Japanese principal farm imports.
41




42
Despite the reliance on imported foodstuffs, the agricultural policies that Japan pursued in the 1970s and 1980s (as prescribed by the Agricultural Basic Law, enacted in 1961) have been policies of self-sufficiency. As a result, in the recent years Japan has persistently kept its agricultural markets closed to international trade.
To achieve import control, the government has used a complex intervention system comprised of a variety of policy instruments, administrative measures, and implementing institutions. These include conventional trade measures such as import tariffs and quantitative restrictions (import quotas), but also state trading and trade controlled indirectly by the government. In particular, trade in certain products (wheat, rice) is carried out directly by the Food Agency of the Japanese government. In other cases (livestock, sugar), trade is carried out by parastatal agencies the Livestock Industry Promotion Corporation (LIPC), and the Japan Raw Silk and Sugar Price Stabilization Corporation (JRSSPSC), respectively. Participation of private traders in external trade is typically subject to strict licensing procedures and administrative guidance (Fitchett, 1988).
Due to insufficient information on certain policies and lack of data, this study does not provide a comprehensive representation of the trade policies and administrative programs in Japan's agriculture. In terms of import value the examined commodities, represent 20% to 30% of the value of Japanese imports in food and live animals during 1971-87. Nonetheless, this study investigates a subset of agricultural policies which comprises the most well known trade restrictions and have been most often criticized by Japan's trading partners (such as the beef and citrus quotas or the ban on rice imports).




43
In particular, this study measures (ceteris paribus) the level of protection by means of the TRI in the following farm products: (1) beef, (2) pork, (3) poultry, (4) wheat, (5) rice, and (6) fresh oranges. An outline of the domestic marketing policies and related border measures pertaining to these products during the period 1970-1987, (Australian Bureau of Agricultural and Resource Economics, 1988; Fitchett, 1988; OECD, 1987) is presented below and summarized in Table 3.1.
The Beef Industry
Although the beef industry is a minor sector in Japan's agriculture (in terms of production value), it attracts special attention because it is one of the industries the government wishes to expand and, at the same time, enjoys the support of the politically powerful cooperative movement (Australian Bureau of Agricultural and Resource Economics, 1988; Longworth, 1983).
For the period under examination13 (1970 -1987), Japan's beef policy included two major instruments: a price stabilization scheme, and import quotas as a means to achieve the price stabilization objective. The implementation of these policies has been assigned to a quasi-governmental agency known as the Livestock Industry Promotion Corporation (LIPC).
The basic mechanism behind the beef price stabilization scheme is the following: every year the government, in consultation with the LIPC and other bodies (consumer groups, unions, etc.) determines a price band (i.e. a range between an "upper" and a
13 In 1988, Japan agreedto abolish its beef quotas by April 1, 1991.




44
Table 3. 1: Japan's agricultural policies, 1970-87.
COMMODITY TYPE OF TRADE POLICY STATUTORY
SUPPORT BODY
Pork wholesale floor variable levy on L.I.P.C.
and ceiling imports
price
Chicken private price 20% tariff on
stabilization imports
band
Beef stabilization (i) import quotas L.I.P.C.
price band (ii) 25% tariff
Wheat Government State trading Food Agency
sets purchasing
and resale
price
Rice Government State trading Food Agency
sets purchasing
and resale
price
Oranges -(i) import quotas
(ii) 20% to 40%
tariff
"base" stabilization price). This price band is set for representative beef categories (steer, wagyu herd, etc.) in the representative wholesale markets, Tokyo and Osaka. In addition, the government determines (usually twice a year) an import quota for beef; the quota is "global", in the sense that it is not directed towards any specific country. Suppliers however must be able to meet Japanese quarantine requirements.
With the beef price band and the quota levels predetermined, the LIPC intervenes in the market to keep the beef wholesale prices within the stabilization band. It does so




45
by buying and storing beef when prices fall below the price band and releasing beef from its stock when prices move above the price band. The LIPC may stock and subsequently release both domestically produced and imported beef. In practice however, the LIPC manipulates the beef market by regulating the flow of imports via the quota levels (Australian Bureau of Agricultural and Resource Economics, 1988). The LIPC buys imported beef from licensed traders by competitive tender and releases it into the wholesale domestic markets by auction.
In addition to quota restrictions, Japanese beef imports are subject to a 25% ad valorem tariff. The associated tariff revenue is collected by the government (Ministry of Finance) and used for the development and assistance of the domestic livestock industry. More detailed presentations of the beef stabilization regime can be found in Australian Bureau of Agricultural and Resource Economics (1988), OECD (1987), and Longworth (1983).
The Pork and Poultry Industries
The government's support policies for the Japanese pork industry are similar to those on beef; in the case of pork however, the primary policy instruments are priceridden rather than direct control over imports. As with beef, the government sets a stabilization price band for each fiscal year and assigns to the LIEPC the role of supervising the pork market (i.e. absorbing from or releasing into the market the necessary quantities of pork and controlling imports).




46
However, in 1971 quantitative import control on pork was abolished. Since then, pork imports have been left to private traders (licensed and approved by the LIPQ, while the LIPC keeps the price of -pork within the stabilization price band by means of a flexible tariff system. In particular, pork imports are subject to the higher of either a 5 % ad valorem tariff or a differential duty. This differential duty is applied when the import price is lower than the central price of the stabilization band and is defined as the difference between the central price of the stabilization band and the import price.
The key aspect of this tariff system is that the price of imported pork is always equal to or higher than the central price of the stabilization band. Additionally the tariff is equal to at least 5 % (or higher) of the world price. A rather interesting side-effect of this mechanism is the preference of Japanese importers for higher quality (and higher price) pork cuts (loins, bucks, etc.) so as to subject their imports to the 5 % duty rather than the differential duty (OECD, 1987).
In the poultry sector, direct government assistance is minimal. There are no conventional intervention forms of price stabilization or quantitative import restrictions. A price stabilization fund for broilers was set up in 1970 by the National sales Federation of Agricultural Cooperatives and the national Purchase Federation of Agricultural Cooperatives, without government participation; the government nevertheless assists poultry producers through research, and disease prevention programs (USDA, 1983). Additionally, poultry imports are subject to an ad valorem tariff with base rate of 20%.




47
The Rice Industry
For cultural, historical, and religious reasons, rice is the most significant agricultural crop in Japan. Rice farmers are known to be one of the most influential groups in Japanese politics. As a consequence, Japanese governments have intervened extensively in the rice market over the years influencing its production, marketing, and trade.
Today the rice policy regime in Japan is a comprehensive policy mix comprising
(i) domestic supply control measures, (ii) state trading (import control), and (iii) pricing determined by the government. The government administers its programs in the rice sector through a governmental body, the Food Agency. The Food Agency buys rice directly from producers and sells it to wholesalers. Every year the government decides both the purchase price (price at which the Food Agency buys from rice producers) and the resale price (price at which the Food Agency sells to wholesalers). For most of the period under examination, the average purchase price has been set higher than the resale price generating considerable government deficits.
In addition to regulating the purchase and resale price for rice, the government also controls international trade on rice. Foreign trade in rice is carried out exclusively by the Food Agency; importers must apply for government permission and sell all rice imported to the government (Food Agency). As a result the domestic rice market is effectively insulated from the world markets. Japan has maintained a firm policy stance in rice full self sufficiency. This means that the government would not import or permit rice to be imported as long as domestic production can meet domestic demand.




48
Practically, no imports have been permitted since 1970, except for small quantities of glutinous and cracked non-glutinous rice for particular processed products (OECD, 1987).
Rice exports have been made only occasionally to dispose of accumulated stocks. The only time that there were notable rice exports is the period 1979-83 when Japan exported to other Asian countries about 3 million tones of surplus rice at favorable repayment conditions (long term, low interest credit). The reaction of the U.S. however led to Japan terminating those export sales.
The high level of support to rice growers coupled with falling rice consumption, and increasing yields has resulted in considerable stockpiles of unwanted rice. To cope with rice oversupply, Japanese administrations have introduced two control schemes: (i) various rice land diversion programs designed to curtail domestic supply, and (ii) the voluntarily marketed rice program.
Concerning diversion of paddy fields to other uses, there have been four programs over the 1970-87 period designed to reduce rice supply and increase the production of other priority crops. In all four programs, an acreage reduction target was set annually, and diversion payments were offered to participant farmers. Participation in the programs has been voluntary. The focus of each program however has been different.
The Rice production Control and Diversion Progrwn (1971-75) aimed primarily to reduce rice production and substitute other crops for rice. 'ne Comprehensive Paddy Field Utilization Program (1976-78) aimed to reduce rice production but increase the self sufficiency of certain crops (with priority given to soybeans, feed crops, and vegetables).




49
The Paddy Field Utilization Re-orientation Program (1978-86) focused more on the reduction of the farm size than the reduction of rice surpluses. Finally, the Paddy Field Farming Establishment Program (1987-92) was designed to improve farm productivity and establish a regional crop rotation program, besides curtailing rice production.
The voluntarily marketed rice program was introduced in 1969 in order to provide an alternative channel for marketing rice (besides the Food Agency). The major difference between voluntarily marketed and government marketed rice is that the government does not fix the purchase price and lets the market mechanism work. Nevertheless subsidies and assistance are provided for the smooth marketing of voluntarily marketed rice. Because its purchase price is not fixed, the voluntarily marketed rice is usually of higher quality. Consequently, the price of voluntarily marketed rice brands are about 25 % higher than the government marketed rice at the wholesale level and about 35% higher at the retail level (OECD, 1987). Thus the voluntarily marketed and government marketed rice complement each other in the preferences of the consumers for product differentiation and higher quality rice. It must be also noted that although the voluntarily marketed rice is subsidized the government expenses are still lower than they would have been if all produced rice was bought by the government (OECD, 1987).
The Wheat industry
Besides rice, the wheat industry is also important in Japan, and a close relationship exists between the two crops as a result of policies aimed to reduce the




50
excess rice production and encourage the cultivation of alternative crops. While wheat production was declining in the 1960s, the rice land diversion programs described earlier caused the production of wheat to rise and the level of wheat self sufficiency to increase in the 1970s. Nevertheless Japan imports about 80% of the wheat it consumes (Fitchett, 1988).
Similar to rice, the marketing of wheat is effectively controlled by the government via the Food Agency. The Agency purchases wheat from individual growers at a predetermined price and consequently sells it to wholesalers also at a fixed price. Although growers have the option of marketing their output privately, in practice the Food Agency buys almost all the wheat harvest, as it offers prices substantially higher than the international market. In addition the Food Agency fully controls all wheat imports; importers must seek government approval and sell all their importation to the Agency.
As in the case of domestically produced wheat, the government also sells imported wheat to domestic users at a fixed price. For domestically produced wheat the difference between the government purchase and resale price is substantially high, creating considerable deficits. By contrast, in the case of imported wheat the government resale price is usually above the international price at which the government buys imported wheat. Thus, the government gain from the sale of imported wheat helps offset the deficit created by paying high prices to domestic wheat growers; however this gain may fluctuate widely from year to year due to the fluctuations in world prices and exchange




51
rates (e.g. the sharp appreciation of the yen over the period 1985-87 lowered considerably the prices paid by wheat importers).
The Fresh Citrus Industry
Japan's fruit industry is largely dominated by citrus fruits the major crops being oranges and lemons. Domestic orange varieties include tangerines (such as mikan, hassaku, and iyokan oranges), the summer-orange natsu-mikan, and the navel orange. The domestically produced mikans are the most important fruit in terms of area planted and volume of production (Australian Bureau of Agricultural and Resource Economics, 1988).
Contrary to the grain industry, the horticultural industry in Japan does not receive any direct price support. This is not to say, however, that the industry does not enjoy government assistance and protection. A variety of measures such as quotas, import licenses, blending requirements, import duties, quarantine regulations etc. are in place to assist domestic growers against foreign competitors.
In the case of mandarin oranges, evidence suggests (Baker and Mori, 1985) that imported oranges do not seriously affect the domestic industry because they are marketed mainly between April and September when only small quantities of the domestic mikans are marketed. Nevertheless, for the time period examined in this study, imports of fresh oranges from all sources were subject to import quotas. In addition, an ad valorem tariff was imposed on orange imports at a rate of 20% for the period from June 1, through




52
November 30, and 40% for the period December 1, through May 31.1 Japan's
persistence in protecting fresh citrus has been attributed to the fact that citrus growing was encouraged as an alternative to rice production. At the same time, there has been extensive financial involvement of the agricultural movement in the fruit industry; it was therefore feared that abandoning import controls would reduce demand for the domestic orange varieties (Australian Bureau of Agricultural and Resource Economics, 1988).
Japan's Balance of Trade Function on Farm Imports
As shown earlier, deriving the TRI for the aforementioned traded goods requires first the specification of a (partial equilibrium) budget constraint that relates the aggregate Japanese consumer expenditure to the sum of the GNP function and trade revenues, associated with these goods., Before specifying the relevant functions however the notation to be used is presented below. For notation convenience, let:
h: the price vector of tariff-ridden goods (pork price = hp,,, chicken price
= hch),
p: the price vector of quota-controlled goods (beef price = p1,f, oranges
price = p)
p: the producer price vector of state-traded goods (wheat price =pvh, rice
price =p,,
14 In 1988 after negotiations with the U.S. Japan agreed to terminate import quotas on oranges and tangerines by.April 1991 and retain the same tariff after the quotas are lifted.




53
s: the consumer (user) price vector of state-traded goods (wheat price =
s,,, rice price = s,),
a*: the international vector price of state-traded goods (arWb for wheat a,.
for rice),
p': the international price for the quota controlled goods (Pbf for beef, p'o
for oranges),
Q: the vector of quota levels (Qbf for beef, Qo for oranges),
A: the riceland diverted from rice production in each period,
c: the subsidy received by rice growers under the paddy field diversion
programs.
On the production side, it is assumed that the farm products examined here are produced via production processes separable from each other. That is to say, from the producer's view-point the output level for each of these products depends only on its own price and on the prices of the respective inputs. This implies that joint production processes are assumed away and results in an additive GDP or revenue function expressed as
G(h, p, p; w,V) = gPk(hpk,w) + gh(hh, w) + gwh(pwh, w) + gr (p,
+ gbf (Pbf, W) + go (P,, W) + wIV (3.1)
The first six terms on the right hand-side of (3. 1) are the profit functions for the tariff-ridden goods (pork and chicken), the quota controlled goods (beef), and the state-




54
traded goods (wheat and rice), respectively. The last term to the far right represents payments to primary factors V employed in these agricultural sub-sectors. Since the analysis here is partial equilibrium, the prices w and supplies V of primary factors are considered exogenous.
The preceding GDP or revenue function can be equivalently written as the sum of returns to all factors associated with tariff-ridden, state-traded, and quota-controlled goods considered in this study, i.e.
G(h,p,p; w, V) = G'(hpk, hch, Pwh, PriW) + W/"V1 +
+ G2 (Pbf'Po, w) + w1V2 (3.2)
where G('), V are returns to all factors associated with tariff-ridden and state-traded goods, while G2(), V2 are returns to factors associated with the quota-controlled goods.
On the demand side, the aggregate expenditure function associated with these goods can be written as
e = e (hpk, hch, Swh, srl pbt, po, u) (3.3)
Then the excess expenditure over the revenue function G(') (i.e. the trade expenditure function) is given by the difference E(-) = e() G(') (3.4)
Given the fact that some of the goods under examination are subject to import quotas, the aggregate consumer minimizes expenditure only on the goods not subject to a quota. This implies that the trade expenditure function on goods other than those




55
which are quota-controlled is conditional on the quota levels. This leads to the distorted trade expenditure function defined as
() = @(hpk, ch, swh, Sri, Qbf, QO, U)
- G1 (hpk, hch, pwh, Pri; W) w'V' (3.5)
The excess (net) expenditure function of the aggregate consumer over all the goods under examination is then equal to E(hpk, hch, SWh, si, Ph Pri, Qbf L,, U) +
+ (pbfPo) (QbQo) (3.6)
or
E(h,s,p,Q,u) + p'Q (3.7)
By Shephard's lemma, the derivatives of E() with respect to h yield the (net) imports of tariff-ridden goods,
Eh(h,s,p,Q,u) = h(*) G(*) = m(h,s,p,Q,u) (3.8)
With respect to the state-traded goods (wheat and rice), the vector of their production levels is given by the derivative vector of the relevant profit function, i.e.




56
G H VP = Y(p)(39
while the vector of quantities demanded is given by the derivative vector of the distorted expenditure function e(-) i.e.
9, (h, s, Q u) = X(h, s, Q u) (3.10)
Additionally, the quota levels are defined as
0 = e (1) G (-) (3.11)
Separately, the government programs for the aforementioned agricultural products generate net government revenue (which may be positive or negative). In particular, this revenue consists of:
(i) the portion of the quota rents in beef and fresh oranges retained in Japan.
These portions can be approximated by the number of quota licenses awarded to foreign importers. However given the restrictive trading status in beef imports (regulation of imports by the parastatal LIPC) the portion of beef quota retained abroad may be expected to be zero. The portion
pertaining to fresh oranges is denoted below by wi.
(ii) the tariff duties and variable levies from the importation of chicken and
pork,
(iii) the implicit subsidy (positive or negative) to consumers from fixing the
consumer (selling) price in wheat and rice, and the implicit subsidy to producers from fixing the producer (purchase) price for wheat and rice.
More formally, the (net) government revenue is equal to the following sum:




57
(Pb P; ) Qbf + (i-) ( 0o ) Q
+ 'pk Mp() + T ch m ()
- (Pwh G- h) Ywh() + (sh _oh) X ()
_ (1rO1) yrli(.) + (SriO' ) X"i() (3.12)
The first line above denotes government revenues from quota rents. In particular, the first term is the quota rent on beef retained in Japan while the second term is the portion of the quota rent on fresh oranges. The second line is the government revenues from imposing tariffs and levies on chicken and pork imports, respectively. The third and fourth lines are the government revenues (positive or negative) from the policies applied on domestically produced wheat and on rice; the first terms are the implicit subsidy to producers (the difference between producer price and international price) while the second terms are the implicit subsidy (positive or negative) to the consumer (user).
In more compact notation these government revenues can be expressed as (1-u) (p-p.)'Q + Tlm(h,s,p,Q,u)
- (p-a)I Y(p) + (s-O)'X(h,s,Q,u) (3.13)
Further, it is assumed that the government trade revenues specified above are redistributed costlessly to the aggregate consumer, in a lump-sum fashion. In addition, the government funnels into the rice sub-sector subsidies for diverting paddy fields to other uses. The respective government outlay on rice diversion can be expressed as




58
(c-A). By definition, the value of consumption expenditure in the farm sub-sectors examined here less the value of the respective domestic product must equal total transfers to these sub-sectors from the government. These total transfers are the sum of the relevant government trade revenues shown in (3.13) and the riceland diversion outlay. Thus the balance-of-trade function is defined as
B(h,s,p,Q,A,u) = E(h,s,p,Q,u) + p'/Q
- (1-o) (p-p*)'Q I'm (h,s,p,Q,u) +
+ (p-a*)'Y(p) (s-o*)'X(h,s,Q,u) cA (3.14)
To facilitate the discussion, the prices of all price-constrained, traded goods are henceforth denoted by r, i.e.
S= hpk, ch, Swh, p) (3.15)
Given the policy of self-sufficiency on rice and the fact that rice imports were zero in the period studied here while exports took place on exceptional basis (surplus disposal), rice may be treated as a (quasi) non-traded good. As discussed in Chapter 2, in such a framework a trade regime involving price restrictions, quantitative import control, and intervention in factor markets (paddy field programs) the TRI is defined as




59
13 = 13: B(hkP, Ih 0, s'
h ch Sri, P ,~
1 1
1 Qbf Q0 3.6
P= 0 1 (3.16)')
That is, the TRI is the proportionate factor # by which one must discount the prices of the price-constrained, traded goods, as well as, the quota levels in period 1 so that the farm sector described by the function B() in (3.14) returns to the welfare level uo of the previous period. Moreover the rate of change in the TRI is given by total differentiation of (3.14) with respect to the policy variables r, Q, s, p,, and A:
m BclQj n __B__r+ Bxi B! Ri +
B, ag BBQ-Br B7
BsSri Ri + Bp Pri + BAA (3.17)
13 (~Q-~.i 13(B~Q-B.',) 13(BQQ-Bvlx)
where a hat (^) over a variable denotes a proportionate change, e.g. (dQ/Qj).
The TRI, in other words, equals the sum of the percentage changes in all trade restrictions (i.e. tariff rates and quota levels) with each restriction being weighted by its contribution to the total welfare cost of the initial trade structure. At the same time, the policy changes in non-traded goods and their inputs are considered as seen in the second line of (3.17).
Computation of this TRI requires first specification and evaluation of the policy derivatives B,, BQ, B,, BA. In this way one can calculate the weights by which the rates of change in policy variables -r, Q,s, p, A are aggregated into a single measure the TRI. To specify the policy derivatives B,, BQ, B,, BA one needs to differentiate the




60
balance of trade function B() in (3.14) with respect to all policy variables. A price (or a quota) appearing as a subscript in the following expressions denotes the partial derivative of the relevant function with respect to this price (or quota); for example the term
my (3.18)
denotes the partial derivative of net pork imports mPk(), with respect the pork price hPk.
Similarly, the term
pbof (3.19)
denotes the partial derivative of the price of fresh oranges po with respect to the beef quota Qbf. Differentiation of the function B() with respect to policy variables yields
pk ch w X I
( B h 4 k p k p k p k 0 'p k
hp Ipk ch wh ri 0 ch
Bh sh mhe X 0 -o*
Bh Ch Swh-Owh +
B mh M; m M X;," X ,zM 0 sri ori1
+ Io /0o
)w P o h
0 0 0 0 -PhGh
Ph ,
+ 0 h, (3.20)
.0.
+()oPhh1




61
pk 0h Wh r' ( Pk (0 mbf MOb2 XObf Xobtf ITch
B0_ 4kc Mjh vh Xr i Swh-Owh
+ (0 QO (Pbf Pbl* (3.21)
+~ 0 iQ O t',lI
B =(mjk ich, Xwh XzI) +,1 Wh'ho (3.22)
S.-1 1 ii 5r1 0r' X8ri Sw + w OpSri Si (Fr.
B~1 (~*i) yiI (3.23)
BA -R -L C = WR -c (3.24)
where WR is the return to land in rice production, and R is the amount of land (paddy fields) employed in rice production. Given that A denotes riceland diverted from rice production, (aR/aA)= -1.
In more compact notation, expressions (3.20), (3.21) are written as




62
B = + W 0P, Q (3.25)
BO = E i + (O PO Q (I-Q) (p-p*) (3.26)
where 4, #, are tariff vectors, I is the identity matrix and 0 is a diagonal matrix with the portions of quota rents on beef (wobf = 0), and oranges (w) retained abroad, on the main diagonal.
The evaluation of the derivatives B.,, BQ, and B, can be carried out by using the equalities (2.31) and (2.32) shown in Chapter 2, which are reproduced here for convenience
m0=E,0 =E E1; m.=,-- =-E E ,-E ,~ (3.27)
(3.27)
m, = Ez,, = ExzEpp m, = Ew = E,, EzEp E,
pO = -f2, = Epp p, = 9, =-p Ep, (3 .28)
Specifically, one can compute
-pk -pk -pk
- Ch -Ch -Ch
m, mhc m3 Wh Wh ch
:K;X X.-h




63
-pk -pk1
mP mPl [ bf-1 Qf Qb,
mPbr P I -1 0 ohl-hlb
-Pch __C ~ IF E b b (3.29)
m~hfmP ,.1 Qh0, Qhc0 !L
[PMt P0
where m() denote the first derivatives of the standard (instead of the distorted) trade expenditure function E('). The terms mQ, p., and pQ can be computed in an analogous fashion.
It may be noted that the evaluation of the policy derivatives B1., BQ, B,, and BA requires information on:
(i) the respective price elasticities on import demand and supply,
(ii) domestic prices, international prices, and the imported quantities of the quotacontrolled goods (beef, fresh oranges).
An empirical estimation of these elasticities is the task of the next section while the price and quantity data are discussed in Appendix C.
Demand Elasticities A System-wide Approach
In estimating import demand elasticities for the farm products considered in this study the system-wide (or differential) approach developed by Theil (1967) is utilized. This focuses on systems of consumer demand equations rather than individual equations. Moreover, instead of initially selecting a particular functional form for the consumer's utility function to generate the demand system, one differentiates the first-order conditions of the standard utility maximization problem to obtain a system of differential demand equations (i.e. expressed in terms of changes in prices and quantities).




64
Specifically, manipulation of the utility maximization-first order conditions yields the following demand equation for the ith good:
N
wid(logq) = 6id(logQ) + 4E Od(logp,-logP/) (3.30)
where
Pi q = i =1,...,N (3.31)
SM
is the respective budget share and
0 (piqi) N (3.32)
= M =1
denotes its marginal share. Additionally, d(logQ) is a Divisia volume index of the form N (3.33)
d(loge) = wid(logq)
d(logP') is a Frisch price index defined as tt (3.34)
d(logP') = Od(logpi) (334)
T.=1
and 4 is the reciprocal of the income elasticity with respect to i (the Lagrange multiplier),
1 alogM (3.35)
A tlogM
Also, the following qualities hold




65
N N N N N
E O E Oi = () E EOi = E ) = 1 (3.36)
i=1 j=1 i=1 j=1 i=1
In empirical implementation of the differential approach, based on the number and the similarity of goods examined, the assumption of (weak) separability is frequently utilized (Theil, 1980). This assumes that the various goods in the consumer's utility function may be divided into a number of commodity sub-groups, so that the overall utility function is some increasing function of the group sub-utilities. The consumer is assumed to first allocate the budget among the different groups and, in a second stage, the expenditure for each group is further allocated among the goods belonging to that group (multistage budgeting). Groups are assumed to be either strongly or weakly separable with each other; the terms blockwise dependence and block independence are also used. Within each group however the goods are no longer separable.
For estimation purposes, the general form-differential demand system can be parametrized in a number of ways; a particular parametrization is the Rotterdam model where the demand parameters are assumed constant over time. The Rotterdam specification assumes that the coefficients Oij, 4 are constant over time, and wi, d(logqc), and d(logp) are approximated as
wjt = 1 (wit+wit-,) d(logqit) =dq1t=1ogq1t-1oqq2tj
d(logpit) dp1t=1ogpit ogpiti3.37)




66
Then the corresponding discrete-time versions of the Divisia and Frisch indexes are
dP N to N (3.38)
ii. i-1
and the differential demand equation of the ith good takes the estimable form N (3.39)
Witdqit = OdQt + wayjdp(+e39)
j.1
where
nsj = < (O, OiOj) i,j=1, ...N (3.40)
and eg is a random disturbance term which is assumed (Theil, 1980) to follow a multinormal distribution with zero mean and covariance
cov(ei,eje) = 02 (Oj 6,0j) i,j=1, ...,N (341)
The Oj's account for the price effect on the quantity demanded, keeping the utility level constant (compensated effect) and are known as the Slutsky coefficients of the Rotterdam model. Note that given the linearity of (3.39) in its parameters, the income elasticity i and the compensated price elasticities Eij can be readily obtained as C i11 (3.42)
Wi Wi
From (3.36) and (3.40) it can be seen that




67
N (3.43)
i i=i1 = 0N
i-i
This implies that the NxN Slutsky matrix [i] has rank N-1. Additionally, it is shown (Theil, 1980) that the covariances of the disturbance terms ei, are given by
cov(e,e t) = 02(0ij 0,0) = 0 2 W (3.44)
that is, all disturbance covariances (and variances) are proportional to the corresponding Slutsky coefficients. Thus the disturbances el ,... ,eN, add up to zero and have a singular covariance matrix. This means that in implementing Rotterdam demand systems one equation may be dropped out and the other N-I equations can be jointly estimated by maximum likelihood techniques.
The linearity of the Rotterdam model in its parameters allows us to impose on the demand equations the standard constraints of demand theory by means of linear restrictions. In particular, the properties of adding-up, homogeneity, and Slutsky symmetry can be imposed as follows:
(a) Adding-up
N N (3.45)
i-1Ot = -7 (134)
(b) Homogeneity
N (3.46)
_ J= 0O,




68
(c) Slutsky symmetry
Oi = Oi (3.47)
and their compatibility can be checked by means of the likelihood ratio test (LRT).
In this study the absolute version of the Rotterdam model is used under the assumption that the farm products considered here are divided into two separable groups. The first group includes meat and grains (i.e. beef, pork, poultry, wheat, and rice). Fresh oranges are assumed to belong in a second group that includes all citrus fruit."5 Moreover it is assumed that the aggregate consumer's welfare function on farm imports consists of the sub-utilities deriving from these two groups which are assumed to be separable from each other. Estimation of the demand systems generated from these two groups yields conditional price and expenditure elasticities (i.e. elasticities which depend upon only the prices, quantities and income expenditure allocated to the particular group).
Typically, the quantity and value of imports are used as data source for the estimation of import demand elasticities. In this study however, preliminary estimations using the reported import quantities and import value of the goods under examination (FAG-Trade Yearbook, various issues), produced unsatisfactory elasticity estimates. This may be attributed to the fact that the markets examined in this study are severely distorted-quantity restrictions imposed on beef imports combined with state trading in the
11 Fresh oranges were separated from the rest of the goods, as preliminary estimations including all six goods in a single group showed that cross price relations between oranges and the rest of the goods are insignificant.




69
case of wheat and rice. Thus, using trade data on a model based on the implicit assumption of undistorted markets, such as the Rotterdam model, is likely to yield questionable estimates (e.g. positive own-price elasticities).
However this problem may be overcome by recalling that imports defined as the partial derivatives of the trade expenditure function E(-) are the difference between the quantity demanded and quantity supplied (see equations (3.8) and (3. 11) in section 3.7 of this Chapter). Thus import demand elasticities can be computed as the difference between elasticities of demand and supply. Accordingly, the first group (meat and grains) was estimated for demand (rather than import demand) price elasticities. The price data used in this estimation are the respective domestic wholesale prices (reported for each commodity in Appendix C) while the quantity data are the respective total consumption volumes (gross food) reported in OECD -Food Consumption Statistics (196478, 1979-88).
The econometric estimates and test statistics for the first group (e.g. beef, pork, poultry, wheat, rice) are reported in Tables 3.2 and 3.3, respectively. Table 3.2 presents the estimated conditional price coefficients and expenditure coefficients with the standard constraints of demand theory (i.e. homogeneity and Slutsky symmetry) imposed. AUl own-price estimates are negative as expected. The own-price estimates of beef, pork, and rice are statistically different from zero at ci= 0.05 level of significance while those of poultry and wheat are statistically different from zero at at =0. 11 level and a= =0. 14 level, respectively. Concerning the cross-price Slutsky coefficients, a positive sign indicates substitutes, while a negative sign indicates complementary goods. Of the cross-




70
Table 3.2: Parameter estimates of a Rotterdam model on meat (beef, pork, poultry) and grains (wheat,rice), homogeneity and symmetry imposed, 1965-87.
Conditional price coefficients i'ij Expend.
coeff.
Beef Pork Poultry Wheat Rice 0i
Beef -.0308 .0252 .0013 .0037 .0006 .4488
(.012)' (.009) (.0055) (.0020) (.014) (.056)
Pork -.0446 -.0001 -.0038 .0233 .3565
(.014) (.0072) (.0028) (.018) (.0582)
Poultry -.0096 -.0062 .0146 .1619
(.0073) (.0027) (.009) (.0305)
Wheat -.0034 .00976 .0003
(.0031) (.003) (.010)
Rice -.0483 .0329
(.030) (.0944)
a Asymptotic standard errors in parentheses.
Table 3.3: Hypothesis testing of the Rotterdam model.
Log of likelihood Likelihood ratio X2
Model function test (LRT) (.05)
Unrestricted 389.852
Homogeneity 386.889 5.926 9.49(4)"
Homogeneity and 380.603 12.572 12.59(6)
Symmetry
Homogeneity and 18.498 18.31(10)
Symmetry vs.
unrestricted
a Numbers inside the parentheses indicate number of restrictions imposed.




71
price estimates, the wheat-rice term and the beef-pork term are statistically different from zero at a = 0. 01 level and positive; this implies that wheat and rice, as well as, beef and pork are substitutes. Additionally, the beef-wheat term and the poultry-rice term are statistically significant from zero at i = 0. 10 level and e = 0. 15 level respectively; they are also positive indicating substitutability between beef and wheat, and poultry and rice. Finally, the poultry-wheat term is significant from zero at the U=.025 level and negative indicating complementarity between wheat and poultry.
The validity of homogeneity and symmetry restrictions is then checked by means of a likelihood ratio test (LRT). The relevant test statistic is
LRT = -2 [ logL, logLu] (3.48)
where LR is the log-likelihood value of the model with the restriction(s) imposed and Lu is the log-likelihood value without the restrictions. The LRT statistic has an asymptotic x2(r) distribution, where r is the number of restrictions imposed (i.e. the degrees of freedom equal the difference between the number of parameters in the model without restrictions and with restrictions).
The computed values of the LRT statistic appear in the second column of Table 3.3. The following three hypotheses were tested: (a) the null hypothesis of homogeneity against the unrestricted version of the model, (b) the null hypothesis of symmetry against homogeneity, (c) the null hypothesis of both symmetry and homogeneity against the unrestricted model. The null hypotheses (a) and (b) cannot be rejected at the 0.05 significance level, while the null hypothesis (c) cannot be rejected at the 0.02 significance




72
level. This implies that the estimated demand equation system complies with the properties of homogeneity and symmetry in prices as required by the standard consumer demand theory.
Conditional (and compensated) price elasticity estimates can be computed by dividing the relevant Slutsky price parameter by the budget share of good i, that is
eli = 1"(3.49) However for the group tested here (meat,grains) this generates price elasticities of demand rather than imports. The conditional own-price elasticities of beef, pork, poultry, wheat, and rice are reported in Table 3.4, while Table 3.5 gives their conditional cross-price elasticities. Of the goods belonging to this group, the calculation of the TRI requires demand elasticities for grains (wheat and rice) but import demand elasticities for the three types of meat (beef, pork, and poultry). As mentioned earlier, import demand elasticities for meat can be obtained as the difference between demand and supply elasticities.
It may be noted that the differential approach used in demand analysis can be also used to generate a system of supply equations, since utility maximization and production maximization are mathematically identical optimization problems. Hence the Rotterdam model can be also utilized to estimate supply elasticities. However due to insufficient data on producer prices, no supply elasticities could be estimated. As an alternative, exogenous information on supply elasticities was used. In preparation for the Uruguay Round of GATT negotiations, the Economic Research Service (ERS) of the U.S.




73
Table 3.4: Conditional own-price elasticities of demand Rotterdam model on meat (beef, pork,poultry) and grains (wheat, rice), 1965-87.
Year Beef Pork Poultry Wheat Rice
1970 -0.270 -0.323 -0.172 -0.053 -0.077
1971 -0.242 -0.273 -0.142 -0.055 -0.084
1972 -0.224 -0.257 -0.152 -0.060 -0.085
1973 -0.167 -0.235 -0.140 -0.061 -0.097
1974 -0.201 -0.244 -0.138 -0.060 -0.090
1975 -0.200 -0.217 -0.150 -0.069 -0.092
1976 -0.187 -0.231 -0.146 -0.060 -0.093
1977 -0.173 -0.237 -0.166 -0.061 -0.093
1978 -0.160 -0.235 -0.176 -0.062 -0.095
1979 -0.175 -0.232 -0.152 -0.062 -0.094
1980 -0.182 -0.231 -0.143 -0.055 -0.095
1981 -0.196 -0.219 -0.134 -0.056 -0.096
1982 -0.182 -0.244 -0.143 -0.058 -0.093
1983 -0.171 -0.238 -0.142 -0.054 -0.096
1984 -0.166 -0.24 1 -0.138 -0.054 -0.098
1985 -0.157 -0.275 -0.147 -0.054 -0.094
1986 -0.147 -0.281 -0.149 -0.055 -0.096
1987 -0.135 -0.286 -0.171 -0.057 -0.097




74
Table 3.5: Conditional cross-price elasticities of demand Rotterdam model on meat and grains (beef, pork, poultry, wheat, rice), 1965-87.
Year Beef Beef Poultry Poultry Wheat
w.r.t. w.r.t. w.r.t. w.r.t. w.r.t.
pork wheat wheat rice rice
1970 0.221 0.032 -0.111 0.262 0.150
1971 0.198 0.029 -0.091 0.216 0.157
1972 0.183 0.027 -0.098 0.231 0.170
1973 0.137 0.020 -0.090 0.213 0.171
1974 0.165 0.024 -0.089 0.210 0.170
1975 0.164 0.024 -0.097 0.229 0.195
1976 0.153 0.022 -0.094 0.222 0.169
1977 0.141 0.021 -0.107 0.253 0.173
1978 0.131 0.019 -0.113 0.268 0.175
1979 0.143 0.021 -0.098 0.232 0.175
1980 0.149 0.022 -0.092 0.218 0.157
1981 0.160 0.024 -0.086 0.203 0.158
1982 0.149 0.022 -0.092 0.218 0.165
1983 0.140 0.021 -0.092 0.217 0.154
1984 0.136 0.020 -0.089 0.210 0.153
1985 0.129 0.019 -0.095 0.225 0.153
1986 0.121 0.018 -0.096 0.227 0.155
1987 0.111 0.016 -0.110 0.260 0.162
(continued)




75
Table 3.5 continued.
Year Pork Wheat Wheat Rice Rice
w.r.t. w.r.t. w.r.t. w.r.t. w.r.t.
beef beef poultry poultry wheat
1970 0.182 0.057 -0.095 0.023 0.016
1971 0.154 0.059 -0.099 0.025 0.017
1972 0.145 0.065 -0.108 0.026 0.017
1973 0.133 0.065 -0.109 0.029 0.020
1974 0.138 0.064 -0.108 0.027 0.018
1975 0.123 0.074 -0.124 0.028 0.019
1976 0.131 0.064 -0.107 0.028 0.019
1977 0.134 0.065 -0.110 0.028 0.019
1978 0.133 0.066 -0.111 0.029 0.019
1979 0.131 0.066 -0.111 0.029 0.019
1980 0.131 0.059 -0.099 0.029 0.019
1981 0.124 0.060 -0.100 0.029 0.019
1982 0.138 0.063 -0.105 0.028 0.019
1983 0.135 0.058 -0.098 0.029 0.019
1984 0.136 0.058 -0.097 0.030 0.020
1985 0.155 0.058 -0.097 0.029 0.019
1986 0.159 0.059 -0.098 0.029 0.019
1987 0.161 0.061 -0.102 0.029 0.020




76
Department of Agriculture (USDA) constructed a trade database known as the Trade Liberalization Database (TLIB) (USDA, 1989b). This database includes a set of price elasticities assembled from a survey of global agricultural models and commodity market studies. For developed countries such as Japan, the principal sources of these elasticities are: the Ministerial Trade Mandate (MTM) developed by OECD, the Grain Livestock, and Sugar model developed by Tyers and Anderson, and the Grain, Oilseed and Livestock model developed by the USDA (USDA, 1989b). The price elasticities of the TLIB have been used in numerous commodity models and comprise one of the most complete sets of price elasticities available. This study utilizes supply price elasticities on Japan developed by the OECD and reported in the TLIB. These elasticities which exclude cross commodity effects are reported as:
f(s).,,=0.23, E(s)(,.j = 1.5, E(s)(p,,)= 1.5, E(s)(W,, .)=0.44, E(s)(, )=0.3
Table 3.6 reports import demand elasticities for beef, pork, and poultry computed by utilizing the preceding supply estimates, as well as, the conditional demand elasticities for wheat and rice. In summary, Tables 3.5 and 3.6 report the (estimated) information requirements which will be used next in the computation of the TRI.
Computing the Combined TRI for a Subset of Japan's Farm Imports. 1971-1987
In this section the calculation of the combined TRI of Japan imports of beef, pork, poultry, wheat, rice and fresh oranges is presented. The derivatives of the respective balance of trade function were evaluated for each period, utilizing the elasticity estimates reported above, along with information on imported quantities, domestic, and




77
Table 3.6: Own-price elasticities of imports for meat (beef, pork, poultry) and ownprice elasticities of demand for grains (wheat, rice), 1970-87.
Beef Pork Poultry Wheat Rice
Year
1970 -0.500 -1.823 -1.672 -0.053 -0.077
1971 -0.472 -1.773 -1.642 -0.055 -0.084
1972 -0.454 -1.757 -1.652 -0.060 -0.085
1973 -0.397 -1.735 -1.640 -0.061 -0.097
1974 -0.431 -1.744 -1.638 -0.060 -0.090
1975 -0.430 -1.717 -1.650 -0.069 -0.092
1976 -0.417 -1.731 -1.646 -0.060 -0.093
1977 -0.403 -1.737 -1.666 -0.061 -0.093
1978 -0.390 -1.735 -1.676 -0.062 -0.095
1979 -0.405 -1.732 -1.652 -0.062 -0.094
1980 -0.412 -1.731 -1.643 -0.055 -0.095
1981 -0.426 -1.719 -1.634 -0.056 -0.096
1982 -0.412 -1.744 -1.643 -0.058 -0.093
1983 -0.401 -1.738 -1.642 -0.054 -0.096
1984 -0.396 -1.741 -1.638 -0.054 -0.098
1985 -0.387 -1.775 -1.647 -0.054 -0.094
1986 -0.377 -1.781 -1,649 -0.055 -0.096
1987 -0.365 -1.786 -1.671 -0.057 -0.097




78
international prices (a detailed description of price and quantity data is presented in Appendix C).
As shown in equation (3.17), the combined TRI is then calculated by weighing the percentage change of the policy variables ('r,Q,A) in each period with their respective share in the total dead-weight loss due to trade distortions, given by (B'QQ B'T'x). The combined TRI for the period 1971-1987 assuming full quota rent retention16 in the home country (Japan), is presented in Table 3.7. Inspection of the table reveals that the rate of change of the combined TRI is basically shaped by changes in the riceland diversion programs and the beef quota and, to a lesser extent, pork and poultry prices. In contrast, the contributions of rice and wheat are minimal reflecting the persistent policies of regulating the producer and consumer prices in both crops throughout the period under examination. The rate of change of the index increases with lower prices r for tariffridden and state-traded goods, higher quotas Q, and less land employed in rice production. Hence, the index rises as the trade distortions (-r,Q), and the misallocation of resources (riceland) are getting reduced.
The level of the index can be computed by using the simple difference equation shown in Chapter 2. One can use either the chain-principle (i.e. in each period, use as basis the level of the TRI in the previous period, starting with 00 = 1) or use = 1
16 In the absence of any information on the portion w of orange imports controlled by foreign importers, the values w=O, w=0.2, wi=0.4 were alternatively used to account for full quota retention in the home country, low quota retention abroad, and high quota retention abroad, respectively. Computations using W=0.2, W=0.4 produced little change in the estimated TRI, apparently due to the small contribution of fresh oranges in the overall index.




79
Table 3.7 The combined Trade Restrictiveness Index of Japan's agricultural imports, 1971-1987.
Pork Poultry Wheat Rice Wheat
Year (demand) (demand) (supply)
1970
1971 -0.02459 -0.03309 -3.3e-06 0.000114 -0.00113
1972 -0.01698 0.02461 0.000022 -0.00074 -0.00068 1973 -0.01367 -0.0158 -0.00007 0.000029 -0.00117
1974 -0.0369 -0.03618 -0.00021 -0.00284 -0.00325
1975 -0.09505 -0.02049 -0.00002 -0.00294 -0.00142 1976 -0.00117 -0.0052 -0.00014 -0.0013 -0.00087
1977 0.006232 0.001564 -0.00003 -0.00112 -0.01491 1978 0.025769 0.013365 0.000001 0 -0.00057
1979 0.037185 0.007626 3.11 e-07 -0.00072 -0.00102 1980 -0.00702 -0.00944 -0.00006 -0.00063 -0.00408 1981 -0.04444 -0.00927 -0.00002 -0.0005 -0.00143
1982 0.013591 0.009637 -4.2e-06 -0.00104 0
1983 -0.00216 0.004764 -0.00005 0 -0.00026
1984 0.004486 0.002236 0.000001 -0.00105 0
1985 0.031799 0.010745 -4.3e-07 -0.00113 0.000712 1986 0.006694 0.004628 -1.3e-06 -0.00035 0.002144 1987 0.012805 0.008617 0.000009 0 0.035188
(continued)




80
Table 3.7 continued.
Rice Beef Oranges Paddy TRI TRI
Year (supply) programs level
1970 1
1971 -0.00243 0.152008 0.002241 0.862268 0.955389 1.95539 1972 -0.004 0.119979 0.004022 0.101089 0.227323 2.30784 1973 -0.00598 0.367431 0.000886 -0.00698 0.324668 3.07404 1974 -0.02111 -0.54561 0.002035 -0.70616 -1.35023 0.43768 1975 -0.01334 -0.07086 0.000711 -0.0775 -0.2809 0.26200 1976 -0.00501 0.318762 0.000586 -0.06777 0.237889 0.26966 1977 -0.00305 -0.0596 -0.00038 0.017219 -0.05408 0.26460 1978 -0.00006 0.672784 0.005136 0.31105 0.427475 0.60641 1979 -0.00015 0.126254 0.000945 0.037107 0.207231 0.74634 1980 -0.00225 -0.02736 0.004886 0.166777 0.120817 0.87809 1981 -0.00035 -0.00084 0.001199 0.094097 0.038447 0.92321 1982 -0.00137 -0.00378 0.00375 0.006114 0.026898 0.94833 1983 -0.00223 0.059956 0.002609 -0.04552 0.017113 0.96209 1984 -0.00284 0.029695 -0.00012 -0.01438 0.018021 0.97874 1985 0 0.021505 0.009789 -0.01489 0.058526 1.03568
1986 0 0.113283 0.001288 0.009488 0.137174 1.17742
1987 0.004595 0.130345 0.001122 -0.00027 0.192411 1.40320




81
for each period. In the latter case, the level of the TRI coincides with its rate of change. Figure 3.1 shows the rate of change of the TRI, while Figure 3.2 shows the level of the TRI using the chain-principle.
The index shows considerable variation in the first half of the period under examination (1971-77) and a relatively smooth pattern in the second half (1978-97). In particular, during the period 197 1-73 the index shows a positive but rapidly decreasing rate of change as beef quota, orange quota and riceland diversion all rise at a diminishing rate, while the domestic prices of pork and poultry rise, too. In 1974, the rate of the TRI turns negative implying a severe reduction in the magnitude of the TRI and therefore a dramatic increase in trade protection for that period. This severe drop of the index reflects the decision of the Japanese government to suspend the beef quota in late 1973 and to completely close the beef market to imports in 1974 until the second half of 1975 (Australian Bureau of Agr. and Res. Economics, 1987). This was coupled with an increase of almost 12% in the wholesale pork price, a 13.5% increase in the wholesale price of poultry, and a 79.5% reduction in the amount of riceland diverted from rice production.
In the period 1975-77 the index shows a recovery as its rate of change becomes less negative in 1975 and turns positive in 1976, almost zero in 1977 (reflecting stable policies during 1976-77) and again positive in 1978 reflecting higher beef imports, higher riceland diversion and lower pork and poultry prices. Thereafter the index shows a smoother variation. Specifically, for the period 1978-84 the index shows a positive but decreasing rate of change as the beef quota was slightly reduced in 1980, 1981, 1982,




82
1
0.6
0.4
u 0.2
-0.2
-0.4
-1
0.2
-0.4
-i .2
1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 Year a
Figure 3.1 The rate of change of the combined TRI, 1971-87.




83
3.2
3
2.8 2.6
S 2.4
S 2.2 2,2
2
S 1.8 1.6
1.4
- 1.2
-J I
0.8 0.6
0.4 0.2
1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 Years
Figure 3.2 The level of the combined TRI (1970= 1), 1970-87.




84
while pork and poultry prices rose and the diversion rate of riceland became smaller and even negative in 1983, 1984.
After 1984 the index rate of change shows a steady increase indicating a trade liberalization process at least for the period 1984-87. This is consistent with the socalled 1984 U.S.-Japan Beef and Citrus Understanding in August 1984, Japan agreed to expand its import quotas for fresh oranges and grain-fed beef (mainly supplied by the U.S.) (USDA, 1986). It also reflects decreasing domestic prices on pork and poultry, stable or even decreasing domestic prices in wheat and rice, and very small variation in the riceland diversion programs.
Concerning the estimated level of the combined TRI, it may be noted that when this is computed according to the chain-principle, the dramatic drop of the index around 1974 affects the magnitude of the TRI in the subsequent years. As a result, the index shows a slow recovery which. gains momentum after 1984. Levelwise the index in the last period (1987) is 40% higher than in the first period (Figure 3.2). This implies that using the trade regime of the initial period (1970) as reference-point, the regime of the last period (1987) is clearly less restrictive. Nevertheless, the magnitude of the TRI in 1973 is three times higher than in the first period. Therefore, one must also note that the lower distortions in the end of the examined period are still more strict compared to those in the beginning of the 1970s.




85
Comparing the TRI with the Respective PSE/CSEs
As explained in Chapter 2, the TRI is not directly comparable with other measures of trade protection. Nonetheless, Anderson and Neary (1992) suggest decompositions of the TRI which allow some comparison with the more conventional measures of producer and consumer subsidy equivalents. Specifically, one may define restrictiveness indexes separately for production and consumption distortions in an analogous fashion to the full TRI.
In the application considered here, one may define a distortion index 1 for traderelated production distortions as
OP- a[P: B(h1P, h1 P l P Qf Q A o] (3.50)
pkhchf ,Pwhf ,,ri"
The rate of change of this index is computed analogously to that of the TRI, considering however only the policy variables related to production i.e. hk, hoh, Pwh, Qbf, Q., p1,, A and ignoring policy variables related to consumption (i.e. S,,h, Sj).
Hence 0' is interpreted as the equiproportionate change in trade-related production distortions (accounting, at the same time, for production distortions in non-traded goods) which is welfare-equivalent to policy-changes from period 0 to period 1.
A similar distortion index can be derived for trade-related consumption distortions, i.e. one may define a distortion index 3' as




86
[ 1 h I3 c h csi, QCt Q 0) 1 (3.51)
. =3CB(hpk',hcO, sc ; u
The rate of change of 03 is calculated by considering only the policy variables related to consumption, i.e. hpk, hch, sh, S Qbt, Qo and ignoring production policy variables, i.e. Pwh, p,, A. Hence, can be interpreted as the equiproportionate change in trade-related consumption distortions (considering however the effect of consumption distortions in non-traded goods) which is welfare-equivalent to these policy-changes from period 0 to 1. Since the distortion indexes P, 3c are defined similarly to the full TRI, an increase their magnitude implies less trade restrictiveness.
The distortion indexes 0P, #c may be compared to the rate of change in the conventional PSEs, CSEs. The Economic Research Service (ERS) of the U.S. Department of Agriculture (USDA) has calculated PSEs, CSEs for a number of farm products in 27 countries for the period 1982-87. Table 3.8 presents the estimated Japanese PSEs, CSEs for beef, pork, poultry, wheat, rice, and mandarin oranges (USDA, 1990). These PSEs, CSEs are expressed as total income transfer per unit of output, i.e. yen per metric ton, (Y/MT). An aggregate PSE, CSE is then calculated by weighing the individual PSEs, CSEs by the respective production (consumption) value shares, i.e.
PSET = 9j PiYi)pSEi, CSET =i x CSE,(3.52)
i p'y 1 q'x
where i= beef, pork, poultry, wheat, rice, mandarin oranges, pjy is the production value of the ith product, and qxi is the consumption value of the ith product. The percentage




87
Table 3.8 A comparison of distortion indexes with PSE(CSE) measures.
Agreggate % % Agreggate % %
PSE change in change in CSE change in change in
PSE OP CSE
Y/MT Y/MT
Year
1982 310951 -263950
1983 342365 0.09175 0.01717 -305985 0.13737 0.06958 1984 336236 -0.0182 0.01908 -305721 -0.0008 0.03757
1985 328767 -0.0227 0.05964 -305535 -0.0006 0.07745
1986 391267 0.15973 0.13735 -387632 0.21179 0.13277 1987 396640 0.01354 0.19380 -403549 0.03944 0.15847
rate of change of these agreggate PSE, CSE is then reported along with the percentage change of the distortion indexes flP, j1.
The production distortion index 13" shows positive rates of change throughout the period 1982-87 thus implying diminishing production- related distortions. In contrast, the aggregate PSE shows negative rates of change (thus implying less distortions) only in 1984, 1985. For 1983, 1986 and, 1987 the aggregate PSE shows positive rates of change suggesting higher production distortions. Identical conclusions are drawn by inspecting the rate of change of the aggregate CSE relative to that of the distortion index fc. It is therefore clear that the TRI approach leads to different conclusions on trade liberalization from the existing measures of PSE(CSE).
At least two reasons may be given for these contradictory results. First the PSEs, CSEs include policies which are not taken into account in the calculation of the traderelated distortion indexes fl", #' since they are not directly related to trade. Second, the




88
aggregate PSEs, CSEs and the trade-related distortion indexes I3~ 3~are constructed in fundamentally different ways.
In estimating the aggregate PSE(CSE) one basically weights the price changes of the individual commodities by their value shares (in production or consumption). Hence, the overall rate of distortions is shaped by goods with high production (consumption) value. Such an aggregation -is not derived from any theoretical basis; therefore the relative importance assigned to the price variations of the individual commodities is ad hoc and open to criticism. It must also be noted that the PSEs/CSEs consider the variation in price in both the cases of tariff-ridden and quota-controlled goods. However, in the case of quota-controlled goods, the variation in quantity is more relevant since the distortion is founded in the quantity available to the consumer rather than the price.
This is indeed the case in the distortion indexes fl' and #3C wherein the overall distortion rate is shaped by changes in the price or quantity of the individual goods depending on whether goods are price or quantity controlled. In addition, the distortion rate of each commodity is weighted by its contribution to the total welfare loss associated with all trade distortions in place. This aggregation of the individual distortions is explicitly founded in the underlying economic structure (balance of trade function) thus providing theoretical justification in the resulting distortion index. It follows that methodologically, the information on trade liberalization provided by the TRI approach, gains validity over more conventional ad hoc measures such as the PSEs/CSEs given its theoretically robust foundation.




CHAPTER 4
THE NEW GOODS PROBLEM AND THE TRI
A long time concern in index number theory has been accommodating for the disappearance of old commodities and the introduction of new ones into the market. The construction of a price index between two consecutive time periods requires data on the prices of the commodities in both periods. However in the case of a disappearing (new) commodity, its price is not observed in the second (first) period.
One theoretical explanation for the appearance (disappearance) of commodities in the marketplace is based on the concept of the reservation price. Conventionally the reservation price is defined as the price at which the consumer is just indifferent between purchasing or not purchasing a commodity. Hence, when a new good appears in the market it is inferred that its price in the previous period was at its reservation level resulting in quantity demanded equal to zero. Similar reasoning holds for the case of a disappearing good.
In practice, the problem of new (disappearing) goods has generally been circumvented either by setting the reservation prices equal to zero or by ignoring them in the period they appear for the first time (and including them into the index in the subsequent periods wherein price data become available). Diewert (1980) however 89




90
showed that both practices result in Fischer price indexes which are upward biased. Moreover, the bias is more severe when the reservation price is arbitrarily set equal to zero than when the new (disappearing) goods are simply ignored when they first appear (disappear).
Theoretically, the correct procedure for treating new (disappearing) goods is to form an estimate of the reservation price and then use this estimate in the price index. One solution to this problem proposed by Griliches (1961) is the hedonic (or continuous characteristics) approach. According to this approach the price of a commodity is viewed as some function of certain continuous characteristics of the commodity (e.g. in Griliches' seminal work on automobiles these characteristics included horsepower, length, weight, etc). Estimates of the price of a new good can then be obtained from hedonic regressions using data on the characteristics of the good from the periods that it is available.
Abstracting from the statistical and theoretical difficulties associated with hedonic regressions, the major draw-back of the hedonic approach is the fact that it can be implemented only if data on the characteristics of a commodity are available. Although a set of continuous characteristics may be easily identified in such industries like automobiles or computers, in many other cases such as agricultural commodities, it is most likely that the relevant characteristics of a product are either not identified or not readily measured. In such cases alternative techniques are needed for the treatment of the new (disappearing) goods problem.




91
The TRI in the Presence of New Goods
This Chapter generalizes the standard TRI to account for new goods (in the sense of newly traded goods), by utilizing the work of Feenstra (1990), and Feenstra and Markusen (1991) who adjust price, quantity, and productivity indices in the presence of new goods.
Their approach to the 'new goods problem' starts with the presumption that data on the characteristics of the goods under examination are not available; this implies that no reservation price can be estimated. To circumvent this difficulty, it may be assumed that the reservation price of a new good is infinite in the period(s) wherein the good is not available, and falls to some finite price when the good first appears in the market.
Given the assumption of an infinite-reservation price, the underlying utility (welfare) function may be represented by a C.E.S. functional form with elasticity of substitution greater than unity. In this case, the reservation price for any good is infinity, since quantity approaches zero only for arbitrarily high prices.
This approach leads to an adjustment of price (quantity) indices by a factor which depends on the expenditure share of the new goods when they first appear, as well as on the value of the elasticity of substitution. Thus the 'opportunity cost' of this approach is that instead of estimating reservation prices (which are infinity by assumption), one must estimate the elasticity of substitution.
Drawing on the theoretical results of Feenstra and Markusen (199 1), the standard TRI can be adapted to allow for different ranges of traded goods between any two




92
periods of time. The basic idea of such a generalization is to alter the reference utility (welfare) of the standard TRI.
To start with, if the same range of goods is available to the consumers over time then the utility levels for periods 0 and 1 respectively, are written as
u1(x;M) and u(x;M)
where x denotes the quantity vector of all goods (quota constrained and tariff-ridden goods) in the respective periods. The important point of this formulation is that welfare levels in periods 0 and 1 depend only on the quantities consumed (which may be different). However the number of goods in the welfare function remains constant.
In this situation, the appropriate TRI is that presented in Chapter 3. That is, taking the initial welfare u(x;M) as the reference point, the TRI is defined as the factor of proportionality 3 by which period-1 policy vector (X,Q) would have to adjust in order to make the balance of trade function hold for the period-0 welfare u(x;M0). Formally this TRI is defined as
P : BUn'I, C uO(x;M0); Z) = 0 (4.1)
Suppose however that the range of traded goods in period 1 is M' > Mo. Then the utility level in period 1 is U1(x;M'). If the range M' were available in period 0, then the corresponding utility would be u0(x;M) instead of uo(x;M).
Under this setting, the TRI may now be defined as a compensating measure in which the trade restrictions (w-,Q) of the goods which are common in both periods change so that u(x;M1), not u(x;M), is attained. To be more explicit, suppose that




93
there are M goods in period 0 and just one new imported good in period 1 (so that M' = MO + 1). Then the TRI can be defined as the equiproportionate change in the vectors (r,Q1) of the goods M, so that the economy returns not to the initial utility level u(x;M), but to the hypothetical level u(x;M').
For notational convenience let u(x0;M) i and u(x;M) u. Following the same reasoning as in defining the standard TRI, one can derive the offsetting uniform change in trade distortions (ir,Q) required to return the aggregate consumer to the utility level 9i (instead of the initial level u). Indeed, one may write
3'-3': z(, ; Z) = 0 (4.2)
This expression should be interpreted as follows: first, it equalizes the change to the foreign exchange requirement B(') due to an arbitrary change in ir, Q, and u to a hypothetical proportionate change in v and Q. Then, starting in period 1 it asks the question: what is the hypothetical proportionate change in (p,'r) which returns the economy to welfare level i rather than the actual level u?
This hypothetical proportionate change, which may be termed the generalized TRI, is determined by totally differentiating (4.2) and then solving for the rate of change
gpj ~ dQB~d a+ dig (4. 3)
B B;Q- 7c BQ B.ir B~,-~i
This generalized version of the TRI suggests that the change of rate of the proportionality factor 0' equals the sum of the percentage changes in the trade policy




Full Text

PAGE 1

TRADE RESTRICTIVENESS OF JAPANESE AGRICULTURAL IMPORT POLICIES By CHRISTOS J. PANTZIOS 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 1993

PAGE 2

ACKNOWLEDGMENTS My appreciation is extended to a number of people for their support during my doctoral studies. I would first like to express my gratitude to Dr. T. G. Taylor and Dr. E. Dinopoulos. As chairman of my supervisory committee, Dr. Taylor not only contributed to my education, but he also influenced my way of thinking and my ideas as an economist. His continuous encouragement, support, and friendship were important factors for me during my studies at the University of Florida and are deeply appreciated. Although the external member of my supervisory committee, Dr. Dinopoulos took a very close interest in my work and was instrumental in the inception and theoretical justification of my research. I would like to express my appreciation for his encouragement throughout my research effort, and also for his friendship. My deep appreciation also goes to Dr. L. Polopolus and Dr. W. G. Boggess for recommending my application at the Food and Resource Economics Department (F .R.E.) at the University of Florida and effectively initiating my Ph.D effort. Special thanks are also owed to Dr. Boggess for his unequivocal support as a member of my supervisory committee and also his friendship throughout my Ph.D. studies at the University of Florida. My sincere appreciation is also extended to the rest of the members of my supervisory committee, Dr. J. L. Seale Jr. and Dr. G. F. Fairchild, for their contribution to my research. ii

PAGE 3

I would also like to express my thanks to the staff of the System Support Center, and the secretarial personnel at the F.R.E. for their help and support during my studies. Last but not least, I would like to extend my thanks to my friends in Gainesville FL, for their concern and support during the last two years of my studies and especially to Ms Trinidad Reyes for her help, encouragement and moral support. iii

PAGE 4

TABLE OF CONTENTS ACKNOWLEDGMENTS . . . . . . . . ii ABSTRACT . . . . . . . . . . VI CHAPTER 1 PROBLEM STATEMENT AND OBJECTIVES OF THE STUDY Introduction ..................................... Problem Statement and Objectives ....................... CHAPTER 2 1 1 4 MEASURES OF PROTECTION . . . . . . . 7 The Average Tariff Measuring the Height of the Tariff 'Wall' . 7 The Nominal and Effective Rate of Protection, and the Domestic Resource Cost . . . . . . . . 10 The Uniform Tariff Equivalent (UTE). . . . . . 17 The Producer (Consumer) Subsidy Equivalent (PSE and CSE) . 18 PSE and Trade Distortions . . . . . . . 22 The Trade Restrictiveness Index (TRI) . . . . . 24 Operationaliz.ation of the TRI . . . . . . 26 The TRI in the Presence of Tariffs and Quotas . . . . 29 Comparisons Among the Measures of Protection . . . . 37 CHAPTER 3 THE TRI APPROACH IN JAPAN'S AGRICULTURE . . . . 41 Japan's Agricultural Trade Policies . . . . . . 41 The Beef Industry . . . . . . . . 43 The Pork and Poultry Industries . . . . . . 45 The Rice Industry . . . . . . . . 47 The Wheat Industry . . . . . . . . 49 The Fresh Citrus Industry . . . . . . . 51 Japan's Balance of Trade Function on Farm Imports . . . 52 Demand Elasticities A System-wide Approach . . . . 63 Computing the Combined TRI for a Subset of Japan's Farm Imports, 1971-1987 . . . . . . . . 76 iv

PAGE 5

Comparing the TRI with the Respective PSE/CSEs . . . 85 CHAPTER 4 THE NEW GOODS PROBLEM AND THE TRI . . . . . 89 The TRI in the Presence of New Goods . . . . . 91 Operationalizing the TRI in the Presence of New Goods . . 94 An Application of the Generalized TRI the Case of Japan s Meat Imports ; . . . . . . . . 100 Estimation of a C.E.S. Aggregator Function .......... ....... 104 Calculating the TRI of Japan's Meat Imports . . . . 111 CHAPTER 5 SUMMARY AND CONCLUSIONS ........................... 124 APPENDIX A PROOF OF THE PROPOSITION ........ ...... .. .......... 131 APPENDIX B STOCHASTIC SPECIFICATION OF A C.E.S. AGGREGATOR FUNCTION 133 APPENDIX C DESCRIPTION OF THE DATA AND THEIR LIMITATIONS ........... 142 REFERENCES ...................... .. ..... ........ 159 BIOGRAPHICAL SKETCH .. ............. .......... ... ... 165 V

PAGE 6

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 TRADE RESTRICTIVENESS OF JAPANESE AGRICULTURAL IMPORT POLICIES By Christos J. Pantzios December 1993 Chairperson: Timothy G. Taylor Major Department: Food and Resource Economics This study is an attempt to re-evaluate existing measures of trade protection and empirically investigate new ones in search of measures that aggregate satisfactorily diverse sets of trade distortions and provide a valid estimation of trade restrictiveness across countries and over time. A review of the existing literature reveals that from a methodological point of view, the Trade Restrictiveness Index (TRI), has some clear advantages over the rest of the trade protection measures The study implements the TRI by investigating the overall restrictiveness of a subset of Japan s agricultural policies. The level of trade protection is measured, ceteris paribus, by means of the TRI for the following farm products: (1) beef, (2) pork (3) poultry, (4) wheat, (5) rice, and (6) fresh oranges. The computed index indicates a dramatic rise in trade protection in the early 1970s, followed by a slow process of easing vi

PAGE 7

price and quantity restrictions so that a slight trade liberalization trend appears towards the end of the examined period. This study also attempts a generalization of the standard TRI by deriving the index in the presence of different sets of traded goods over time. Under the maintained assumption of a C.E.S. welfare function, it is shown that when newly traded goods are introduced over time, the standard TRI is adjusted by a factor that depends on the relative expenditure value of the new goods, and the elasticity of substitution of the underlying C E S. welfare function. This generalized version of the TRI is empirically investigated by calculating a partial TRI pertaining to Japanese meat imports under the maintained hypothesis that beef imports are differentiated products, distinguished by origin (supplier country). The standard TRI is adjusted in 1969 to account for the introduction of U S. beef in the Japanese meat market. This exercise shows that the TRI may be shaped by both policy changes and cross-commodity effects and as a result the adjustment in the generalized TRI may be either upwards or downwards. vii

PAGE 8

CHAPTER 1 PROBLEM STATEMENT AND OBJECTIVES OF THE STUDY Introduction The rapid growth of international trade during the last two decades has resulted m an increasing interdependence of the world's economies. The globalization of markets however, has been often accompanied by heavy government intervention This is especially true in agricultural trade where government intervention is the rule rather than the exception. In attempting to insulate their farm sectors from the variability of world markets, governments have devised national agricultural policies whose diversity and complexity obscures the global interdependence of agricultural markets. Nowhere is this clearer than in the contrast between the agricultural policies of the European Community (EC) countries and the U S and the resulting declining export power of the US farm sector. To resolve this conflict, trade in agricultural products has been included for the first time in the Uruguay round of the General Agreement on Tariffs and Trade (GA TT) negotiations. A major objective of the negotiations is to achieve a progressive reduction of government assistance to agriculture in developed countries. Given the complexities and regulated environment of agricultural sectors, evaluation of the level of protection ( overall trade restrictiveness) poses a challenge for 1

PAGE 9

2 both theorists and applied economists. In principle, a straightforward way to carry out such an evaluation is through an aggregate scalar measure ( one that combines the various types of distortions and interventions). The resulting index would provide a useful and relatively simple means for international and intertemporal comparisons of trade protection. However, the use of an such an aggregate measure may easily result in miscalculations of the degree of protection. In order to quantify the effects of diverse and often overlapping policies one must cope with a number of methodological and practical issues. Schwartz and Parker ( 1988) outline at least four criteria that an ideal aggregate measure of trade protection should satisfy. First, the measure should be a consistent aggregation across products, countries, and over time so that the resulting measurements can be meaningfully ranked and compared. Second, the measure should be relatively simple so that it is easily understood. Third, it should be flexible enough to capture the effects of diverse policies. Fourth, it should make the effects of trade distorting policies transparent, thus separating trade distortions from inefficiencies in resource allocation. Measuring the magnitude of the distortions which government policies inflict on the trade flows of a country has long been a concern for economists. Over the years, researchers have developed various measures designed to quantify in a scalar measure the level of trade distortion. At the same time, there has been a continuous evolution of trade policies as global economic interdependence has progressed and the importance of the international marketplace was recognized.

PAGE 10

3 In agriculture, one may observe two major evolutions related to trade policy tools. First, there has been a shift towards quantitative restrictions instead of tariffs, as the predominant means of protection. These range from quotas to various supply control schemes and marketing orders. Second, government assistance to the farm sector has taken numerous forms ranging from direct income transfers to structural policies such as research and extension expenditures. While some of these cause distortions to trade others cause distorting effects only to the domestic market. Thus any particular good may be affected by a number of different policies. This evolution of trade policies and the inclusion of agricultural issues in the GA TI negotiations have revitalized the interest of economists and trade negotiators in obtaining reliable indicators of trade protection. A comprehensive quantification of trade restrictiveness can provide a sound basis for trade negotiations and subsequently a consistent monitoring of the trade liberalization process. Thus an examination of the relevant existing measures on trade protection has been initiated in recent years, along with a search for new ones which can better cope with the new realities of agricultural protection. Separately, the growth in international trade has revived the importance of the 'new goods-problem' as trade among the world's economies involves different sets of goods over time. Indeed, observed trade patterns suggest that new products play an increasingly important role in international trade. Product differentiation leads to the development of intra-industry trade and the importation of new varieties of goods across countries. It is also common practice for countries to change the trading status of goods

PAGE 11

4 by liberalizing old goods and protecting newly developed ones; thus the list of goods subject to trade restrictions differs over time. In addition, regional or global trade agreements (such as the North American Free Trade Agreement (NAFT A) or a possible future GA IT agreement on liberalizing trade in farm products) are expected to facilitate the importation of goods which were not traded earlier (or were traded in negligible quantities) due to trade impediments. In the area of agricultural trade, the 'new goods problem' can be linked to the Armington assumption (1969) which is often used in empirical work. The Armington assumption states that imports and domestically produced goods are imperfect substitutes in consumption or production. Also exports are imperfect substitutes for domestically consumed goods. Such differentiation is rationalized on the grounds of packaging, taste, safety requirements etc. --something which is very relevant in the case of trade in certain agricultural products. As the liberalization process of a particular country develops, agricultural imports from countries trading for the first time in certain products with the country in question can be viewed as a special case of the 'new goods problem'. Problem Statement and OQjectives The preceding discussion makes clear that the empirical measurement of the size of trade distortions remains a complicated, but timely task. The intent of this study is to contribute to the attempts being made to evaluate and generalize aggregate measures that satisfactorily combine diverse sets of trade distortions and provide valid estimation of the trade restrictiveness across countries and over time.

PAGE 12

5 Among the available aggregate measures of protection the Trade Restrictiveness Index (TRI) developed recently by Anderson and Neary (1991) seems to be a particularly promising approach in evaluating trade distortions. This study reviews the existing measures of protection and subsequently focuses on the TRI approach in an attempt to examine empirically the performance of this newly developed protection index and furthermore investigate the effects of trading in new products on the measured level of trade protection. The first objective of this study is an empirical implementation of the TRI approach. Given that the TRI is a recently developed concept in the measurement of trade protection very few empirical constructions of the index exist (Anderson, 1991; Anderson and Bannister, 1991) In particular this study applies the TRI approach in the agricultural sector of the Japanese economy in an attempt to subject the concept of the TRI to empirical testing and investigate its applicability; compute a measurement of trade restrictiveness which has a robust theoretical derivation and explicitly incorporates considerations missing from other standard measures, notably both production and consumption side interventions cross commodity relationships, and quantity restrictions; and examine how various government interventions peculiar to agriculture can be modelled and subsumed in one parameter-measures.

PAGE 13

6 The second objective of this study is to generalize the TRI to account for the fact that different sets of goods may be traded over time, in an attempt to incorporate the 'new goods-problem' in the measurement of trade restrictiveness. An overview of this study is as follows. Chapter 2 provides a review of the literature on trade protection measures with particular emphasis on the TRI, and then a comparison of the available measures. Chapter 3 presents an empirical application of the TRI in Japan's agricultural sector. In particular, Japan's government policies on beef, pork, poultry, wheat, rice, and fresh oranges during the period 1970-87 are considered, and the respective partial equilibrium TRI is constructed. The index aggregates the government programs in all six commodities in a scalar measure and provides an estimation as how the overall level of protection on these commodities changed over the examined period. Chapter 4 provides a discussion of the 'new goods' problem in index number theory and demonstrates how the standard TRI can be generalized to account for the presence of newly traded goods over time. Subsequently, it presents an empirical exercise on this generalization of the standard TRI. Specifically, a TRI pertaining only to Japan meat imports is constructed with the additional assumption that beef imports from the three major supplier countries (Australia, New Zealand, and the United States) are considered as different varieties of beef. Since U.S. beef imports were negligible in the 1960s, the meat TRI is computed for the period 1969-87, and adjusted in 1970 for the introduction of U.S. beef. Finally, conclusions and remarks are summarized in Chapter 5.

PAGE 14

CHAPTER 2 MEASURES OF PROTECTION The methodologies that have been used over the years to quantify the degree of trade protection reveals that the major protection measures can be classified into four groups: (a) the Average Tariff; (b) the Nominal and Effective Rate of Protection (NRP and ERP), and the Domestic Resource Cost (DRC); (c) the Producer/Consumer Subsidy Equivalent (PSE/CSE); and (d) the Trade Restrictiveness Index (TRI). A discussion of the merits and shortcomings of these measures is the objective of this chapter. The Average Tariff Measuring the Height of the Tariff 'Wall' Early attempts to measure the effect of protection on trade flows focused on tariffs (League of Nations 1927, Crawford 1934). In particular, researchers devised various methods to compute the overall height of the individual tariff levels (in the sense of a tariff 'wall') for the purpose of international comparisons across countries or intertemporal comparisons of a particular country. The procedures which have been generally employed involve the calculation of the percentage of all import duties to a 7

PAGE 15

8 certain basis (e.g. the value of all imports) or equivalently, the calculation of weighted and unweighted averages of import duties. 1 It was early recognized, however (Loveday 1929, Haberler 1936), that such computations have grave theoretical pitfalls and thus they can be quite misleading. When the height of the (aggregate) tariff is computed as the percentage of all duties collected over the value of total imports, one gets the rather absurd result that a more protective tariff regime yields a smaller percentage (i.e. a lower tariff 'wall'). This is because as duties become more protective, the value of total imports declines. In fact, if all duties became prohibitive, imports would be zero and the tariff 'wall' would be also zero. The same fallacy applies to the measurement of the tariff 'wall' as the percentage of imports subject to tariffs over all imports. Calculations of a weighted average of all import duties faces the same grave objections. The basic difficulty is associated with choosing the proper weights of an average tariff rate. It is clear that weighing the various duties by the value shares of their own imports produces the same sort of distorting results. Low duties correspond to high relative levels of impoits, thus they are given large weights; high duties are given small weights and prohibitive duties are given zero weights. To circumvent these difficulties, various alternatives have been suggested (Haberler 1936, Lerdau 1957, Balassa 1965). A possible alternative would be the share 1 Unweighted averages are obtained by computing a simple average of the import duties upon total imports. Weighted averages are obtained by assigning different weights on the duties of individual imports. For an empirical application see, Research and Policy Committee of the Committee for Economic Development, 1964.

PAGE 16

9 of the imported good in total exports. Others include the share of the imported good in the volume of the world trade or its share in the volume of production ( or consumption) in one or more countries. Nonetheless, even the weights based on these shares can be biased given the volume of world trade, and domestic production and consumption are influenced by the import duties already in place. On the other hand, the calculation of unweighted averages of import duties fails to take into account the relative importance of individual imports. Besides the aforementioned difficulties in computing a reliable average of all import duties, there is a great deal of ambiguity surrounding the very concept of 'the height of the tariff. In other words, even if all calculation problems are overpassed, it is still not clear what is to be established or indicated by such a measurement. It is only clear that an estimate of the average tariff alone by no means provides an assessment about the degree of a country's total protection. Several reasons are cited in the literature (Bieda 1963, Towle 1956). First, there is a variety of other equally important measures upon which the protection of a country may rest (e.g. quantitative restrictions) that are not incorporated in the average tariff. Second, even if all trade restrictions were expressed in terms of a tariff, the true degree of a country's protection would still not be revealed. Different countries are expected to have different elasticity of demand for an imported good and different elasticity of domestic supply. Thus, any particular level of tariff will yield a less protective effect when the elasticity of domestic supply is low rather than high. In tum, elasticity of demand for imported goods depends on tastes, availability of close substitutes etc., while

PAGE 17

10 elasticities of domestic supply depend on a country's resources and technological constraints. Third, the degree of protection of final goods may well be affected by protectionist measures imposed on the respective inputs (raw materials and intermediate goods). Consequently, there is a clear difference between nominal and effective rates of tariffs wherein the latter include all duties levied on inputs. This last consideration led researchers to develop the concept of the Effective Rate of Protection (ERP) which examines net or effective rather than nominal tariff rates. The Nominal and Effective Rate of Protection, and the Domestic Resource Cost The idea that a distinction should be drawn between the nominal and the effective protection of an economic activity was first developed by Barber (1955) and further elaborated by Johnson (1960), Humphrey (1962) and Corden (1963, 1966, 1971, 1985) among several others. Large-scale empirical contributions of the concept of effective protection are given in Balassa (1965) and Basevi (1966). The effective rate of protection deals with the true or net rate of protection associated with an economic activity which produces a.final or value-added1product by using purchased material inputs that are themselves traded, and thus may be subject to distorting government policies. In this case the question becomes: Does the nominal protection on the final product indicate the true protection rate of the associated economic 2 Value added is the value of the final output less the cost of purchased intermediate inputs.

PAGE 18

11 activi t y, and if not, how is the protection of the final product affected by government polici e s on its intermediate inputs? Formally, the ERP is defined as the percentage change in the value-added per unit (or effective price) of an economic activity with and without the existing government intervention. Thus, algebraically the ERP of the jth economic activity is given by the ratio where = vf v}' v}' ( 2. 1) v 0 i = the value-added per unit of the final product under the existing protective structure, vwi = the value-added per unit of the final product in the absence of any distortions (i.e. under free trade). The ERP so defined is based on the following assumptions: (i) the ratios of intermediate inputs to output (i.e. what is often called the physical input-output coefficients) are fixed 3 and identical for all firms, (ii) a small country-framework is adopted implying the elasticities of demand for all exports and supply for all imports are 3 The fixed input-output coefficients assumption needs to be interpreted with caution as it does not mean that there is a fixed coefficient production process. The basic idea is that the various value-added production functions are functions of purchased intermediate inputs, as well as, primary factors such as, capital labor and so on. Then assumption (i) still permits substitution among primary factors along isoquants but rules out substitution among intermediate inputs or intermediate inputs and primary factors. It is shown that calculation of ERPs always tends to overstate the true effective rates, if there are indeed substitution relationships among the intermediate inputs or intermediate inputs and primary factors (Corden 1971, 1985).

PAGE 19

12 infinite, (iii) all tradeable goods remain traded even after tariffs and other distortions have been imposed, so that the domestic market price of each importable is simply given by its international price plus tariff. 4 The algebraic formula for the ERP of the jth economic activity can be further elaborated. Consider first the simple case of a traded final product j, which has a single intermediate input i, subject to a tariff, in its value-added production function. Let: a;j = the physical input-output coefficient of the intermediate input i, under free trade, = the nominal tariff rate on the final product, t; = the nominal tariff rate on the intermediate input, pwj = the world price of the final product. Then by definition, it holds Combining equations (2.1), (2.2) and (2.3) yields (2.2) (2.3) 4 Removing this assumption implies that part of a tariff may be redundant. Since a redundant tariff has no effect of any kind, all calculations should then be based on the utilized part of the tariff-this could require detailed price data not always available.

PAGE 20

13 ERP = .1 .1 = --t .1; t 1 tJ a J t ( 1 ) ( ) J 1 a iJ 1 a iJ J 1 a iJ (2.4) Thus the ERP is a combination of two effects. The first, given by the term V(l a;) is the proportional increase in effective price (value-added per unit) v 0 i, resulting from the nominal tariff on the final product. The second, given by [a;/(1-a;i)l is the proportional fall in the value-added per unit, resulting from the tariff on the intermediate input i. Clearly, an increase in the tariff rate imposed on input i reduces the rate of protection on the final product and vice versa. Furthermore, let Then and a\ = the physical input-output coefficient of intermediate input i, under the existing distortions. ( 2. 5) 1 a! .1] 1 = tj ti I (1 + tj) (1 + t1> a1J = ( 2. 6) I 1 a1J ( 1 + tj) (1 + t)

PAGE 21

14 Equation (2.6) is a 'deflated' version of equation (2.4) expressing the ERPj in terms of the distorted input-output coefficient a\ instead of the undistorted (free trade)
PAGE 22

15 The NRP 0 measures the percentage difference (i e. the 'wedge') between the domestic consumer price and the world (free-trade) price. Algebraically it is defined as the ratio (2.8a) where pci is the domestic consumer price. Correspondingly, the nominal rate of protection to producers (NRPP) measures the percentage difference between the domestic price received by the producers and the world (free-trade) price. Thus it is given by the ratio D w NRP = Pi Pi P1 w Pi (2.8b) Assuming ad valorem tariffs on final and intermediate goods, the ERPi in equation (2.4) can also be expressed as n NRPP1 L aij NRPc i=l 1 ERPi = ___ ___;;;....;;. ___ n (2.9) 1 I:; aiJ ll Thus the ERPi can be equivalently interpreted as a weighted average of the relevant nominal protection rates in the production of the final good j and the consumption of its intermediate inputs i = 1 ... n Finally, the ERP bears also similarities with the concept of Domestic Resource Cost (or DRC). The DRC was developed by government planners in Israel during the 1950s as a means of project evaluation under conditions involving distorted official

PAGE 23

16 exchange rates and distorted prices of tradeable goods (Bruno, 1972). In a general sense, the Domestic Resource Cost per unit of the ith economic activity, DRC;, is defined as the ratio NVAi (2.10a) where the numerator, DC;, is the total value added of the domestic resources per unit of output employed in the ith activity, measured at opponunity cost. The denominator NV A; is the value added per unit of output of the ith activity, measured at world prices (i.e. its international value added) Originally, the DRC method was used as a normative ex ante criterion of social comparative advantage, in ranking development projects. That is, the DRC criterion was used as an indicator of ranking future investments according to the real cost of net foreign exchange earned or saved. However, the DRC can also be viewed as a measure of the opportunity cost that a country incurs in order to sustain its existing import substitutes. In this sense, the DRC can be interpreted as an index of the social cost of trade protection and a means for evaluating the magnitude of trade distortions. From the definition of the DRC, it is clear that both the ERP and DRC involve value adding activities in their measurement. A comparison between the two reveals that the ERP and DRC are in theory identical under only exceptional circumstances. In particular, it is shown (Krueger, 1972) that the following relationship is true (2.10b)

PAGE 24

17 under these stringent conditions: (1) all goods are traded (or tradable), (2) there are no transportation costs, (3) resources are perfectly mobile within the domestic economy but perfectly immobile international! y, and ( 4) all domestic output/input markets are perfectly competitive. It follows that the ERP may be a sufficient measure for economies where tariffs are the predominant impediment to trade and factor markets are fairly competitive. For economies where, in addition to tariffs, there are quantitative restrictions, institutional constraints (e.g. government involvement in trade-related sectors), and market imperfections ( resulting in prices which do not reflect the true opportunity cost), the DRC has a conceptual advantage over the ERP. In these cases, evaluation of domestic resources at imputed shadow prices rather than market prices is a superior method to encompass as much as possible non-price distortions in measuring the cost of protection. This significant advantage comes, of course, at a higher information cost. Implementation of the DRC method requires besides input-output coefficients, shadow values for the domestic inputs which are usually estimated from programming models. The Uniform Tariff Equivalent (UTE}. Based on the concept of ERP, Corden ( 1966) defines a single aggregate measure of the various individual tariffs, which he calls the Uniform Tariff Equivalent (UTE). The UTE is defined as the uniform tariff which (if applied) would keep the value of imports at the same level as the existing (nonuniform) tariffs. In other words, the UTE

PAGE 25

18 is the uniform tariff which is equivalent to the existing tariffs in its effect on the total value of imports. By considering imports of two final goods and linear demand curves, Corden shows that the UTE is a weighted average of the ERPs on these two final goods; the respective weights are combinations of their demand elasticities, import values under free trade and input-output coefficients. The UTE is then compared with average tariffs, calculated as weighted averages of all individual tariffs by using as weights either actual imports in the presence of distortions or domestic output. The diversion between the UTE and the average tariff is examined under various conditions on demand elasticities and input-output coefficients. These exercises provide a useful insight on the important factors ignored by the average tariff-approach mainly, demand elasticities, traded inputs and the reference point of the comparison. An additional empirical application of the concept of the UTE appears in Balassa ( 1965). The Producer (Consumer) Subsidy Equivalent (PSE and CSE) Developed by Josling (FAO, 1973; PAO, 1975), the notions of the producer (consumer) subsidy equivalent (or PSFJCSE approach) are designed to provide an aggregate measure of government support policies in a particular sector of the economy. The concept of the PSE is straightforward. It is the subsidy that would be necessary to replace all current government policies applied to the agricultural sector of a particular country in order to leave the producer's income unchanged. Thus the PSE

PAGE 26

19 measures total income transfer resulting from any policy that can be linked to incomes and it can be computed at any level of govemment--local regional or national. The consumer subsidy equivalent (CSE) is defined in a symmetric fashion. Several features of the PSE deserve particular attention. Unlike the measures mentioned above, the PSE combines both price and nonprice policies (ranging from import quotas to direct payments to farmers, disaster payments and so on). Like the previous measures, however, the PSE is commodity-specific, evaluated as the absolute sum of money received as support by the producers of that commodity. Then it can be expressed in relation to several bases: (i) PSE per unit of output (i e. PSFJvolume of quantity produced) (ii) PSE as a percentage of domestic production valued at domestic prices (iii) PSE as a percentage of domestic production valued at world prices (iv) PSE as a percentage of actual net farm income (thus serving as an indicator of income dependency). The definition of PSEs and CSEs is inherently flexible. One may decide to include or exclude various government programs. Thus, while the first calculations of PSEs and CSEs included only commodity specific policies, a recent OECD study (OECD, 1987) broadened the policies covered in the PSE to include structural support programs, which are not necessarily commodity specific (e g. research and extension). Additionally, the Economic Research Service (ERS) of the U.S. Department of Agriculture (USDA) has extended the OECD measurement of the PSE to include the effects of exchange rate distortions in the case of developing countries (USDA ERS, 1988).

PAGE 27

20 Concerning the estimation of PSEs (and CSEs), it must be emphasized that different countries have quite different sets of policies in their agricultural sectors. Thus a standard framework is needed for comparisons across countries and across commodities 6 Several important points must be noted on the use and the interpretation of PSEs and CSEs. By aggregating a variety of government policies into a single measure, the PSE and CSE allow comparisons of government support across countries, commodity markets and types of policies. Additionally, they can indicate which forms of government support are most important in different countries and, when examined over time, they can indicate the intertemporal changes of government support. As mentioned earlier, international comparisons of the PSE and CSE require a standard procedure and a common set of the policies covered. To the extent that policies excluded from this common framework provide a significant amount of support, the resulting comparisons may well be biased. Additionally, in intertemporal comparisons a country may change its government policy profile towards (or away from) the set of policies covered by the PSE (CSE). Therefore interpreting comparisons of PSEs (CSEs) requires consideration of these issues. It is erroneous to conclude that, if all governmental programs were removed, incomes would decline by the value of the transfers estimated by the PSE. Income in 6 In addition to estimating PSEs and CSEs for each commodity, 'pooled' PSEs and CSEs are also calculated and compared across countries (USDA ERS, 1987). Theses are weighted averages of the individual PSEs weighted by each commodity's share with the total production value of the covered commodities.

PAGE 28

21 the absence of the existing government programs would depend on the new levels of prices, production, consumption and trade. PSEs only measure the total transfer to producers under current policy and market conditions. The PSE and CSE measures do not capture the large-country effects on the world prices (i.e. they are implicitly based on a small-country framework). Additionally, PSEs and CSEs do not reveal the distributional effect of government programs within a particular sector (e.g. PSEs cannot show whether total transfers in the dairy industry are received equally by all dairy producers). Changes in the PSEs and CSEs can be merely due to changes in the world reference prices or controlled exchange rates--both of which may be quite variable over time. Almost all traded goods are priced in U.S. dollars; thus when the dollar appreciates, the world reference price observed by countries other than the U.S. rises (and vice versa). Then for countries supporting the producer price above the world price, the price 'wedge' is now narrower than it would have been under constant U.S. exchange rates (unless their agricultural policies are responsive to world prices). Therefore changes in the magnitude of the PSE (CSE) can be markedly affected by controlled fluctuations in exchange rates. The PSE does not directly measure the effect of supply control policies (e.g. uncompensated acreage reduction programs). An effective supply-control program

PAGE 29

22 reduces production from what it would have been in the absence of such program ; therefore total government transfers (and thus the PSE) are lower under the program. 7 Finally and most importantly, government policies which yield the same PSE/CSE across countries do not necessarily imply that these countries have the same degree of trade distortions Different types of government policies produce different trade effects, 8 while the effect of government programs on trade depends on the country s share in international trade. Additionally, producers and consumers across countries respond differently to the same type of government intervention due to political and social factors, market characteristics, and resource and technology constraints. PSE and Trade Distortions Given the fact that the link between government support policies and the impact of that support on trade can be very weak,9 the use of the PSE/CSE as a yardstick of trade liberaliz.ation has come under criticism. It is argued that if the PSE/CSE is used in trade negotiations countries may try to achieve lower levels of PSEs (CSEs) by 7 However, supply control programs may be implicitly taken into account in the PSE/CSE measures. For example, if a country is large enough to affect world prices, its supply control programs will raise world prices The price-effect of the supply control program is then captured in the PSE/CSE calculations if the corresponding price data are used. Tangermann et al (1987) propose practical ways for countries employing supply-control programs to receive negotiating 'credits' on the PSEs for the commodities concerned. 8 For example deficiency payments stimulate domestic production but leave consumption unaffected; import quotas also raise domestic production but at the same time, reduce demand. 9 An illustrating diagrammatic exposition of the differences between support measures and measures of trade distortion appears in Roningen and Dixit (1991).

PAGE 30

23 merely eliminating government programs which bear little relationship to trade, thus leaving their trade barriers intact. To cope with this deficiency, economists have suggested versions of the PSE including only policies which are likely to affect trade. Thus Schwartz and Parker (1988) proposed the modified PSE as a more appropriate measure of trade distortion. The modified PSE includes only policies with well-defined price effects which are more likely to reflect on trade; it excludes policies with ambiguous price effects such as structural programs. Rondingen and Dixit (1991) proposed a more direct measure of trade distortion, termed the Trade Distorted by Support-Index (or TDS). The TDS index is a volume measure 10 of distorted trade. In particular, it measures the change in the volume of the existing net trade if a country eliminates completely all support to a commodity The TDS in volume terms is generally expressed as = [domestic market support] + [direct payments to farmers/consumers] + [other farmer support] [offsets to support] (2.11) where e., ed = own-price supply and demand elasticities, respectively, 10 However, the TDS can be readily expressed in percentage form (i.e TDS/[volume of production (consumption)]) or in value form (VTDS) by multiplying the TDS by the world (border) price.

PAGE 31

q., qd = observed production and consumption quantities, respectively, sm = market support ratio (i.e. the support level per unit of the commodity compared with domestic prices), sci sP = direct support rate to farmers and consumers, respectively, s; = support ratio for all other types of assistance to producers, and sso = set-aside offset policies, usually policies that require production (consumption) discipline for farmers (consumers) to be eligible for direct payments 24 The primary contribution of the modified PSE and the TDS is that they are more transparent measures of the trade distorting effects of government policies than the PSE. Methodologically however they share a lot in common with the PSE/CSE approach. Hence they may be viewed as 'trade-oriented' versions of the PSE 11 The Trade Restrictiveness Index (TRI) Unlike the previous protection measures which lack a clear theoretical basis, the TRI (Anderson and Neary, 1991a; 1991b; 1992) is explicitly based on the underlying economic structure of a trading economy. Conceptually, the TRI is a general equilibrium distance function measure (Debreu 1951; Deaton,1979) for a trading economy. Conventionally the distance function is defined with some reference utility level u and an arbitrary quantity bundle 11 The modified PSE is literally a trade version of the PSE; the TDS can be viewed as a volume-(trade) version of the PSE.

PAGE 32

25 q as the factor by which the bundle q must be deflated ( or inflated) in order to attain exactly utility u. Thus the distance function is a measure of the inefficiency of q relative to u. The TRI utilizes the same notion in the space of trade distortions; in the case of only tariffs this space is the space of the domestic prices of the traded goods while in the case of only quotas, it is the quantity space of the traded goods. Moreover, instead of a utility function, the starting point in the case of the TRI is the (general equilibrium) budget constraint of a trading economy. Such a function equates aggregate expenditure on traded goods to a gross revenue or GNP function and trade revenues (tariffs and quota rents). This permits the aggregate utility (welfare) level and the instruments of trade distortion (i.e. tariffs,quotas) to be subsumed conveniently in a single function called the balance-of trade function. Based on the economy's balance-of-trade function the TRI is defined as a compensating variation measure of the trade distortions towards a reference utility (welfare) level. Specifically, given a final period 1 and an initial period O the TRI is defined as the uniform rate by which the instruments of trade distortion (tariff rates, quota levels) must be deflated (or inflated) in order to take the economy back to the initial welfare level. Similar to the distance function therefore, the TRI can be viewed as a measure of the trade inefficiency associated with a set of trade distortions, relative to a reference welfare level.

PAGE 33

26 Operationalization of the TRI The TRI stems from the same theoretical reasoning as the cost-of-living-index or Consumer Price Index (CPI). To demonstrate this similarity a short digression on the concept of the CPI will first be presented. The CPI is the uniform rate of change in all consumer prices, which produces an equivalent rise in the expenditure required to maintain the same level of utility for a representative consumer. Formally the CPI is based on the consumer's expenditure function e(p,u) where pis the vector of consumer prices u is the consumer's utility level and e( is the minimum level of income required to achieve the utility level u when the consumer faces prices p. Since e(p,u) is an optimal value-function, its gradient with respect to p gives the vector of the consumer's demands, X. Given arbitrary changes in prices p, the level of income required to support utility u is X'Vp, where Vp is vector of first order partial derivatives (i.e. the gradient vector) of the price vector, p. If the price vector p were proportional to its initial level by some factor a (i.e. p = p 0 a where p 0 is the initial price vector) then the level of income required to support utility u, given a change in the prices p, is X'p 0 da. Then the uniform rate of change in all consumer prices which creates the same rise in the required expenditure as the arbitrary change Vp is X 1 p 0 da. = X 1 'ilp da. = x' 'ilp x'po (2.12) where a hat(") over a variable denotes a proportionate change, i.e. Vp/p. For an initial value of a= 1, da is also a percentage change and represents the CPI.

PAGE 34

27 At this point it must be stressed that the preceding formula of the CPI (bearing the opposite algebraic sign) can be obtained as a solution to the following problem: what is the (hypothetical) uniform reduction in consumer prices (instead of income transfer) which compensates the consumer (i.e. preserves initial utility u) given some arbitrary change in prices dp. As earlier drl' = x' 'vp = -~(X1P1)P1 (2.13) X 1 p 0 f:t X 1 p where the minus sign denotes the direction of the compensatory change in prices p. The notion of the TRI is rooted in the same reasoning as the compensatory price change da'. This is however where the similarities between a consumer and a trading economy end. The budget constraint of a price taking consumer links his expenditure to his monetary income. By contrast the budget constraint of a trading economy links the aggregate expenditure on all goods (domestic and imported) to the value of the national product (GDP) and the government revenues resulting from trade intervention (i.e. tariff revenues and quota rents). Thus a change in the tariff vector causes not only a change in aggregate expenditure but also a change in government tariff revenue. Consequently, the starting point in the derivation of the TRI is the economy's budget constraint. The specification of this budget constraint can be conveniently carried out using standard dual techniques (e.g. Dixit and Norman, 1980). In particular, using dual functions the consumption and the production sectors of a trading economy can be modelled in terms of an aggregate expenditure function e(7,u) and a revenue (or GNP) function g(7,v), respectively (where 1r denotes the price vector of traded goods, u

PAGE 35

28 aggregate utility or welfare, and v the vector of resource endowments). Given the fundamental requirement that imports must be paid for (i.e. trade must balance), the budget constraint of a trading economy becomes e ( 1t u) g ( 1t v) = b ( 2. 14) where b denotes the value of net imports; a positive b implies a trade surplus while a negative b implies trade deficit; for the purpose of this analysis, b is considered to be exogenous. Moreover, the above equilibrium condition can be re-written in terms of the function B(1t,u,v) = e(1t,u) g(1t,v) P = 0 ( 2. 15) which is termed the balance-of-trade function. Essentially, this function represents the overall equilibrium condition of the economy in each time period and can be expressed as B(1t, u; z) = O ( 2. 16) where z represents all exogenous variables, such as factor endowments international prices of traded goods, tastes, etc. The balance of trade function B( can be specified to take into account the fact that trading economies impose price restrictions, or quantity constraints, or both on their imports (exports). Below, the balance of trade function (and subsequently the TRI) is defined with respect to an economy imposing both tariffs and quotas (the cases of only

PAGE 36

29 imports or only quotas may be considered as special cases and can be found in Anderson (1991)) The TRI in the Presence of Tariffs and Quotas Consider a competitive trading economy imposing both tariff and quota restrictions, wherein all goods are tradeable (non-traded goods can be considered in the background as quota-constrained goods whose quota levels are set equal to zero). The vector of goods subject to tariffs is denoted by m and the respective price vector by 1r Quota levels are denoted by Q and the price vector of quota-constrained goods is p Finally, the world prices for the tariff-ridden goods and quota-constrained goods are p* and 1r*, respectively. The economy under examination is assumed to be small (price-taking) so that p* and 1r* are exogenous to the analysis. To facilitate the analysis, one can define E(p,1t,U) = min 0 ,m [p 1 Q + 1t 1 m I U(Q,m) = u] (2.17) as the trade expenditure function E(p,r,U). In words, E() is the expenditure of the aggregate consumer in the traded quantities (net imports) of all goods and has the standard properties of an expenditure function: it is linearly homogeneous and concave in (p, r) and its derivatives with respect to prices are the economy's excess demand functions i.e.

PAGE 37

30 En ( 1t p, u) = m ( 1t p, u) EP = Q(rt,p, u) (2.17a) However, assuming that the quotas are binding the aggregate consumer can be regarded as minimizing expenditure over the tariff-ridden goods only. Thus optimizing behavior can be described equally well by the distorted trade expenditure function, defined as E(Q,1t,U) = minm [1t'm I U(Q,m) = u] ( 2. 18) Alternatively one may write E( Q, 1t, U) = maxP [E(p, rt, u) p 1 Q] (2.19) and its first derivatives yield E1eQ,1t,u) = E1e[p(Q,1t,u),1t,u] = : ffi [p (QI 1t I U) I 1t I U] (2.20a) E 0 ( Q, 1t u) = P ( Q, 1t u) (2.20b) The derivative (2.20a) follows from Shephard's Lemma since the distorted expenditure function is the minimum-value function for tariff-ridden goods. The derivative (2.20b) implies that relaxing the quota by one unit reduces the expenditure of

PAGE 38

31 tariff-ridden goods by -p. Thus it gives the consumers' marginal willingness to pay for quota-restrained goods. Turning to the quota rent(s) (p p*)'Q, it may be recognized that usually these rents are shared between the home country and its trading partners. 12 Let w denote the fraction of the quota rents accruing to foreigners (supplying countries); then the quota rent retained in the home country is (1 w)(p -p*)'Q. The budget constraint of such an economy (assuming away any trade balance surplus or deficit) in domestic prices, is given by E(Q,p,U) + p 1 Q = t 1 m + (1-w) (p-p*)'O (2.21) In words, the left-hand side equals the expenditure of the aggregate consumer in all goods. In equilibrium this must equal the revenues from the trade restrictions retained at home. These are given by the right-hand side. More conveniently this equilibrium condition can summarized in terms of the balance-of-trade function, defined as 12 The fraction w may be interpreted as the fraction of quota licenses awarded to foreign importers. A common practice in international trade is to impose tariffs on goods already constrained by quotas. In this case w is also equal to the tariff rate imposed on quota controlled goods, under the assumption that all imports are handled by foreign importers.

PAGE 39

32 B(Q,1t,u;y) = E(Q,1t,u) + p 1 Q t 1 m (1-w) (p-p*) 1 0 = 0 (2.21a) As noted earlier B( summarizes the equilibrium condition of the home economy. It may also be interpreted as the net foreign exchange requirement (in the sense of total income), necessary to support the utility level u for the aggregate consumer facing given levels of tariffs or quotas. Consider now a relaxation in both trade restrictions of this economy (i.e. the tariffs rates and the quota levels). Then the compensating income transfer required to move the aggregate consumer back to the initial utility u, is (2.22) Assume now that the prices of tariff-ridden goods 1r, and the quota levels Q are proportional to their initial levels at the same proponionality factor P (i.e. Q = Cf#, -r = -r 0 P, where 1r 0 Q 0 are the initial levels of prices and quotas respectively). Then the compensating income transfer, described in (2.22) is written (2.23) By setting (2.22) equal to (2.23), one can solve for the unifonn proponional change in tariffs and quotas which offsets the arbitrary change in (-r,Q) in the sense that it returns the aggregate utility (welfare) to the initial level u. Thus:

PAGE 40

33 B 1 d1t w (2.24) Note that for the initial value {3 = 1, d/3 is also a percentage change (d{3/{3). The above expression is a scalar measure of the change in the overall trade restrictiveness of the economy as a result of any arbitrary change in trade policies. In particular, the percentage change d{3 has the interpretation of an index number of trade distortion: it is the uniform proportional change in trade distortions which returns the economy back to its initial welfare level, given some arbitrary change in these trade distortions. This index is due to Anderson and Neary (1991) and it is termed 'Trade Restrictiveness Index' (henceforward TRI). It may be noted that the derivatives -B. and -BQ can be interpreted as the marginal cost of tariff-increases and the shadow price of quotas, respectively. Then the welfare costs associated with tariffs and quotas are -B.QQ and B' ,...,.., respectively Accordingly, the term (B' QQ B' ,..-r) equals the negative of the total welfare cost of the initial trade structure, and it is termed the 'shadow value of distorted trade' (Anderson and Neary, 1991). As seen in (2.24), the TRI equals the sum of the (percentage) changes in all trade restrictions (,r,Q) where each component (-ri and Qj) is weighted by its contribution to the total welfare cost of the initial trade structure. As mentioned above, the balance of trade function can be interpreted as giving the net foreign exchange required to support a certain welfare level u Thus the TRI can also be thought of as measuring the uniform

PAGE 41

34 change in trade distortions which yields an offsetting change in the foreign exchange required to maintain the initial welfare. A more formal definition of the TRI is given below: (2.25) In words, the TRI is the uniform rate (:3 by which the domestic prices of the tariff ridden goods must be raised, and the levels of the quota-controlled goods must be reduced (tightened) so that the economy returns to the initial protective regime consistent with welfare level u 0 If the trade policies ( 1r, Q) do not change between periods O and 1, {:3= 1. As the prices of tariff-ridden goods get lower and quota levels expand, the rate of change d/31{3 of the TRI (and subsequently its magnitude) rises. Thus, an increase in the size of the TRI indicates that the economy moves towards free trade. Totally differentiating (2.25) for the reference welfare level u 0 yields: (2.26) Converting to proportional changes yields (2.27) + where Bi = 0B/01ri, B; = oB/oQ;, and a hat (A) over a variable denotes a proportionate If distortions ( 1r, Q) decrease in period 1, then returning the economy to the initial protective regime associated with welfare u 0 implies lower quotas (so that dQ/Q is

PAGE 42

35 positive) but higher tariffs and consequently higher domestic prices (so that dTIT is negative). In that case, (2.27) may be also written as M L (BjQj) Qj j (2.28) I I BoO Bw 1t Turning to the derivatives B., BQ, differentiation of B() in (2.21a) with respect to T, Q (and recalling that m, pare themselves functions of (T,Q,u) from (2.20) yields B~ = t 1 ~ + u> 0 1 Pw (2.29) (2.30) Moreover, Anderson and Neary (1992) show that the derivatives of the import demand functions m(-r,Q,u) and the inverse demand functions p(-r,Q,u) can be expressed in terms of either the distorted or the standard trade expenditure function. Taking into account equations (2.17a) and (2.20) one may show that the following equations hold (2.31) (2.32) The derivative matrix E1 PP is generally expected to be negative definite. Finally, it may be noted that the TRI can be also extended to incorporate the effect on trade protection from policies applied to: (i) non traded goods and (ii) distortions in factor markets (Anderson and Bannister, 1991; Anderson and Neary 1992b). In

PAGE 43

36 particular, consider the case where some goods are not traded and their price vector is denoted by s. The TRI in this case is defined as (2.33) Note that the index is still defined over (T,Q) only; however the price vector s is an additional variable in the analysis as it reflects the policies which may be applied on the non traded goods. The TRI can now be derived by total differentiation of (2.33) with respect to 1r, Q, and s M .E (BJQJ) QJ (B~Q B~1t) B 1 ds s (2.34) It may be noted that a new complication arises here as the level of the TRI appears in the third denominator of the right-hand side. This difficulty may be overcome by multiplying (2.34) by (3 and transforming it into a first-order differential equation in (3, i.e. (2.35) where D 1 represents the first two terms on the right-hand side of (2.34) and D 2 represents the third term. Solving for (3 yields

PAGE 44

37 A = [ A + D1 l e Di t D1 Pt P t-1 D D 2 2 (2.36) However in empirical applications of the index with discrete time-data, (2. 35) can be approximated as simple difference-equation and be solved as (2.37) This formula can be applied to each period ~,t 3 with the normalization that {3 = 1 in the initial period t 1 The introduction of distortions in factor markets can be treated in an analogous fashion (Anderson and Neary, 1992). Comparisons Among the Measures of Protection In attempting comparisons among these protection measures, one should bear in mind that they do not include the same set of policies in their definitions. Additionally, some of these policies cause trade distortions while others do not. The focus of the PSE/CSE approach on income-transfers makes it the least transparent measure of trade distortion. The ERP, by including government policies on intermediate input-prices, is a superior indicator of government intervention over the NRPP which captures only border measures and measures taken at the level of the final products; quantity restrictions nonetheless, are considered only implicitly in the ERP. The NRPc on the other hand, provides a good analytical measure of trade distortions for consumers. The PSE/CSE measure is the most flexible as it covers a wide range of government policies. However its flexibility opens the possibility for policy changes

PAGE 45

38 which reduce the size of the PSE (CSE) without affecting the existing trade distortions. Similarly the NRPp, by focusing only on policies applied on the final goods, opens up loopholes though which trade distortions are manipulated while the level of NRPP remains unaffected (e.g. by applying government intervention on inputs). The PSE does not include the effects of government policies on intermediate product prices By contrast, the ERP by explicitly including policies affecting intennediate inputs, is a more reliable indicator in the sense that certain targets of ERP can be negotiated and attained via either input or output (final good) policies. Moreover the DRC is superior to the ERP in capturing, at least implicitly, nonprice trade distortions. In tenns of calculation, the PSE (CSE) and the modified PSE are most easily calculated by adding up budget expenditures on income transfers The ERP, on the other hand requires explicit estimates of input-output coefficients. The DRC can be a more accurate measure than the ERP; it requires however estimates on both input-output coefficients and shadow prices for the domestic factors. The TDS and modified PSE are most appropriate analytical measures of trade distortion in production; however calculation of the TDS requires data on supply, demand elasticities. Apart from the above mentioned differences, the ERP (NRP) approach, the DRC and the CSE/PSE approach share a number of common methodological traits. First, they all are partial equilibrium measures, meaning that the prices of non-traded goods and goods in other (e.g. non-agricultural) sectors are held constant. Cross commodity substitution effects for both producers and consumers are ignored The small-country framework is maintained thus ignoring the (potentially) important effect government

PAGE 46

39 policies may cause on world prices. Supply-control programs and income stabilization programs are not modelled explicitly in any of the measures (the TDS is an exception by including set-aside policies, while the DRC may implicitly capture nonprice distortions via imputed shadow prices). Last but not least, none of the aforementioned measures has any theoretical foundation. From a methodological point of view, this may be their most serious drawback. Aggregate measures should summarize the changes of their individual components in the sense of an index number. The change in value of an index number is generally a weighted average of changes in the components of the index while the respective weights are explicitly derived from the economic structure associated with the index This makes the weights consistent with the underlying economic theory and the resulting index meaningful. The TRI approach remedies some (but not all) of the above shortcomings. First, the TRI is defined in theory as a general equilibrium measure. Second, it is explicitly derived from the underlying economic structure gaining theoretical consistency. Third, the TRI incorporates explicitly both the consumption and production sector of the economy and also allows cross-commodity relationships (cross price elasticities are explicitly included in the formula that computes the TRI); this permits a theoretically consistent aggregation over commodities and (or) sectors. Fourth, by specifically incorporating expenditure and revenue functions in its definition, the TRI can potentially allow the explicit modelling of supply control policies (e.g uncompensated acreage

PAGE 47

40 reduction or production quotas). Fifth, the effect of purely domestic policies can be included in the definition of the TRI thus enhancing its comprehensiveness. Nonetheless, the TRI approach still maintains the small country framework and has rather high information requirements (price elasticities are needed in its computation). Additionally, in its practical implementations, the TRI is still likely to be specified at a partial-equilibrium context unless the researcher has access to a complete general equilibrium model. Furthermore, the use of elasticities has been criticized as a practical means of trade negotiations. With respect to that, it must be stressed that elasticities can be viewed as weights that differentiate the importance of the various commodities included in the analysis and thus they can easily be interpreted even by non experts (Roningen and Dixit, 1991). Most importantly, the price elasticities in the formula of the TRI can be viewed as a theory-based means of aggregating the degree of trade protection across the commodities under examination.

PAGE 48

CHAPTER 3 THE TRI APPROACH IN JAPAN'S AGRICULTURE This chapter presents an empirical implementation of the TRI. The focus of this application is the farm sector of Japan and the respective policies affecting the trade flows of agricultural products. Japan provides a typical example of a country that protects the domestic agricultural sector predominantly by tight border measures along with a wide array of support programs within the farm sector itself. Indeed in recent years Japan, along with the European Economic Community (EC), have been charged with being two major world markets fairly closed to international trade in farm goods. In the subsequent sections the TRI pertaining to a subset of Japan's farm sector is first specified, the respective information requirements are then empirically estimated, and the index is computed for the period 1971-87. Japan's Agricultural Trade Policies With relatively poor endowments of natural resources and arable land, Japan is one of the world's largest net importers of farm products (OECD, 1987). It is also the largest and one of the most stable overseas markets for U.S. agricultural products. The U.S. consistently accounts for a significant share of the Japanese principal farm imports. 41

PAGE 49

42 Despite the reliance on imported foodstuffs, the agricultural policies that Japan pursued in the 1970s and 1980s (as prescribed by the Agricultural Basic Law, enacted in 1961) have been policies of self-sufficiency. As a result, in the recent years Japan has persistently kept its agricultural markets closed to international trade. To achieve import control, the government has used a complex intervention system comprised of a variety of policy instruments, administrative measures, and implementing institutions. These include conventional trade measures such as import tariffs and quantitative restrictions (import quotas), but also state trading and trade controlled indirectly by the government. In particular, trade in certain products (wheat, rice) is carried out directly by the Food Agency of the Japanese government. In other cases (livestock, sugar), trade is carried out by parastatal agencies the Livestock Industry Promotion Corporation (LIPC), and the Japan Raw Silk and Sugar Price Stabilization Corporation (JRSSPSC), respectively. Participation of private traders in external trade is typically subject to strict licensing procedures and administrative guidance (Fitchett, 1988). Due to insufficient information on certain policies and lack of data this study does not provide a comprehensive representation of the trade policies and administrative programs in Japan's agriculture. In terms of import value the examined commodities, represent 20% to 30% of the value of Japanese imports in food and live animals during 1971-87. Nonetheless, this study investigates a subset of agricultural policies which comprises the most well known trade restrictions and have been most often criticized by Japan's trading partners (such as the beef and citrus quotas or the ban on rice imports).

PAGE 50

43 In particular, this study measures (ceteris paribus) the level of protection by means of the TRI in the following farm products: (1) beef, (2) pork, (3) poultry, (4) wheat, (5) rice and (6) fresh oranges. An outline of the domestic marketing policies and related border measures pertaining to these products during the period 1970-1987, (Australian Bureau of Agricultural and Resource Economics, 1988; Fitchett, 1988; OECD, 1987) is presented below and summarized in Table 3.1. The Beef Industry Although the beef industry is a minor sector in Japan's agriculture (in terms of production value), it attracts special attention because it is one of the industries the government wishes to expand and, at the same time, enjoys the support of the politically powerful cooperative movemen t (Australian Bureau of Agricultural and Resource Economics, 1988; Longworth_, 1983). For the period under examination 13 (1970 -1987), Japan's beef policy included two major instruments: a price stabilization scheme, and import quotas as a means to achieve the price stabilization objective. The implementation of these policies has been assigned to a quasi-governmental agency known as the Livestock Industry Promotion Corporation (LIPC). The basic mechanism behind the beef price stabilization scheme is the following: every year the government, in consultation with the LIPC and other bodies (consumer groups, unions, etc.) determines a price band (i.e. a range between an "upper" and a 13 In 1988, Japan agreed to abolish its beef quotas by April 1, 1991.

PAGE 51

44 Table 3.1: Japan's agricultural policies, 1970-87. COMMODITY TYPE OF TRADE POLICY STATUTORY SUPPORT BODY Pork wholesale floor variable levy on L.I.P C and ceiling imports price Chicken private price 20 % tariff on stabilization imports band Beef stabilization (i) import quotas L.I.P C. price band (ii) 25 % tariff Wheat Government State trading Food Agency sets purchasing and resale price Rice Government State trading Food Agency sets purchasing and resale price Oranges (i) import quotas (ii) 20% to 40% tariff "base 11 stabilization price). This price band is set for representative beef categories (steer, wagyu herd, etc ) in the representative wholesale markets, Tokyo and Osaka. In addition, the government determines (usually twice a year) an import quota for beef; the quota is II global 11 in the sense that it is not directed towards any specific country. Suppliers however must be able to meet Japanese quarantine requirements. With the beef price band and the quota levels predetermined, the LIPC intervenes in the market to keep the beef wholesale prices within the stabilization band. It does so

PAGE 52

45 by buying and storing beef when prices fall below the price band and releasing beef from its stock when prices move above the price band. The LIPC may stock and subsequently release both domestically produced and imported beef. In practice however the LIPC manipulates the beef market by regulating the flow of imports via the quota levels (Australian Bureau of Agricultural and Resource Economics, 1988). The LIPC buys imported beef from licensed traders by competitive tender and releases it into the wholesale domestic markets by auction. In addition to quota restrictions Japanese beef imports are subject to a 25 % ad valorem tariff. The associated tariff revenue is collected by the government (Ministry of Finance) and used for the development and assistance of the domestic livestock industry. More detailed presentations of the beef stabilization regime can be found in Australian Bureau of Agricultural and Resource Economics (1988), OECD (1987), and Longworth (1983). The Pork and Poultry Industries The government's support policies for the Japanese pork industry are similar to those on beef ; in the case of pork however, the primary policy instruments are price ridden rather than direct control over imports. As with beef, the government sets a stabilization price band for each fiscal year and assigns to the LIPC the role of supervising the pork market (i.e. absorbing from or releasing into the market the necessary quantities of pork and controlling imports).

PAGE 53

46 However in 1971 quantitative import control on pork was abolished. Since then pork imports have been left to private traders (licensed and approved by the LIPC), while the LIPC keeps the price of pork within the stabilization price band by means of a flexible tariff system. In particular, pork imports are subject to the higher of either a 5 % ad valorem tariff or a differential duty. This differential duty is applied when the import price is lower than the central price of the stabilization band and is defined as the difference between the central price of the stabilization band and the import price. The key aspect of this tariff system is that the price of imported pork is always equal to or higher than the central price of the stabilization band. Additionally the tariff is equal to at least 5 % (or higher) of the world price. A rather interesting side-effect of this mechanism is the preference of Japanese importers for higher quality (and higher price) pork cuts (loins, bucks, etc.) so as to subject their imports to the 5 % duty rather than the differential duty (OECD, 1987). In the poultry sector, direct government assistance is minimal. There are no conventional i ntervention forms of price stabilization or quantitative import restrictions. A price stabilization fund for broilers was set up in 1970 by the National sales Federation of Agricultural Cooperatives and the national Purchase Federation of Agricultural Cooperatives, without government participation; the government nevertheless assists poultry producers through research, and disease prevention programs (USDA, 1983). Additionally, poultry imports are subject to an ad valorem tariff with base rate of 20%.

PAGE 54

47 The Rice Industry For cultural, historical, and religious reasons, rice is the most significant agricultural crop in Japan. Rice farmers are known to be one of the most influential groups in Japanese politics. As a consequence, Japanese governments have intervened extensively in the rice market over the years influencing its production, marketing, and trade. Today the rice policy regime in Japan is a comprehensive policy mix comprising (i) domestic supply control measures, (ii) state trading (import control), and (iii) pricing determined by the government. The government administers its programs in the rice sector through a governmental body, the Food Agency. The Food Agency buys rice directly from producers and sells it to wholesalers. Every year the government decides both the purchase price (price at which the Food Agency buys from rice producers) and the resale price (price at which the Food Agency sells to wholesalers) For most of the period under examination, the average purchase price has been set higher than the resale price generating considerable government deficits. In addition to regulating the purchase and resale price for rice, the government also controls international trade on rice Foreign trade in rice is carried out exclusively by the Food Agency; importers must apply for government permission and sell all rice imported to the government (Food Agency). As a result the domestic rice market is effectively insulated from the world markets. Japan has maintained a firm policy stance in rice full self sufficiency. This means that the government would not import or permit rice to be imported as long as domestic production can meet domestic demand.

PAGE 55

48 Practically, no imports have been permitted since 1970, except for small quantities of glutinous and cracked non-glutinous rice for particular processed products (OECD 1987). Rice exports have been made only occasionally to dispose of accumulated stocks. The only time that there were notable rice exports is the period 1979-83 when Japan exported to other Asian countries about 3 million tones of surplus rice at favorable repayment conditions (long term, low interest credit). The reaction of the U S however led to Japan terminating those export sales. The high level of support to rice growers coupled with falling rice consumption, and increasing yields has resulted in considerable stockpiles of unwanted rice To cope with rice oversupply, Japanese administrations have introduced two control schemes : (i) various rice land diversion programs designed to curtail domestic supply, and (ii) the voluntarily marketed rice program. Concerning diversion of paddy fields to other uses there have been four programs over the 1970-87 period designed to reduce rice supply and increase the production of other priority crops. In all four programs, an acreage reduction target was set annually, and diversion payments were offered to participant farmers. Participation in the programs has been voluntary. The focus of each program however has been different. The Rice production Control and Diversion Program (197175) aimed primarily to reduce rice production and substitute other crops for rice. The Comprehensive Paddy Field Utiliwion Program ( 197678) aimed to reduce rice production but increase the self sufficiency of certain crops (with priority given to soybeans, feed crops, and vegetables).

PAGE 56

49 The Paddy Field Utilization Re-orientation Program ( 1978-86) focused more on the reduction of the farm size than the reduction of rice surpluses. Finally, the Paddy Field Fanning Establishment Program (1987-92) was designed to improve farm productivity and establish a regional crop rotation program, besides curtailing rice production. The voluntarily marketed rice program was introduced in 1969 in order to provide an alternative channel for marketing rice (besides the Food Agency) The major difference between voluntarily marketed and government marketed rice is that the government does not fix the purchase price and lets the market mechanism work. Nevertheless subsidies and assistance are provided for the smooth marketing of voluntarily marketed rice. Because its purchase price is not fixed, the voluntarily marketed rice is usually of higher quality. Consequently, the price of voluntarily marketed rice brands are about 25 % higher than the government marketed rice at the wholesale level and about 35% higher at the retail level (OECD, 1987) Thus the voluntarily marketed and government marketed rice complement each other in the preferences of the consumers for product differentiation and higher quality rice. It must be also noted that although the voluntarily marketed rice is subsidized the government expenses are still lower than they would have been if all produced rice was bought by the government (OECD, 1987) The Wheat Industry Besides rice, the wheat industry is also important in Japan, and a close relationship exists between the two crops as a result of policies aimed to reduce the

PAGE 57

50 excess rice production and encourage the cultivation of alternative crops. While wheat production was declining in the 1960s, the rice land diversion programs described earlier caused the production of wheat to rise and the level of wheat self sufficiency to increase in the 1970s. Nevertheless Japan imports about 80% of the wheat it consumes (Fitchett, 1988). Similar to rice, the marketing of wheat is effectively controlled by the government via the Food Agency. The Agency purchases wheat from individual growers at a predetermined price and consequently sells it to wholesalers also at a fixed price. Although growers have the option of marketing their output privately, in practice the Food Agency buys almost all the wheat harvest, as it offers prices substantially higher than the international market. In addition the Food Agency fully controls all wheat imports; importers must seek government approval and sell all their importation to the Agency. As in the case of domestically produced wheat, the government also sells imported wheat to domestic users at a fixed price. For domestically produced wheat the difference between the government purchase and resale price is substantially high, creating considerable deficits. By contrast, in the case of imported wheat the government resale price is usually above the international price at which the government buys imported wheat. Thus, the government gain from the sale of imported wheat helps offset the deficit created by paying high prices to domestic wheat growers; however this gain may fluctuate widely from year to year due to the fluctuations in world prices and exchange

PAGE 58

51 rates (e.g. the sharp appreciation of the yen over the period 1985-87 lowered considerably the prices paid by wheat importers). The Fresh Citrus Industry Japan's fruit industry is largely dominated by citrus fruits the major crops being oranges and lemons. Domestic orange varieties include tangerines (such as mikan, hassaku, and iyokan oranges), the summer-orange natsu-mikan, and the navel orange. The domestically produced mikans are the most important fruit in terms of area planted and volume of production (Australian Bureau of Agricultural and Resource Economics, 1988). Contrary to the grain industry, the horticultural industry in Japan does not receive any direct price support. This is not to say, however, that the industry does not enjoy government assistance and protection. A variety of measures such as quotas, import licenses, blending requirements, import duties, quarantine regulations etc. are in place to assist domestic growers against foreign competitors. In the case of mandarin oranges, evidence suggests (Baker and Mori, 1985) that imported oranges do not seriously affect the domestic industry because they are marketed mainly between April and September when only small quantities of the domestic mikans are marketed. Nevertheless, for the time period examined in this study, imports of fresh oranges from all sources were subject to import quotas. In addition, an ad valorem tariff was imposed on orange imports at a rate of 20% for the period from June 1, through

PAGE 59

November 30, and 40% for the period December 1, through May 31. 14 52 Japan's persistence in protecting fresh citrus has been attributed to the fact that citrus growing was encouraged as an alternative to rice production. At the same time, there has been extensive financial involvement of the agricultural movement in the fruit industry; it was therefore feared that abandoning import controls would reduce demand for the domestic orange varieties (Australian Bureau of Agricultural and Resource Economics, 1988) Japan s Balance of Trade Function on Farm Imports As shown earlier, deriving the TRI for the aforementioned traded goods requires first the specification of a (partial equilibrium) budget constraint that relates the aggregate Japanese consumer expenditure to the sum of the GNP function and trade revenues, associated with these goods. Before specifying the relevant functions however the notation to be used is presented below. For notation convenience, let: h: the price vector of tariff-ridden goods (pork price = ~1r;, chicken price p: the price vector of quota-controlled goods (beef price = Pbf, oranges price = pJ, p: the producer price vector of state-traded goods (wheat price =pwh, rice price = pJ, 14 In 1988 after negotiations with the U.S. Japan agreed to terminate import quotas on oranges and tangerines by April 1991 and retain the same tariff after the quotas are lifted

PAGE 60

53 s: the consumer (user) price vector of state-traded goods (wheat price = Swb, rice price = sJ, u: the international vector price of state-traded goods (uwh for wheat uri for rice), p: the international price for the quota controlled goods (p\r for beef, 0 for oranges), Q: the vector of quota levels (~r for beef, Q 0 for oranges), A: the riceland diverted from rice production in each period, c: the subsidy received by rice growers under the paddy field diversion programs. On the production side, it is assumed that the farm products examined here are produced via production processes separable from each other. That is to say, from the producer's view-point the output level for each of these products depends only on its own price and on the prices of the respective inputs. This implies that joint production processes are assumed away and results in an additive GDP or revenue function expressed as ( 3 .1) The first six terms on the right hand-side of (3.1) are the profit functions for the tariff-ridden goods (pork and chicken), the quota controlled goods (beef), and the state

PAGE 61

54 traded goods (wheat and rice), respectively. The last term to the far right represents payments to primary factors V employed in these agricultural sub-sectors. Since the analysis here is partial equilibrium, the prices w and supplies V of primary factors are considered exogenous The preceding GDP or revenue function can be equivalently written as the sum of returns to all factors associated with tariff-ridden, state-traded, and quota-controlled goods considered in this study, i.e. G(h, p,p; w, V) = Gl (hpk'hch' Pwh' Pr i w) + w' vl + (3.2) where G 1 ( ), V 1 are returns to all factors associated with tariff-ridden and state-traded goods, while G 2 ( ), V 2 are returns to factors associated with the quota-controlled goods. On the demand side, the aggregate expenditure function associated with these goods can be written as (3.3) Then the excess expenditure over the revenue function G( (i.e. the trade expenditure function) is given by the difference E() = e() G() (3.4) Given the fact that some of the goods under examination are subject to import quotas, the aggregate consumer minimizes expenditure only on the goods not subject to a quota. This implies that the trade expenditure function on goods other than those

PAGE 62

55 which are quota-controlled is conditional on the quota levels. This leads to the distoned trade expenditure function defined as ( 3. 5) The excess (net) expenditure function of the aggregate consumer over all the goods under examination is then equal to or E(h,s,p,Q,u) + p 1 Q (3.7) By Shephard's lemma, the derivatives of E() with respect to h yield the (net) imports of tariff-ridden goods, Eh(h,s,p,Q,u) = eh() Gt() = m(h,s,p,Q,u) (3.8) With respect to the state-traded goods (wheat and rice), the vector of their production levels is given by the derivative vector of the relevant profit function, i.e.

PAGE 63

56 (3.9) while the vector of quantities demanded is given by the derivative vector of the distorted expenditure function e( i.e. e 9 (h,s, Q u} = X(h,s, Q u} (3.10) Additionally, the quota levels are defined as Q = e ( } G ( } p p (3.11) Separately, the government programs for the aforementioned agricultural products generate net government revenue (which may be positive or negative). In particular, this revenue consists of: (i) the portion of the quota rents in beef and fresh oranges retained in Japan These portions can be approximated by the number of quota licenses awarded to foreign importers However given the restrictive trading status in beef imports (regulation of imports by the parastatal LIPC) the portion of beef quota retained abroad may be expected to be zero The portion pertaining to fresh oranges is denoted below by w 0 (ii) the tariff duties and variable levies from the importation of chicken and pork (iii) the implicit subsidy (positive or negative) to consumers from fixing the consumer (selling) price in wheat and rice and the implicit subsidy to producers from fixing the producer (purchase) price for wheat and rice. More formally, the (net) government revenue is equal to the following sum:

PAGE 64

57 (p -a )Yri() + (s .-a-)xri() ri ri ri ri (3.12) The first line above denotes government revenues from quota rents. In particular, the first term is the quota rent on beef retained in Japan while the second term is the portion of the quota rent on fresh oranges. The second line is the government revenues from imposing tariffs and levies on chicken and pork imports, respectively. The third and fourth lines are the government revenues (positive or negative) from the policies applied on domestically produced wheat and on rice; the first terms are the implicit subsidy to producers (the difference between producer price and international price) while the second terms are the implicit subsidy (positive or negative) to the consumer (user). In more compact notation these government revenues can be expressed as (1-w) (p-p) 1 0 + -r 1 m (h, s, p, Q, u) (p-a) 1 Y(p) + (s-a) 1 X(h, s, Q, u) (3.13) Further, it is assumed that the government trade revenues specified above are redistributed costlessly to the aggregate consumer, in a lump-sum fashion. In addition, the government funnels into the rice sub-sector subsidies for diverting paddy fields to other uses. The respective government outlay on rice diversion can be expressed as

PAGE 65

58 (c A). By definition the value of consumption expenditure in the farm sub-sectors examined here less the value of the respective domestic product must equal total transfers to these sub-sectors from the government. These total transfers are the sum of the relevant government trade revenues shown in (3.13) and the riceland diversion outlay. Thus the balance-of-trade function is defined as B(h,s,p,Q,A,u) = E(h,s,p,Q,u) + p 1 Q (1-w) (p-p") 1 Q t 1 m (h,s,p,Q,u) + + (p-o") 1 Y(p) (s-o) 1 X(h,s,Q,u) cA {3.14) To facilitate the discussion, the prices of all price-constrained, traded goods are henceforth denoted by 1r i.e. { 3. 15) Given the policy of self-sufficiency on rice and the fact that rice imports were zero in the period studied here while exports took place on exceptional basis (surplus disposal), rice may be treated as a (quasi) non-traded good. As discussed in Chapter 2, in such a framework a trade regime involving price restrictions, quantitative import control, and intervention in factor markets (paddy field programs) the TRI is defined as

PAGE 66

59 1 1 1 OM Oo 1 o) ] Pri p, 7f ,A U 0 (3.16) That is, the TRI is the proportionate factor {3 by which one must discount the prices of the price-constrained, traded goods, as well as, the quota levels in period 1 so that the farm sector described by the function BO in (3.14) returns to the welfare level u 0 of the previous period. Moreover the rate of change in the TRI is given by total differentiation of (3.14) with respect to the policy variables 1r, Q, Sn, p 11 and A: m p = J where a hat (A) over a variable denotes a proportionate change, e.g. (dQ/Qj). The TRI, in other words, equals the sum of the percentage changes in all trade restrictions (i.e. tariff rates and quota levels) with each restriction being weighted by its contribution to the total welfare cost of the initial trade structure. At the same time, the policy changes in non-traded goods and their inputs are considered as seen in the second line of (3.17). Computation of this TRI requires first specification and evaluation of the policy derivatives B,.., BQ, B 11 BA. In this way one can calculate the weights by which the rates of change in policy variables 1r, Q,s, p, A are aggregated into a single measure the TRI. To specify the policy derivatives B,.., BQ, B 11 BA one needs to differentiate the

PAGE 67

60 balance of trade function B(-) in (3.14) with respect to all policy variables. A price (or a quota) appearing as a subscript in the following expressions denotes the partial derivative of the relevant function with respect to this price (or quota); for example the term (3.18) denotes the partial derivative of net pork imports mPk(), with respect the pork price ~11: Similarly, the term 0 Pbt ( 3. 19) denotes the partial derivative of the price of fresh oranges p 0 with respect to the beef quota Qbf Differentiation of the function B( with respect to policy variables yields mpk mch X"'h xr1 0 'tplc B hpk hpk hpk hpk hpk mpk mch X"'h xri 0 'tch B heh heh heh heh heh = Swh-Owh + BSvt, mpk mch X"'h xri 0 BPwh Swt, Swt, Swt, Swt, Sr1-0ri 0 0 0 0 y"'h ( Pwh-o;h) p..,, 0 Phpk 0 (3.20) + (1)0 Phcb oo 0 Ps..,, 0

PAGE 68

61 mtk mch wh Ii 'tpk (;;:)=bt Obt Xobt Xobt 't ch + mt,,k mch Xwh xri Swh-Owh Oo Oo Oo 8 ri-ari (3.21) 'tpk pk ch wh r i 't ch BSri=
PAGE 69

62 = E "' + wo P o lilt 'f' 1t (3.25) B E,Ir + ,., 0 Po Q 0 O 't" "" (3.26) where
PAGE 70

63 -pk -pk ffipbL ffipo [(Qbf)-1 (Q:) -,] bf Qbf 0:11 -ch -ch PbL Qhp1c heh (3.29) ffipbL mPo 0 Qhocb Xwh Xwh 0 Po Qhp1c Oswb PbL Po where m( denote the first derivatives of the standard (instead of the distorted) trade expenditure function E( ). The terms mQ, p,.., and PQ can be computed in an analogous fashion. It may be noted that the evaluation of the policy derivatives B,.., BQ, Bri, and BA requires information on: (i) the respective price elasticities on import demand and supply, (ii) domestic prices, international prices, and the imported quantities of the quota controlled goods (beef, fresh oranges). An empirical estimation of these elasticities is the task of the next section while the price and quantity data are discussed in Appendix C. Demand Elasticities A System-wide Approach In estimating import demand elasticities for the farm products considered in this study the system-wide (or differential) approach developed by Theil (1967) is utilized. This focuses on systems of consumer demand equations rather than individual equations. Moreover, instead of initially selecting a particular functional form for the consumer's utility function to generate the demand system, one differentiates the first-order conditions of the standard utility maximization problem to obtain a system of differential demand equations (i.e. expressed in terms of changes in prices and quantities).

PAGE 71

64 Specifically, manipulation of the utility maximization-first order conditions yields the following demand equation for the ith good: N wid(logqJ = 8id(logQ) + L8iJd(logpJ-logP 1 ) (3.30) j l where p.q w = -2.......:. i M i=l, ... ,N (3.31) is the respective budget share and (3.32) denotes its marginal share. Additionally, d(logQ) is a Divisia volume index of the form N d(logQ) = L wid(logqi) il d(logP') is a Frisch price index defined as N d(logP 1 ) = L 8id(logpJ il (3.33) (3.34) and cp is the reciprocal of the income elasticity with respect to (the Lagrange multiplier), Also, the following equalities hold 1 = a1og ct, alogM {3.35)

PAGE 72

N :E eij j=l N N :E :E eij = i=l j=l 65 (3. 36) In empirical implementation of the differential approach, based on the number and the similarity of goods examined, the assumption of (weak) separability is frequently utilized (Theil, 1980). This assumes that the various goods in the consumer's utility function may be divided into a number of commodity sub-groups, so that the overall utility function is some increasing function of the group sub-utilities. The consumer is assumed to first allocate the budget among the different groups and, in a second stage, the expenditure for each group is further allocated among the goods belonging to that group (multistage budgeting). Groups are assumed to be either strongly or weakly separable with each other; the terms blockwise dependence and block independence are also used. Within each group however the goods are no longer separable. For estimation purposes, the general form-differential demand system can be parametrized in a number of ways; a particular parametrization is the Rotterdam model where the demand parameters are assumed constant over time. The Rotterdam specification assumes that the coefficients 0;j,
PAGE 73

Then the corresponding discrete-time versions of the Divisia and Frisch indexes are N dPt: = L Wir:dPit:' il N dOr: = L wir:dqir: il and the differential demand equation of the ith good takes the estimable form where N W1tdqit = ei dQt + L 1t ijdPjt + eit jl i,j=l, ... ,N 66 (3.38) (3.39) (3.40) and eit is a random disturbance term which is assumed (Theil, 1980) to follow a multinormal distribution with zero mean and covariance i,j=l, ... ,N (3.41) The 8/s account for the price effect on the quantity demanded, keeping the utility level constant ( compensated effect) and are known as the Slutsky coefficients of the Rotterdam model. Note that given the linearity of (3.39) in its parameters, the income elasticity .{"; and the compensated price elasticities E;; can be readily obtained as (3.42) From (3. 36) and (3. 40) it can be seen that

PAGE 74

N :E 1tij = 0 jl i=l, ... ,N 67 (3.43) This implies that the NxN Slutsky matrix ( Ti;] has rank N-1. Additionally, it is shown (Theil 1980) that the covariances of the disturbance terms eit are given by (3.44) that is, all disturbance covariances (and variances) are proportional to the corresponding Slutsky coefficients. Thus the disturbances e1t, ... ,~t add up to zero and have a singular covariance matrix. This means that in implementing Rotterdam demand systems one equation may be dropped out and the other N-1 equations can be jointly estimated by maximum likelihood techniques. The linearity of the Rotterdam model in its parameters allows us to impose on the demand equations the standard constraints of demand theory by means of linear restrictions. In particular, the properties of adding-up, homogeneity, and Slutsky symmetry can be imposed as follows: (a) Adding-up (3.45) (b) Homogeneity (3.46)

PAGE 75

68 (c) Slutsky symmetry (3.47) and their compatibili t y can be checked by means of the likelihood ratio test (LRT). In this study the absolute version of the Rotterdam model is used under the assumption that the farm products considered here are divided into two separable groups. The first group incl u des meat and grains (i.e. beef, pork, poultry, wheat, and rice). Fresh oranges are assumed to belong in a second group that includes all citrus fruit. 15 Moreover it is assumed that the aggregate consumer's welfare function on farm imports consists of the sub-utilities deriving from these two groups which are assumed to be separable from each other. Estimation of the demand systems generated from these two groups yields conditional price and expenditure elasticities (i.e. elasticities which depend upon only the prices, quantities and income expenditure allocated to the particular group). Typically, the quantity and value of imports are used as data source for the estimation of import demand elasticities. In this study however, preliminary estimations using the reported import quantities and import value of the goods under examination (FAQ-Trade Yearbook, various issues), produced unsatisfactory elasticity estimates. This may be attributed to the fact that the markets examined in this study are severely distorted-quantity restrictions imposed on beef imports combined with state trading in the 15 Fresh oranges were separated from the rest of the goods, as preliminary estimations including all six goods in a single group showed that cross price relations between oranges and the rest of the goods are insignificant.

PAGE 76

69 case of wheat and rice. Thus, using trade data on a model based on the implicit assumption of undistorted markets, such as the Rotterdam model, is likely to yield questionable estimates (e.g. positive own-price elasticities). However this problem may be overcome by recalling that imports defined as the partial derivatives of the trade expenditure function E( are the difference between the quantity demanded and quantity supplied (see equations (3.8) and (3.11) in section 3. 7 of this Chapter). Thus import demand elasticities can be computed as the difference between elasticities of demand and supply. Accordingly, the first group (meat and grains) was estimated for demand (rather than import demand) price elasticities. The price data used in this estimation are the respective domestic wholesale prices (reported for each commodity in Appendix C) while the quantity data are the respective total consumption volumes (gross food) reported in OECD-Food Consumption Statistics (196478, 1979-88). The econometric estimates and test statistics for the first group (e.g. beef, pork, poultry, wheat, rice) are reported in Tables 3.2 and 3.3, respectively. Table 3.2 presents the estimated conditional price coefficients and expenditure coefficients with the standard constraints of demand theory (i.e. homogeneity and Slutsky symmetry) imposed. All own-price estimates are negative as expected. The own-price estimates of beef, pork, and rice are statistically different from zero at a=0.05 level of significance while those of poultry and wheat are statistically different from zero at a=0.11 level and a=0.14 level, respectively. Concerning the cross-price Slutsky coefficients, a positive sign indicates substitutes, while a negative sign indicates complementary goods. Of the cross

PAGE 77

70 Table 3.2: Parameter estimates of a Rotterdam model on meat (beef, pork, poultry) and grains (wheat,rice), homogeneity and symmetry imposed, 1965-87. Conditional price coefficients -rij Expend. coeff. Beef Pork Poultry Wheat Rice (Ji Beef -.0308 .0252 .0013 .0037 .0006 .4488 (.012) (.009) (.0055) (.0020) (.014) (.056) Pork -.0446 -.0001 -.0038 .0233 .3565 (.014) (.0072) ( 0028) (.018) (.0582) Poultry -.0096 -.0062 .0146 .1619 (.0073) (.0027) (.009) (.0305) Wheat -.0034 .00976 .0003 (.0031) (.003) (.010) Rice -.0483 .0329 (.030) (.0944) Asymptotic standard errors in parentheses. Table 3.3: Hypothesis testing of the Rotterdam model. Log of likelihood Likelihood ratio xModel function test (LRT) (.05) Unrestricted 389.852 Homogeneity 386.889 5.926 9.49(4) Homogeneity and 380.603 12.572 12.59(6) Symmetry Homogeneity and 18.498 18.31(10) Symmetry vs. unrestricted Numbers inside the parentheses indicate number of restrictions imposed.

PAGE 78

71 price estimates, the wheat-rice term and the beef-pork term are statistically different from zero at a=0.01 level and positive; this implies that wheat and rice, as well as, beef and pork are substitutes. Additionally, the beef-wheat term and the poultry-rice term are statistically significant from zero at a =0.10 level and a =0.15 level respectively; they are also positive indicating substitutability between beef and wheat, and poultry and rice. Finally, the poultry-wheat term is significant from zero at the a =0.025 level and negative indicating complementarity between wheat and poultry. The validity of homogeneity and symmetry restrictions is then checked by means of a likelihood ratio test (LRT). The relevant test statistic is LRT = -2 [ log LR log Lu] (3.48) where Lit is the log-likelihood value of the model with the restriction(s) imposed and Lu is the log-likelihood value without the restrictions. The LRT statistic has an asymptotic x 2 (r) distribution, where r is the number of restrictions imposed (i.e. the degrees of freedom equal the difference between the number of parameters in the model without restrictions and with restrictions). The computed values of the LRT statistic appear in the second column of Table 3.3. The following three hypotheses were tested: (a) the null hypothesis of homogeneity against the unrestricted version of the model, (b) the null hypothesis of symmetry against homogeneity, (c) the null hypothesis of both symmetry and homogeneity against the unrestricted model. The null hypotheses (a) and (b) cannot be rejected at the 0.05 significance level, while the null hypothesis (c) cannot be rejected at the 0.02 significance

PAGE 79

72 level. This implies that the estimated demand equation system complies with the properties of homogeneity and symmetry in prices as required by the standard consumer demand theory. Conditional (and compensated) price elasticity estimates can be computed by dividing the relevant Slutsky price parameter by the budget share of good i, that is (3.49) However for the group tested here (meat,grains) this generates price elasticities of demand rather than imports. The conditional own-price elasticities of beef, pork, poultry, wheat, and rice are reported in Table 3.4, while Table 3.5 gives their conditional cross-price elasticities. Of the goods belonging to this group, the calculation of the TRI requires demand elasticities for grains (wheat and rice) but import demand elasticities for the three types of meat (beef, pork, and poultry). As mentioned earlier, import demand elasticities for meat can be obtained as the difference between demand and supply elasticities. It may be noted that the differential approach used in demand analysis can be also used to generate a system of supply equations, since utility maximization and production maximization are mathematically identical optimization problems. Hence the Rotterdam model can be also utilized to estimate supply elasticities. However due to insufficient data on producer prices, no supply elasticities could be estimated. As an alternative, exogenous information on supply elasticities was used. In preparation for the Uruguay Round of GATT negotiations, the Economic Research Service (ERS) of the U.S.

PAGE 80

73 Table 3.4: Conditional own-price elasticities of demand Rotterdam model on meat (beef, pork,poultry) and grains (wheat, rice), 1965-87. Year Beef Pork Poultry Wheat Rice 1970 -0.270 -0.323 -0.172 -0.053 -0.077 1971 -0.242 -0.273 -0.142 -0.055 -0.084 1972 -0.224 -0.257 -0.152 -0.060 -0.085 1973 -0.167 -0.235 -0.140 -0.061 -0.097 1974 -0.201 -0.244 -0.138 -0.060 -0.090 1975 -0.200 -0.217 -0.150 -0.069 -0.092 1976 -0.187 -0.231 -0.146 -0.060 -0.093 1977 -0.173 -0.237 -0.166 -0.061 -0.093 1978 -0.160 -0.235 -0.176 -0.062 -0.095 1979 -0.175 -0.232 -0.152 -0.062 -0.094 1980 -0.182 -0.231 -0.143 -0.055 -0.095 1981 -0.196 -0.219 -0.134 -0.056 -0.096 1982 -0.182 -0.244 -0.143 -0.058 -0.093 1983 -0.171 -0.238 -0.142 -0.054 -0.096 1984 -0.166 -0.241 -0.138 -0.054 -0.098 1985 -0.157 -0.275 -0.147 -0.054 -0.094 1986 -0.147 -0.281 -0.149 -0.055 -0.096 1987 -0.135 -0.286 -0.171 -0.057 -0.097

PAGE 81

74 Table 3.5: Conditional cross-price elasticities of demand Rotterdam model on meat and grains (beef, pork, poultry, wheat, rice), 1965-87. Year Beef Beef Poultry Poultry Wheat w.r.t. w.r.t. w.r.t. w.r.t. w.r.t. pork wheat wheat rice rice 1970 0.221 0.032 -0.111 0.262 0.150 1971 0.198 0.029 -0.091 0.216 0.157 1972 0.183 0.027 -0.098 0.231 0.170 1973 0.137 0.020 -0.090 0.213 0.171 1974 0.165 0.024 -0.089 0.210 0.170 1975 0.164 0.024 -0.097 0.229 0.195 1976 0.153 0.022 -0.094 0.222 0.169 1977 0.141 0.021 -0.107 0.253 0.173 1978 0.131 0.019 -0.113 0.268 0.175 1979 0.143 0.021 -0.098 0.232 0.175 1980 0.149 0.022 -0.092 0.218 0.157 1981 0.160 0.024 -0.086 0.203 0.158 1982 0.149 0.022 -0.092 0.218 0.165 1983 0.140 0.021 -0.092 0.217 0.154 1984 0.136 0.020 -0.089 0.210 0.153 1985 0.129 0.019 -0.095 0.225 0.153 1986 0.121 0.018 -0.096 0.227 0.155 1987 0.111 0.016 -0.110 0.260 0.162 ( continued)

PAGE 82

75 Table 3.5 continued. Year Pork Wheat Wheat Rice Rice w.r.t. w.r.t. w.r.t. w r.t. w.r.t. beef beef poultry poultry wheat 1970 0.182 0.057 -0.095 0.023 0.016 1971 0.154 0.059 -0.099 0.025 0.017 1972 0.145 0.065 -0.108 0.026 0.017 1973 0.133 0.065 -0.109 0.029 0.020 1974 0.138 0.064 -0.108 0.027 0.018 1975 0.123 0.074 -0.124 0.028 0.019 1976 0.131 0.064 -0.107 0.028 0.019 1977 0.134 0.065 -0.110 0.028 0.019 1978 0.133 0.066 -0.111 0.029 0.019 1979 0.131 0.066 -0.111 0.029 0.019 1980 0.131 0.059 -0.099 0.029 0.019 1981 0.124 0.060 -0.100 0.029 0.019 1982 0.138 0.063 -0.105 0.028 0.019 1983 0.135 0.058 -0.098 0.029 0.019 1984 0.136 0.058 -0 097 0.030 0.020 1985 0.155 0.058 -0.097 0.029 0.019 1986 0 159 0.059 -0.098 0.029 0.019 1987 0.161 0.061 -0.102 0.029 0.020

PAGE 83

76 Department of Agriculture (USDA) constructed a trade database known as the Trade Liberalization Database (TLIB) (USDA, 1989b). This database includes a set of price elasticities assembled from a survey of global agricultural models and commodity market studies. For developed countries such as Japan, the principal sources of these elasticities are: the Ministerial Trade Mandate (MTM) developed by OECD, the Grain Livestock, and Sugar model developed by Tyers and Anderson, and the Grain, Oilseed and Livestock model developed by the USDA (USDA, 1989b). The price elasticities of the TLIB have been used in numerous commodity models and comprise one of the most complete sets of price elasticities available. This study utilizes supply price elasticities on Japan developed by the OECD and reported in the TLIB. These elasticities which exclude cross commodity effects are reported as: e(s)(bf)=0.23, e(s)(port>= 1.5, e(s)(poulby>= 1.5, e(s)c,.,._>=0.44, e(s)cnce>=0.3 Table 3. 6 reports import demand elasticities for beef, pork and poultry computed by utilizing the preceding supply estimates, as well as, the conditional demand elasticities for wheat and rice. In summary, Tables 3.5 and 3.6 report the (estimated) information requirements which will be used next in the computation of the TRI. Computing the Combined TRI for a Subset of Japan's Farm Imports, 1971-1987 In this section the calculation of the combined TRI of Japan i mports of beef, pork, poultry, wheat, rice and fresh oranges is presented. The derivatives of the respective balance of trade function were evaluated for each period, utilizing the elasticity estimates reported above, along with information on imported quantities, domestic, and

PAGE 84

77 Table 3.6: Own-price elasticities of imports for meat (beef, pork, poultry) and ownprice elasticities of demand for grains (wheat, rice), 1970-87. Beef Pork Poultry Wheat Rice Year 1970 -0.500 -1.823 -1.672 -0.053 -0.077 1971 -0.472 -1. 773 -1.642 -0.055 -0.084 1972 -0.454 -1.757 -1.652 -0.060 -0.085 1973 -0.397 -1. 735 -1.640 -0.061 -0.097 1974 -0.431 -1.744 -1.638 -0.060 -0.090 1975 -0.430 -1.717 -1.650 -0.069 -0.092 1976 -0.417 -1.731 -1.646 -0.060 -0.093 1977 -0.403 -1.737 -1.666 -0.061 -0.093 1978 -0.390 -1. 735 -1.676 -0.062 -0.095 1979 -0.405 -1. 732 -1.652 -0.062 -0.094 1980 -0.412 -1.731 -1.643 -0.055 -0.095 1981 -0.426 -1.719 -1.634 -0.056 -0.096 1982 -0.412 -1.744 -1.643 -0.058 -0.093 1983 -0.401 -1. 738 -1.642 -0 054 -0.096 1984 -0.396 -1.741 -1.638 -0.054 -0.098 1985 -0.387 -1. 775 -1.647 -0.054 -0.094 1986 -0.377 -1. 781 -1.649 -0.055 -0.096 1987 -0.365 -1.786 -1.671 -0.057 -0.097

PAGE 85

78 international prices (a detailed description of price and quantity data is presented in Appendix C). As shown in equation (3.17), the combined TRI is then calculated by weighing the percentage change of the policy variables (7,Q,A) in each period with their respective share in the total dead-weight loss due to trade distortions, given by (B'QQ B' ,..7). The combined TRI for the period 1971-1987 assuming full quota rent retention 16 in the home country (Japan) is presented in Table 3.7. Inspection of the table reveals that the rate of change of the combined TRI is basically shaped by changes in the riceland diversion programs and the beef quota and, to a lesser extent, pork and poultry prices. In contrast, the contributions of rice and wheat are minimal reflecting the persistent policies of regulating the producer and consumer prices in both crops throughout the period under examination The rate of change of the index increases with lower prices 'I" for tariff ridden and state-traded goods, higher quotas Q, and less land employed in rice production. Hence the index rises as the trade distortions (-ir,Q), and the misallocation of resources (riceland) are getting reduced. The level of the index can be computed by using the simple difference equation shown in Chapter 2. One can use either the chain-principle (i.e. in each period, use as basis the level of the TRI in the previous period, starting with {J 0 = 1) or use f3ct l) = 1 16 In the absence of any information on the portion w of orange imports controlled by foreign importers, the values w=O, w=0.2, w=0.4 were alternatively used to account for full quota retention in the home country, low quota retention abroad, and high quota retention abroad respectively. Computations using w=0.2, w=0.4 produced little change in the estimated TRI, apparently due to the small contribution of fresh oranges in the overall index.

PAGE 86

79 Table 3. 7 The combined Trade Restrictiveness Index of Japan's agricultural imports, 1971-1987. Pork Poultry Wheat Rice Wheat Year (demand) (demand) (supply) 1970 1971 -0.02459 -0.03309 -3.3e-06 0.000114 -0.00113 1972 -0.01698 0.02461 0.000022 -0.00074 -0.00068 1973 -0.01367 -0.0158 -0.00007 0.000029 -0.00117 1974 -0.0369 -0.03618 -0.00021 -0.00284 -0.00325 1975 -0.09505 -0.02049 -0 00002 -0.00294 -0.00142 1976 -0.00117 -0.0052 -0.00014 -0.0013 -0.00087 1977 0.006232 0.001564 -0.00003 -0.00112 -0.01491 1978 0.025769 0.013365 0.000001 0 -0.00057 1979 0.037185 0.007626 3.lle-07 -0.00072 -0.00102 1980 -0.00702 -0.00944 -0.00006 -0.00063 -0.00408 1981 -0.04444 -0.00927 -0.00002 -0.0005 -0 00143 1982 0.013591 0.009637 -4.2e-06 -0.00104 0 1983 -0.00216 0.004764 -0.00005 0 -0.00026 1984 0 004486 0 002236 0.000001 -0.00105 0 1985 0.031799 0.010745 -4.3e-07 -0.00113 0. 000712 1986 0.006694 0.004628 -1.3e-06 -0.00035 0.002144 1987 0.012805 0.008617 0.000009 0 0.035188 (continued)

PAGE 87

80 Table 3. 7 continued. Rice Beef Oranges Paddy TRI TRI Year (supply) programs level 1970 1 1971 -0.00243 0.152008 0.002241 0.862268 0.955389 1.95539 1972 -0.004 0.119979 0.004022 0 101089 0.227323 2.30784 1973 -0.00598 0.367431 0.000886 -0.00698 0.324668 3.07404 1974 -0.02111 -0.54561 0.002035 0.70616 -l.35023 0.43768 1975 -0.01334 -0.07086 0.000711 -0.0775 -0.2809 0.26200 1976 -0.00501 0.318762 0.000586 -0.06777 0.237889 0 26966 1977 -0.00305 -0.0596 -0.00038 0 017219 -0.05408 0 26460 1978 -0.00006 0.072784 0.005136 0.31105 0.427475 0.60641 1979 -0.00015 0.126254 0 000945 0.037107 0.207231 0.74634 1980 -0.00225 -0.02736 0.004886 0.166777 0.120817 0.87809 1981 -0.00035 -0.00084 0.001199 0.094097 0 038447 0.92321 1982 -0.00137 0.00378 0.00375 0 006114 0.026898 0 94833 1983 -0.00223 0.059956 0.002609 -0.04552 0.017113 0 96209 1984 -0.00284 0.029695 -0.00012 -0.01438 0.018021 0.97874 1985 0 0.021505 0 009789 -0.01489 0.058526 1.03568 1986 0 0.113283 0.001288 0.009488 0.137174 1.17742 1987 0.004595 0.130345 0.001122 -0.00027 0.192411 1.40320

PAGE 88

81 for each period. In the latter case, the level of the TRI coincides with its rate of change. Figure 3.1 shows the rate of change of the TRI, while Figure 3.2 shows the level of the TRI using the chain-principle. The index shows considerable variation in the first half of the period under examination (1971-77) and a relatively smooth pattern in the second half (1978-97). In particular, during the period 1971-73 the index shows a positive but rapidly decreasing rate of change as beef quota, orange quota and riceland diversion all rise at a diminishing rate, while the domestic prices of pork and poultry rise, too. In 1974, the rate of the TRI turns negative implying a severe reduction in the magnitude of the TRI and therefore a dramatic increase in trade protection for that period. This severe drop of the index reflects the decision of the Japanese government to suspend the beef quota in late 1973 and to comp l etely close the beef market to imports in 1974 until the second half of 1975 (Australian Bureau of Agr. and Res. :Economics, 1987). This was coupled with an increase of almost 12% in the wholesale pork price, a 13.5% increase in the wholesale price of poultry, and a 79.5% reduction in the amount of riceland diverted from rice production. In the period 197577 the index shows a recovery as its rate of change becomes less negative in 1975 and turns positive in 1976, almost zero in 1977 (reflecting stable policies during 197677) and again positive in 1978 reflecting higher beef imports, higher riceland diversion and lower pork and poultry prices. Thereafter the index shows a smoother variation. Specifically, for the period 1978-84 the index shows a positive but decreasing rate of change as the beef quota was slightly reduced in 1980, 1981, 1982,

PAGE 89

82 1 0 8 0 6 ., 4 2 u 0 2 ... 0 GI 0 ., ., I.. -0 2 g -0 4 'O .. -0 6 C i B 8 -1 -1 2 -1 4 1971 1972 1973 197-4 19n 1976 1977 1978 1979 19B 1981 19B2 19B3 198-4 19B:S 1986 1987 Y e a r s Figure 3.1 The rate of change of the combined TRI, 1971-87.

PAGE 90

ii ... 2 8
PAGE 91

84 wh i le pork and poultry prices rose and the diversion rate of riceland became smaller and even negative in 1983, 1984. After 1984 the index rate of change shows a steady increase indicating a trade liberalization process at least for the period 1984-87. This is consistent with the so called 1984 U.S.-Japan Beef and Citrus Understanding in August 1984 Japan agreed to expand its import quotas for fresh oranges and grain-fed beef (mainly supplied by the U.S.) (USDA, 1986) It also reflects decreasing domestic prices on pork and poultry, stable or even decreasing domestic prices in wheat and rice and very small variation in the riceland diversion programs. Concerning the estimated level of the combined TRI, it may be noted that when this is computed according to the chain-principle, the dramatic drop of the index around 1974 affects the magnitude of the TRI in the subsequent years. As a result the index shows a slow recovery which gains momentum after 1984. Level wise the index in the last period (1987) is 40% higher than in the first period (Figure 3.2). This implies that using the trade regime of the initial period (1970) as reference-point the regime of the last period (1987) is clearly less restrictive. Nevertheless the magnitude of the TRI in 1973 is three times higher than in the first period. Therefore one must also note that the l ower distortions in the end of the examined period are still more strict compared to those in the beginning of the 1970s.

PAGE 92

85 Comparing the TRI with the Respective PSE/CSEs As explained in Chapter 2, the TRI is not directly comparable with other measures of trade protection. Nonetheless, Anderson and Neary (1992) suggest decompositions of the TRI which allow some comparison with the more conventional measures of producer and consumer subsidy equivalents. Specifically, one may define restrictiveness indexes separately for production and consumption distortions in an analogous fashion to the full TRI. In the application considered here, one may define a distonion index {3P for trade related production distortions as A.P A.P (hl A.P hl A.P 1 A.P 1 Qbf Oo 1, O) (3 50) [ 1 1 l t-' = t-': B pkt-', cht-' ,Pwht-' ,Pri pp' pp'A, U The rate of change of this index is computed analogously to that of the TRI, considering however only the policy variables related to production i.e. I\., heh, Pwh, Qbf, Q 0 Pn, A and ignoring policy variables related to consumption (i.e. S..Vb, sJ. Hence /3P is interpreted as the equiproportionate change in trade-related production distortions (accounting, at the same time, for production distortions in non-traded goods) which is welfare-equivalent to policy-changes from period O to period 1 A similar distortion index can be derived for trade-related consumption distortions, i.e. one may define a distortion index as

PAGE 93

86 1 1 (h l AC hl AC 1 AC 1 Qbf Oo. B pk t-' ch t-' S wh t-' Sri p c p c (3.51) The rate of change of~ is calculated by considering only the policy variables related to consumption i.e. l\,i.:, heh, -h, S,;, Qbf, Qo and ignoring production policy variables, i.e. Pwb, P,;, A. Hence, can be interpreted as the equ i proportionate change in trade-related consumption distortions (considering however the effect of consumption distortions in non-traded goods) which is welfare-equivalent to these policy-changes from period O to 1. Since the distortion indexes {3P are defined similarly to the full TRI, an i ncrease their magnitude implies less trade restrictiveness. The distortion indexes {JP, {Jc may be compared to the rate of change in the conventional PSEs, CSEs. The Economic Research Service (ERS) of the U.S. Department of Agriculture (USDA) has calculated PSEs, CSEs for a number of farm products in 27 countries for the period 1982-87. Table 3. 8 presents the estimated Japanese PS Es, CS Es for beef, pork, poultry, wheat rice, and mandarin oranges (USDA, 1990). These PSEs, CSEs are expressed as total income transfer per unit of output, i.e. yen per metric ton, (Y/MT). An aggregate PSE CSE is then calculated by weighing the individual PSEs CSEs by the respective production (consumption) value shares, i.e PSET = t(Pi;i)PSEi, i p y CSET = t(qixi)csEi i q 1 X (3.52) where i = beef, pork, poultry, wheat, rice, mandarin oranges, PiYi is the production value of the ith product, and CLXi is the consumption value of the ith product. The percentage

PAGE 94

87 Table 3.8 A comparison of distortion indexes with PSE(CSE) measures. Agreggate % % Agreggate % % PSE change in change in CSE change in change in PSE {3P CSE [JC Y/MT Y/MT Year 1 982 310951 -263950 1 983 342365 0.09175 0.01717 -305985 0.13737 0.06958 1 984 336236 -0.0182 0.01908 -305721 -0.0008 0.03757 1 985 328767 -0.0227 0.05964 -305535 -0.0006 0.07745 1 986 391267 0.15973 0.13735 -387632 0.21179 0.13277 1 987 396640 0.01354 0.19380 -403549 0.03944 0.15847 rate of change of these agreggate PSE, CSE is then reported along with the percentage change of the distortion indexes fJP, 13c. The production distortion index {3P shows positive rates of change throughout the period 1982-87 thus implying diminishing productionrelated distortions. In contrast, the aggregate PSE shows negative rates of change (thus implying less distortions) only in 1984, 1985. For 1983, 1986 and, 1987 the aggregate PSE shows positive rates of change suggesting higher production distortions. Identical conclusions are drawn by inspecting the rate of change of the aggregate CSE relative to that of the distortion index {3c. It is therefore clear that the TRI approach leads to different conclusions on trade liberalization from the existing measures of PSE(CSE). At least two reasons may be given for these contradictory results. First the PSEs, CSEs include policies which are not taken into account in the calculation of the trade related distortion indexes {3P, 13c since they are not directly related to trade. Second, the

PAGE 95

88 aggregate PSEs, CSEs and the trade-related distortion indexes /3P, 13c are constructed in fundamentally different ways. In estimating the aggregate PSE(CSE) one basically weights the price changes of the individual commodities by their value shares (in production or consumption). Hence, the overall rate of distortions is shaped by goods with high production (consumption) value. Such an aggregation is not derived from any theoretical basis; therefore the relative importance assigned to the price variations of the individual commodities is ad hoc and open to criticism. It must also be noted that the PSEs/CSEs consider the variation in price in both the cases of tariff-ridden and quota-controlled goods. However, in the case of quota-controlled goods, the variation in quantity is more relevant since the distortion is founded in the quantity available to the consumer rather than the price. This is indeed the case in the distortion indexes {3P and {3c wherein the overall distortion rate is shaped by changes in the price or quantity of the individual goods depending on whether goods are price or quantity controlled. In addition, the distortion rate of each commodity is weighted by its contribution to the total welfare loss associated with all trade distortions in place. This aggregation of the individual distortions is explicitly founded in the underlying economic structure (balance of trade function) thus providing theoretical justification in the resulting distortion index. It follows that methodologically, the information on trade liberalization provided by the TRI approach, gains validity over more conventional ad hoc measures such as the PSEs/CSEs given its theoretically robust foundation.

PAGE 96

CHAPTER 4 THE NEW GOODS PROBLEM AND THE TRI A long time concern in index number theory has been accommodating for the disappearance of old commodities and the introduction of new ones into the market. The construction of a price index between two consecutive time periods requires data on the prices of the commodities in both periods. However in the case of a disappearing (new) commodity, its price is not observed in the second (first) period. One theoretical explanation for the appearance (disappearance) of commodities in the marketplace is based on the concept of the reservation price. Conventionally the reservation price is defined as the price at which the consumer is just indifferent between purchasing or not purchasing a commodity. Hence, when a new good appears in the market it is inferred that its price in the previous period was at its reservation level resulting in quantity demanded equal to zero. Similar reasoning holds for the case of a disappearing good. In practice, the problem of new (disappearing) goods has generally been circumvented either by setting the reservation prices equal to zero or by ignoring them in the period they appear for the first time (and including them into the index in the subsequent periods wherein price data become available). Diewert (1980) however 89

PAGE 97

90 showed that both practices result in Fischer price indexes which are upward biased. Moreover, the bias is more severe when the reservation price is arbitrarily set equal to zero than when the new (disappearing) goods are simply ignored when they first appear (disappear). Theoretically, the correct procedure for treating new (disappearing) goods is to form an estimate of the reservation price and then use this estimate in the price index. One solution to this problem proposed by Griliches (1961) is the hedonic (or continuous characteristics) approach. According to this approach the price of a commodity is viewed as some function of certain continuous characteristics of the commodity (e g. in Griliches' seminal work on automobiles these characteristics included horsepower, length weight, etc). Estimates of the price of a new good can then be obtained from hedonic regressions using data on the characteristics of the good from the periods that it is avai l able. Abstracting from the statistical and theoretical difficulties associated with hedonic regressions the major draw-back of the hedonic approach is the fact that it can be implemented only if data on the characteristics of a commodity are available. Although a set of continuous characteristics may be easily identified in such industries like automobiles or computers, in many other cases such as agricultural commodities, it is most likely that the relevant characteristics of a product are either not identified or not readily measured. In such cases alternative techniques are needed for the treatment of the new ( disappearing) goods problem.

PAGE 98

91 The TRI in the Presence of New Goods This Chapter generalizes the standard TRI to account for new goods (in the sense of newly traded goods), by utilizing the work of Feenstra (1990) and Feenstra and Markusen (1991) who adjust price, quantity, and productivity indices in the presence of new goods. Their approach to the 'new goods problem' starts with the presumption that data on the characteristics of the goods under examination are not available; this implies that no reservation price can be estimated. To circumvent this difficulty, it may be assumed that the reservation price of a new good is infinite in the period(s) wherein the good is not available and falls to some finite price when the good first appears in the market. Given the assumption of an infinite-reservation price, the underlying utility (we l fare) function may be represented by a C.E.S. functional form with elasticity of substitution greater than unity In this case, the reservation price for any good is infinity, since quantity approaches zero only for arbitrarily high prices This approach leads to an adjustment of price (quantity) indices by a factor which depends on the expenditure share of the new goods when they first appear, as well as on the value of the elasticity of substitution. Thus the 'opportunity cost' of this approach is that instead of estimating reservation prices (which are infinity by assumption), one must estimate the elasticity of substitution. Drawing on the theoretical results of Feenstra and Markusen (1991), the standard TRI can be adapted to allow for different ranges of traded goods between any two

PAGE 99

92 periods of time. The basic idea of such a generalization is to alter the reference utility (welfare) of the standard TRI. To start with, if the same range of goods is available to the consumers over time then the utility levels for periods O and 1 respectively, are written as u 1 (x;M) and u 0 (x;M) where x denotes the quantity vector of all goods (quota constrained and tariff-ridden goods) in the respective periods. The important point of this formulation is that welfare levels in periods O and 1 depend only on the quantities consumed (which may be different). However the number of goods in the welfare function remains constant. In this situation, the appropriate TRI is that presented in Chapter 3. That is, taking the initial welfare u 0 (x;M) as the reference point, the TRI is defined as the factor of proportionality (3 by which period-I policy vector (-r, Q 1 ) would have to adjust in order to make the balance of trade function hold for the period-0 welfare u 0 (x;M). Formally this TRI is defined as p = [ P: B(1t1p, i1 ,uo(x;Mo); z) = 0 l ( 4 .1) Suppose however that the range of traded goods in period 1 is M 1 > M 0 Then the utility level in period 1 is u 1 (x;M 1 ). If the range M 1 were available in period 0, then the corresponding utility would be u 0 (x;M 1 ) instead of u 0 (x;M). Under this setting, the TRI may now be defined as a compensating measure in which the trade restrictions (-r ,Q') of the goods which are common in both periods change so that u 0 (x;M 1 ), not u 0 (x;M), is attained. To be more explicit, suppose that

PAGE 100

93 there are M 0 goods in period O and just one new imported good in period 1 (so that M 1 = M 0 + 1). Then the TRI can be defined as the equiproportionate change in the vectors (i\Q 1 ) of the goods M 0 so that the economy returns not to the initial utility level u 0 (x;M), but to the hypothetical level u 0 (x;M 1 ). For notational convenience let u(x 0 ;M 1 ) = ii and u(x 0 ;M) = u 0 Following the same reasoning as in defining the standard TRI, one can derive the offsetting uniform change in trade distortions ( 1r, Q) required to return the aggregate consumer to the utility level ii (instead of the initial level u). Indeed, one may write P' = [ P': B(P'n 1 ~~,a; z) = o] (4.2) This expression should be interpreted as follows: first, it equalizes the change to the foreign exchange requirement B() due to an arbitrary change in ..-, Q, and u to a hypothetical proportionate change in.,and Q. Then, starting in period 1 it asks the question: what is the hypothetical proportionate change in (p, ..-) which returns the economy to welfare level ii rather than the actual level u? This hypothetical proportionate change, which may be termed the generalized TRI, is determined by totally differentiating (4.2) and then solving for the rate of change (d/3' /(3'): = __ B_b_d_O __ Al I I .., B 0 Q B" 1t B 1 d1t ff da ( 4. 3) a This generalized version of the TRI suggests that the change of rate of the proportionality factor {3' equals the sum of the percentage changes in the trade policy

PAGE 101

94 variables (?r,Q) (each one weighted by its contribution to total welfare cost). Furthermore, it is adjusted by the percentage change in the hypothetical welfare level ii with the respective 'weight' being the ratio of the utility-derivative of the balance of trade function to the total welfare cost. Mathematically, the new features are the numerator i1Bu and the utility rate of change (du/ii) on the far right-hand side of (4.3). Making this generalized TRI operational is the objective of the following section. Operationalizing the TRI in the Presence of New Goods Assume that the preferences of the aggregate consumer are homothetic. Such preferences generate consumer expenditure functions e(1r,p,u) which can also be written as ue(..-,p) where e(..-,p) is the unit-expenditure function. Additionally, given the fact that the measurement of utility is ordinal there is no loss of generality in representing preferences via linearly homogeneous utility functions. Then one may write E(1t,p) = ue (rc,p) g(1t,p) (4.4) ( 4. 5) m = Ew = uew (1t ,p) g._ (1t ,p) ( 4. 6) and t he balance of trade function B( ) pertaining to utility level ii can be written as

PAGE 102

95 B() = a e() -g() t'[u e7( -g 11 J ( 4. 7) Differentiating with respect to ii and premultiplying with ii yields a B = a e() t 1 u e (1-w) (p-p*) a e Q 7( p (4.8) Rearranging the Balance of trade function in ( 4. 7) gives (4.9) Note however that the left-hand side of (4.9) is the derivative iiB 0 Thus the term iiB 11 can be evaluated as (4.10) The rate of change (du/ii) can be estimated by drawing on the work of Feenstra and Markusen (1991). They show that given a constant-returns-to-scale, C E S. production function 0 < 8 < 1 (4.11) and t wo different but overlapping ranges of inputs denoted by M M 1 the respective output levels y = f(x; M) and y' = f(x;M') are related by

PAGE 103

f(x;M 1 ) = f(x;M 0 )p Nl L wixi p = _i_=l __ Mo L wixi i 2 l 96 1/8 ( 4. 12) where w's denote input prices, x's denote input quantities, and () is the exponent coefficient of the underlying C.E.S. function. Drawing on the similarities between the formal structure of the firm's cost minimization problem and the consumer's expenditure minimization problem, an analogous result can be stated in the context of consumption theory. Specifically, given the expenditure minimization problem of the individual consumer II Min(x) E = LP1X1 il s. t. u 0 = u (x;M) where Pi is the (exogenously given) price of the ith good, M ( 4. 13) denotes the set of goods available to the consumer, and u 0 = u(x;M) denotes a certain util i ty level, the following proposition can be stated. Proposition: Given a linearly homogeneous C.E.S. utility function, 0 < 8 < 1 ( 4. 14) the utility levels achieved with two different ranges of goods M 0 and M 1 with M 0 n M 1 ;.c 0 are related as

PAGE 104

u(x;M 1 ) = u(x;M 0 ) p Ml LPiXi p = _i_=l __ No LPiXi il 1 8 97 ( 4. 15) In other words, the ratio of these two utility levels equals the expenditure ratio on the respective sets of goods M and M' raised to the power (1/0). The proof of this proposition is given in Appendix A. Given this proposition, the utility levels u 0 (x;M) = u 0 and u 0 (x;M 1 ) = u 0 can be related as: ( 4. 16) where as noted earlier, p equals the expenditure ratio on the range of goods M 1 to the range M 0 raised to the power (1/0). Since ii is a multiple of u, its rate of change dii/ii, keeping the initial u 0 constant, is du=~ a P (4.17) Further, the term dp/ p can be viewed as the rate of growth of the ratio p between periods O and 1. Thus it may be approximated as = pl po p pl (4.18) If the same range of traded goods exists in both periods then, of course p vanishes and {3 = {3'. Suppose however that M' > M 0 and let N denote traded goods existing only in period 1, so that it holds, M + N = M'.

PAGE 105

98 Since the N new goods exist only in period 1, we have Po = 1, Jll }:pJxl P1 = _i_=l __ M ~pJxl 1 8 M N = fu pJxJ + p}x} Mo EpJxl il 1 6 =R (4.19) In light of expressions (4.9) and (4.17), the rate of change (d/3' !{3') can be finally written as riRI dR 1 (1t-t) 1 g + [p+(l)(p-p)]g ( 1) ....:::..t:...-~ + P 1 (4 20) /-A.-, I I p .., I} B 0 Q-B.1t R The TRI /3' can be thought of as the generalized value-TRI when trade in new products occurs over time. Its rate of change ( d/3' I {3 ') in expression ( 4. 20) consists of two parts: the standard TRI (although evaluated at a different reference equilibrium) and an adjustment factor the second term on the right-hand side of (4.20) which accounts for the presence of newly traded goods in the second time period. With respect to this generalized value-TRI, one can make several observations. First, taking into account different ranges of traded goods over time the value of the standard TRI can change either upwards or downwards, depending on the shadow value of distorted trade. This can be seen by observing the adjustment factor on the right-hand side of (4.20) The numerator and the expression (p 1 p 0 )/p 1 are unambiguously pos i tive. The shadow value of distorted trade (e.g. the denominator (B' QQ B' ,,:i-)) is generally expected to be negative as can be inferred from expressions (2.29)-(2.32)--the trade distortions in other words are welfare-reducing. Therefore the adjustment factor

PAGE 106

99 of the generalized TRI is generally expected to be negative, too. However, there may be situations (e.g., strong cross commodity effects) that render the denominator (B' QQ B' ,.:i-) positive. In this case, the adjustment factor becomes also positive thus enlarging the value of the standard TRI. Intuitively, this change in the size of the standard TRI may be interpreted along the following lines. When only the traded goods, common in both time periods are considered, a certain degree of trade restrictiveness is measured via the standard TRI. If some newly traded goods appear in the second period, the generalized TRI suggests the standard TRI underestimates 17 the level of protection, since it does not take into account the restrictiveness associated with the new goods--this restrictiveness is reflected on the fact that these 'new' goods were not traded at all in the previous period. Second, the value of the generalized TRI depends on whether the 'new' goods are freely traded or restricted. This can be seen via the term (1-1/R) of the adjustment factor in (4.20). If the newly traded goods are priced at the world prices the adjustment factor reduces the standard TRI only in a definitional sense (i.e. the original TRI is evaluated at a different reference point). If however, the newly traded goods are subject to tariffs or quotas, their prices are even higher, and the adjustment factor increases accordingly. This, in turn, reduces the value of the TRI even further. A third point of importance is that the significance of the adjustment factor depends on the value of the new imports. This can be seen by observing the ratio R in 17 However, in situations where the shadow value of distorted trade is positive (i.e., overall, the distortions are welfare-improving) the generalized TRI implies that the standard TRI overestimates the level of protection.

PAGE 107

100 expression (4.19); when the expenditure on new imported goods is large relative to that on the common set of goods M 0 the magnitude of R rises accordingly This is indicative of a situation where lowering the tariffs (and thus the price) of old traded goods is combined with imposing very high tariffs on newly traded products. By contrast, a small expenditure on new imported goods relative to that on the common set M 0 may result in a factor R, near zero In such a case, a restrictiveness measurement based on the standard TRI may be considered adequate. An Application of the Generalized TRI the Case of Japan s Meat Imports In this section, some of Japan's agricultural trade policies discussed in the previous Chapter are used to empirically investigate the generalized TRI. This empirical exercise constructs a TRI pertaining only to Japanese meat imports (beef, pork, and poultry). The interesting feature about this partial index (termed here the meat-TRI) is that it provides an example for adjusting the standard TRI to account for the presence of new' goods (i.e. goods not previously traded). In particular, the beef imports in this meat-TRI are considered to be differentiated products, distinguished by origin (supplier country). Inspection of the data on beef imports by origin (Table C 8 Appendix C) reveals that in the 1970s Japan beef imports were dominated by Australian and New Zealand beef, while U.S. beef imports were negligible. In contrast, U S beef imports rose considerably throughout the 1980s. Therefore one can start the construction of the meat-TRI in the absence of U.S. beef

PAGE 108

101 imports and then adjust it to account for the introduction of a newly traded beef variety (U.S. beef). Beef import differentiation could, of course, be used to compute a generalized TRI for all six products considered earlier. Nevertheless, the small import value of U.S beef in the early 1970s relative to the total import value of the rest of agricultural goods (beef, pork, poultry, wheat, oranges) may yield a negligible adjustment in the standard TRI. Therefore testing the applicability of the generalized TRI is confined to a partial measurement of the trade restrictiveness of meat imports. Following similar steps as in the case of the combined TRI in Chapter 3 a partial equilibrium balance of trade function is specified for Japanese imports on beef, pork, and poultry. Focusing on beef imports, the three major beef suppliers of the Japanese market are Australia, New Zealand, and the U.S.--all three supply more than 90% all beef imports. These suppliers however produce different types of beef. In particular, Australia and New Zealand specialize in the production of grass-fed beef while the U.S. specialize in grain-fed beef. It is often argued (Mori and Gorman, 1985; Mori, Gorman, and Faminow, 1987) that, traditionally, Japanese consumers have a strong preference for tender beef with high a portion of intermuscular fat (beef marbling). Traditional Japanese dishes such as sukiyaki require highly marbled beef, and the Japanese beef grading system is chiefly determined by the extent of marbling. Given the importance of marbling in assessing beef quality, the beef imports may be differentiated by origin (the often called Armington assumption) In other words, Australian, New Zealand, and U.S. beef imports can be

PAGE 109

102 considered as different beef varieties and be treated separately in the balance of trade function on meat imports. Following similar steps as earlier, the balance of trade function on meat imports (assuming complete quota rent retention at home) can be written as (4.21) where (pAus,PrmPus) is the price vector of Australian New Zealand, and U.S. beef, and (~us,Cba,Qus) is the corresponding vector of imported quantities. The TRI on meat imports is then defined as (4.22) and its rate of change can be derived by total differentiation of (4.22) with respect to the respective policy variables. The rate of change of the generalized-meat TRI in the first period that U.S. beef is introduced (i.e. 1969), is given by

PAGE 110

103 (4.23) where j =Australian beef, N. Zealand beef, and i=pork poultry. The derivatives BQ, B,.. refer to the balance of trade function on meat imports (4.21), V 1 is the expenditure on meat including U S. beef in the first period it appears, and yn is the meat expenditure without U.S beef in the same period. For the subsequent periods, wherein price and quantity data are available for all three types of beef imports, the TRI is computed using its standard form (4.24) where j =Australian beef N. Zealand beef, U.S. beef, and i=pork, poultry. It may be recalled that the adjustment factor of the generalized TRI is based on the assumption that the underlying welfare function (here the utility associated with meat consumption) h as a linearly homogeneous C.E.S. specification. Then the information requirements for the computation of the meat-TRI include import demand elasticities on pork poultry, Australian, New Zealand, and U.S. beef, as well as, an estimate for the

PAGE 111

104 exponent coefficient 0 of the underlying C.E.S. welfare function. Information on these requirements can be obtained from empirical estimation of a C.E.S. utility function. Estimation of a C.E.S. Aggregator Function Feenstra (1990) demonstrated an econometric procedure for the empirical estimation of a C.E S aggregator (utility or production) function. His technique basically involves the stochastic specification of a model derived from demand and supply equations of a C E.S. utility (welfare) function. Estimation of this model yields an efficient estimate of the elasticity of substitution a ; a is, of course linked to the exponent coefficient 0 as a = 1 / (1 0). An outline of Feenstra's methodology is presented in Appendix B. Assuming a C.E S utility function which involves i= 1. .. N goods, (only a subset of which may be available in period t), the corresponding expenditure function is given by : P= 1-e e (4.25) where p = {p., ... PN} denotes a vector of prices, b;t are taste parameters, Gt i s the set of goods available to the consumer in period t, and {3=(1-0/8) where 8 is the exponent coefficient of the C.E.S. utility function. The demand equation for the ith good can be written as

PAGE 112

105 (4.26) where si, denotes expenditure shares, and Alog mi 1 = (logmiclogmi 1 1 ) with m =pio sit Further, d 1 =log(e/~ ,) denotes the ratio of unit-expenditure int and t-1 while E; 1 =(logb;c logb; 1 1 ) denotes the random change in the taste parameters b in t and t-1. On the supply side, the supply curve for the ith good is specified m first differences, A, as (4.27) where Alog m; 1 =(logm; 1 -logm; 1 1 ) with m=g 0 X; 1 X; 1 is the quantity supplied, r, denotes the inverse supply elasticity, and V; 1 is a random error. Further, a good k (available to the consumer in all time periods can be used as numeraire and its demand and supply equations can be subtracted from the demand and supply equations, respectively, for all other goods i= 1. .. N, i ;t k. The adjusted demand and supply equations for each good i ;t k are then multiplied with each other to yield the following expression (4.28) where (4.29) (4.30) (4.31a)

PAGE 113

y 2 = p (21-1) ( 1-l) 106 (4.32) This expression relates the logarithmic difference of the price of the ith good, i ;rt. k to the logarithmic difference of its expenditure share. Estimation of the two parameters -y 1 and -y 2 yields an estimate for {3 as can be inferred from (4.32). Subsequently, this estimated {3 can be used to calculate the elasticity of substitution since /3=1/(a-l) In the empirical application considered here, Japan's meat imports are assumed to be associated with a C E.S. utility function so that for each type of meat equation (4.28) holds. Further, New Zealand beef imports were used as numeraire. That is, the differences (dlogp;t dlogp;J and (dlogSjt dlogs;J were formed for Austral i an beef U.S. beef pork, and poultry with New Zealand beef being the numeraire good k. Accordingly, equation (4 28) was specified for the rest of meat imports i.e. (Australian beef U.S. beef, pork and poultry). As shown in Appendix B stacking equation (4.28) for i=Australian beef, US beef, pork and poultry yields an estimable model of the form Y = X -y + w. The data set used for the estimation of this model appears in Table C. 8--A ppendix C. These data are the reported quantity and value of total Japan imports of beef, pork and poultry covering the period 1962-87. Beef imports are dissagregated into three categories: Australian beef, New Zealand beef, and U.S. beef. (United Nations Trade Data Summary). Data on pork and poultry are total imports from all supplying countries

PAGE 114

107 (FAQ Trade Yearbook, various issues). In the case of beef, imports from each of the three major supplying countries (Australia, New Zealand, U.S.A.) are assumed to represent a different variety of beef, as explained earlier. Imports of Australian beef, New Zealand beef, and poultry are reported here for the whole sample period. Negligible imports of U.S. beef (less than 50 metric tons) are omitted so that data on U.S. beef imports start in 1969. For the same reason, pork imports are considered to be zero in 1965, and 1966. Finally, import prices for all five types of meat are calculated as import unit-values (i.e. the ratio of the import value to the respect import quantity). The model Y = X 'Y + w is estimated via instrumental variables (IV) using as instruments a matrix Z of dummy variables for each good i k (shown in equation (B.28)--Appendix B. The instrumental variable (IV) estimator 'Y of the model is consistent but not efficient because it introduces heteroscedasticity in the errors wit. Thus in a second stage, efficient estimates are obtained by using a weighted IV estimator -y. The respective weighing matrix is a diagonal matrix involving the computed residual from the initial IV estimator 'Y. The estimation results of this model are presented in Table 4.1. The upper part of the table shows the estimated parameters along with their standard errors (in parentheses). The consistent estimates (-y 1 ,-y 2 ) (shown in the first row) are computed as in (B.29)--Appendix B. The efficient estimates ('Yi,'Yz) (shown in the second row) are computed as in (B.30)--Appendix B, and their variances are computed as in (B.32) Appendix B. The estimates of A and {3 and their variances are then obtained using the formulas (B.33) through (B.37) in Appendix B.

PAGE 115

108 Table 4 1 Parameter estimates of a C.E.S. welfare function Japan meat imports 196288. Coefficients: Consistent estimate Efficient estimate ')' I 0.05417 (0.0298) 0.0635 (0.0245) Autocorrelation test statistic 1.205 Specification test statistic 0.814 Observationsb: 93 -0.6681 (0.428) -0.6668 (0.4142) Numbers in parentheses denote standard errors. A 0.0896 (0.0995) 0.1011 (0.0964) {3 0.7410 (0.3591) 0.7513 (0.3425) x 2 (3)o os 7.81 x 2 (2)o os 5.99 b The number of observations over years and supplying countries. It may be noted that, as expected, the standard errors of all the efficient estimates are reduced relative to those of the consistent estimates. Focusing on the efficient estimates, the estimated -y 1 is statistically different from zero at a=0.01 level while the estimated -y 2 is statistically different from zero at a =0.12 level. Additionally, the estimated 13 is statistically different from zero at a=0.025 level while the estimated ,._ is statistically different from zero at a =0.15 level. The lower part of Table 4 1 reports statistics used to test the existence of autocorrelation and mispecification in the estimated model. As in Feenstra (1990), the White and Domowitz (1984) test was used to test for the existence of autocorrelation. This test is carried out by first computing the residuals

PAGE 116

109 (4.33) and then performing the regression (4.34) where the computed X, and X 2 are obtained by regressing X 1 and X 2 on the instrument matrix Z. The existence of autocorrelation can be tested under the null hypothesis of no autocorrelation by using the statistic nR 2 which is asymptotically distributed as x2(k); k is the number of regressors, n is the number of observations, and R 2 is the coefficient of determination of the regression (4.34). In addition, the Hausman (1978) test was used to test for mispecification. This specification test contrasts an efficient estimator (such as -y with an inefficient but generally consistent estimator (such as -y), under the null hypothesis of no mispecification. The test statistic has the quadratic form (?*-?)' [Var(?) -Var(y*) J1 (?*-?) (4.35) and is asymptotically distributed as x 2 (k) where k is the number of elements in the vector -y. The computed values of both the Domowitz-White, and Hausman statistics are below the relevant tabulated values at a=0.05 level. Thus, both the null hypotheses of no autocorrelation and no mispecification cannot be rejected. Given consistent estimates of /3 and A, the elasticity of substitution a and the exponent coefficient () are then computed as

PAGE 117

a = i + p p 8 := a 1 a 110 (4.36} At this point it may be noted that, given the small estimated value of A and its relatively large standard error, the parameter A may considered to be zero. As explained in Appendix B, this implies a very flat (almost horizontal) supply curve for meat. This finding is interesting because it implies a meat sector consisting of relatively competitive producers, as one might expect. Furthermore, ignoring 18 the small value of A the estimated values of {3 and a are = 0.6668, a= 2.499, 8=0.599 (4.37} These values are used in the next section for the computation of the TRI in Japan's meat sub-sector. Specifically, the estimated exponent () is used in the calculation of the adjustment factor of the TRI in 1969 as shown in (4.23). Additionally, it may be recalled that the TRI requires own-price and cross-price elasticities of demand for the five types of meat imports. It turns out that these elasticities can be conveniently computed given the estimated elasticity of substitution a. This can be demonstrated by considering equation (B.12) from Appendix B which, for convenience, is repeated below 18 Alternative computations taking into account the value A = 0.1011 yield a = 2.33 and () = 0.570 with insignificant numerical changes to the rest of the analysis.

PAGE 118

Differentiating with respect to logp;, and logpjt yields respectively a1ogxit a1ogpjt 111 (4.38) (4.39) (4.40) These formulas were used to compute own and cross price elasticities of demand for imports of Australian, New Zealand, and U.S. beef, pork, and poultry. Since a C.E.S. functional form is commodity-wise weakly separable, the estimated cross-price elasticities for any commodity i are identical across commodities. Additionally, all goods are price substitutes with each other. These elasticity estimates are reported in Table 4.2 and are utilized next in the calculation of the respective TRI on meat imports. Calculating the TRI of Japan's Meat Imports The TRI pertaining to Japanese imports of Australian, New Zealand, and U.S. beef, pork, and poultry is calculated in an analogous fashion to the combined index. However the TRI of meat imports differs from the combined index as the import demand elasticities for the five types of meat involved here are based on a C.E.S. specification of the underlying welfare function. Concerning the calculation of the quota rents for the three types of beef, the domestic wholesale prices for Australian, New Zealand, and U.S. beef were not

PAGE 119

112 Table 4.2 Own price-elasticities of import demand, assuming a C.E.S. utility function on meat imports, 1969-87. Austra New U.S lian Zealand Year beef beef beef pork poultry 1969 -2.125 -2.420 -2.487 -0.921 -2.042 1970 -1.612 -2.375 -2.432 -1.472 -2.105 1971 -1.457 -2.394 -2.466 -1.666 -2.013 1972 -1.603 -2.438 -2.473 -1.263 -2.219 1973 -1.415 -2.411 -2.350 -1.443 -2.377 1974 -1.521 -2.430 -2.249 -1.606 -2.191 1975 -2.182 -2.461 -2.402 -0.617 -2.334 1976 -2.028 -2.464 -2.346 -0.848 -2.310 1977 -2.013 -2.458 -2.398 -0.928 -2.199 1978 -1.967 -2.443 -2.334 -1.064 -2.189 1979 -1.790 -2.470 -2.263 -1.251 -2.222 1980 -1.692 -2.455 -2.223 -1.424 -2.202 1981 -1.992 -2.455 -2.283 -1.101 -2.165 1982 -1.954 -2.469 -2.191 -1.257 -2.124 1983 -1.951 -2.447 -2.208 -1.201 -2.190 1984 -1.995 -2.455 -2.211 -1.165 -2.169 1985 -2.024 -2.456 -2.154 -1.159 -2.202 1986 -2.123 -2.474 -2.180 -1.105 -2.114 1987 -2.141 -2.472 -2.125 -1.106 -2.152 (continued)

PAGE 120

113 Table 4.2 continued. w.r.t w.r.t w.r.t w.r.t w.r.t Austra New U.S. lian Zealand Year beef beef beef pork poultry 1969 0.374 0.079 0.012 1.578 0.457 1970 0.887 0.124 0.067 1.027 0.394 1971 1.042 0.105 0.033 0.833 0.486 1972 0.896 0.061 0.026 1.236 0.280 1973 1.084 0.088 0.149 1.056 0.122 1974 0.978 0.069 0.250 0.893 0.308 1975 0.317 0.038 0.097 1.882 0.165 1976 0.471 0.035 0.153 1.651 0.189 1977 0.486 0.041 0.101 1.571 0.300 1978 0.532 0.056 0.165 1.435 0.310 1979 0.709 0.029 0.236 1.248 0.277 1980 0.807 0.044 0.276 1.075 0.297 1981 0.507 0.044 0.216 1.398 0.334 1982 0.545 0.030 0.308 1.242 0.375 1983 0.548 0.052 0.291 1.298 0.309 1984 0.504 0.044 0.288 1.334 0.330 1985 0.475 0.043 0.345 1.340 0.297 1986 0.376 0.025 0.319 1.394 0.385 1987 0.358 0.027 0.374 1.393 0.347

PAGE 121

114 available. Thus the total quota rent on beef was distributed among the three supplying countries using as weights the respective import value shares 19 Additionally, inspection of the computed cross price-elasticities reveals that the cross price-elasticity with respect to pork imports is relatively high--indeed for half of the examined years, this elasticity is higher than the own price-elasticity of pork imports. This is due to the fact that the import value-share for pork is predominant, ranging from 40% to 60% of the value of all meat imports. Since own and cross price-elasticities are estimated via equations (4.39) and (4.40), this results in cross price-elasticities with respect to pork that are higher than the respective own price-elasticities of pork. The high substitutability of the rest of meat types with respect to the price of pork renders the derivative with respect to price of pork Sii(pt), negative. Intuitively, this implies that a rise in the price of pork reduces, instead of increasing, the income required to preserve a certain welfare level--in other words, it is welfare-improving. The reason is, of course, the high degree of substitutability of the other meat categories with pork. 19 The total beef quota rent is R bf = Qbf (pbf P;f) where Qbr is total beef imports. This rent is assumed to be distributed among Australia, New Zealand, and the U.S. according to their import value shares. Hence, for Australia RAUS= ( ~A::)[Qbf(pbf-p;f)] The rent per physical unit (i.e. Y /MT) is RAUS = ( ::: ) (pbfP;f)

PAGE 122

115 Moreover, the shadow value of distorted trade associated with pork--the term Bi,(J>k>h(pk)--tums out to be larger than the shadow values associated with the other policy instruments (i e., the poultry price and the beef quotas). As a result, the overall shadow value of distorted activity (B 'QQ B' ,,:r) is positive for most of the examined years. This, in tum, adversely affects the performance of the TRI: now the individual components of the index do not necessarily change in accordance with the changes in the policy instruments (domestic prices of tariff-ridden goods and quota levels). In other words, a relaxation of a quota (or a reduction of the price of tariff-ridden goods--pork poultry) does not necessarily raise the index, as the algebraic sign of the 'weight' may reverse the change of the respective policy instrument. The results of computing the TRI on meat imports for the period 1968-69 through 1986 87 appear in Table 4.3. Table 4.4 then presents the percentage changes in the respective policy instruments--pork prices, poultry prices beef quotas--over the same period. It can be seen that with the exception of a dramatic variation in 1972-1974, this TRI shows relatively low change over time. Furthermore the rate of change of the index is primarily shaped by the changes in domestic pork prices followed by changes in poultry prices and to a lesser extend by changes in beef imports. This is due to the fact that the shadow value of distorted activity associated with pork prices is the largest portion of the overall shadow value of distorted trade. Consequently, percentage changes in pork prices comprise the largest component of the index. Upon close examination, the index captures the abrupt changes of agricultural import policies of Japan in the early 1970s-especially the sharp rise of all beef imports

PAGE 123

116 Table 4.3 The Trade Restrictiveness Index of Japan's meat imports, 1969-87. Year Pork Poultry Australian beef New Zealand beef 1968 1969 -0.19412 -0.06422 -0.02383 -0.00679 1970 0.595683 0 -0.02598 0.008347 1971 -0.3491 0.277394 -0.234 -0.07451 1972 -0.16791 -0.16796 -0.11402 0 002716 1973 2.607768 -2.25562 6.557446 0.85529 1974 -0.50446 0.321219 0.914166 0.206341 1975 -0.58578 0.098689 0.026548 -0.00509 1976 -0.03159 0.094805 -0.76237 -0.02262 1977 0.175971 -0.04749 0.166626 0.017643 1978 -0.11854 0.147654 0.113734 0.039707 1979 -0 88779 0.24539 0.788355 -0.1321 1980 -0.22296 0.409131 0.263379 -0.02411 1981 -0.70429 0.249391 0.139392 -0.07137 1982 0.538008 -0.28383 0.027766 0.103106 1983 -0.19043 -0.27208 -0.29371 -0.26449 1984 -0.53664 0.178829 0.077559 -0.01824 1985 -0. 70198 0.161476 0.016868 -0.01342 1986 -0.0416 0.0256 0.088311 0.000273 1987 -0.02528 0.028785 0.075706 -0.00054 (continued)

PAGE 124

117 Table 4.3 continued. Year U.S beef TRI TRI level Adjusted TRI level TRI in 1969 1968 1 1 1969 -0.28897 0.71103 -0.08517 0.914827 1970 -0.01216 0.565887 1.113393 1.432516 1971 -0.00868 -0.38889 0.680405 0.875424 1972 -0.00258 -0.44975 0.374392 0.481701 1973 2.358343 10.12322 4.164443 5.358063 1974 0.050074 0.98734 8.276162 10.64829 1975 0.072174 -0.39346 5.019839 6.458635 1976 -0.3551 -1.07688 -0.3859 -0.49651 1977 0.338734 0.651479 -0.63731 -0.81998 1978 0.230617 0.413174 -0.90063 -1.15878 1979 0.634138 0.647992 -1.48424 -1.90965 1980 0.07974 0.505185 -2.23405 -2.87438 1981 -0.15864 -0.54552 -1.01533 -1.30635 1982 -0.25787 0.127175 -1.14446 -1.47248 1983 -0.50784 -1.52854 0.604892 0.778267 1984 0.521979 0.223489 0.740079 0.952201 1985 0.101071 -0.43599 0.417414 0.537054 1986 0.175744 0.248324 0.521068 0.670418 1987 0.15782 0.236498 0.6443 0.828971

PAGE 125

118 Table 4.4 Percentage changes in price and quantity distortions of Japanese meat imports, 1969-1987. Year % change % change % change % change % change in in 1\,1: in heh in ~us in ~z Qus 1968 1969 0.089 -0.060 0.334 0.254 1970 -0.212 0.000 0.252 -0.227 0.732 1971 0.079 0.103 0.456 0.373 0.286 1972 0.055 -0.106 0.299 -0.035 0.151 1973 0.075 0.119 0.509 0.591 0.937 1974 0.118 0.135 -1.533 -2.231 -0.235 1975 0.248 0.091 -0.141 0.166 -1.175 1976 0.005 0.034 0.513 0.213 0.685 1977 -0.020 -0.011 -0.061 -0.157 -0.551 1978 -0.059 -0.171 0.075 0.502 0.430 1979 -0.124 -0.076 0.228 -1.258 0.458 1980 0.019 0.074 -0.081 0.125 -0.049 1981 0.094 0.077 -0.069 0.361 0.152 1982 -0.055 -0.062 -0.011 -0.687 0.162 1983 0.011 -0.034 0.054 0.529 0.163 1984 -0.023 -0.017 0.010 -0.020 0.107 1985 -0.211 -0.098 0.012 -0.089 0.092 1986 -0.051 -0.065 0.116 -0.152 0.266 1987 -0.076 -0.198 0.132 0.232 0.257

PAGE 126

119 in 1973 and their abrupt reduction in the next year. Indeed from Table 4.4 it can be seen that in 1973 Australian beef imports increased by 51 % New Zealand beef imports by 59%, and U.S. beef imports by 94% while in 1974 Australian imports plunged by 153%, New Zealand beef imports by 223% and U.S. beef imports by 23 5%. At the same time, pork and poultry prices rose by 7.5% and 12% respectively in 1973 and 12% 13.5% in 1974 The severe variations in the beef quotas are due to the policies of the Japanese government which suspended the beef quota in late 1973 and completely closed the Japanese market to beef imports in 1974. From the second half of the 1970s to the end of the examined period (1987), t he index shows negative rates of change in the years 1976, 1981, 1983, and 1985 suggesting that in periods t = 1976, 1981, 1983, 1985 there was increased restrictiveness relative to the respective (t-1) periods For the rest of this sub-period the index shows in general, low positive rates, suggesting slight liberalization on a year-to-year basis. Additionally, one may observe that during the 1980s the increasing U.S. beef quota is the main contributor of the index, among the three beef quotas. A graphical representation of the rates of change of the meat-TRI is given in Figure 4.1. It may be noted that this graph can also be interpreted as showing the level of the meat-TRI on a year-to-year basis--that is, the level of the index between any t and (t-1) periods. Furthermore, the meat-TRI was adjusted in 1969 to account for the introduction of U.S. grain-fed beef. As demonstrated in Chapter 3, the adjustment term equals the ratio of the production value (evaluated at domestic prices, net of levies or tariff rates) of the tariff-ridden goods over the shadow value of trade distortion, times a factor

PAGE 127

120 11 10 g 8 E 7 u 6 ... 0
PAGE 128

121 involving the expenditure share of the 'new' good(s) relative to the expenditure of all goods in the index. The result of this adjustment is reported in the fifth column of Table 4.3. Given a positive shadow value of distorted trade, the adjustment reduced the negative rate of change of the standard meat-TRI by 70% implying a larger size meat index and therefore lower trade restrictiveness for meat imports in 1969. Figure 4.2 plots the levels of the standard and the adjusted meat-TRI computed according to the chain-principle, against time. It may be recalled that when the level of the TRI is computed via the chain-principle, one compares the magnitude of the index in all subsequent periods with that of the initial period (here 1968) assuming that initially the index equals one. Up to 1972, the level of the index is generally decreasing indicating rising protection. Then in 1973 and 1974 the index level is steeply raised implying a dramatic relaxation of trade protection. Nonetheless, the level of the index plummets from 1974 to 1980, assuming even negative values after 1976. In 1980, the index is at its lowest level relative to the initial (base) period. After 1980 however, the index level starts increasing as one moves towards the end of the period under examination. Yet, in 1987 the size of the index is 35. 5 % smaller than the size of the index in 1968. One may conclude that, using the early 1970s as reference point, there has been increasing restrictiveness in Japan's meat sub-sector throughout the 1980s. Although this restrictiveness is somewhat counterbalanced after 1980, the degree of combined trade restrictiveness of meat imports, at the end of the period is still higher relative to that of the base period.

PAGE 129

122 11 1 0 9 8 7 er I6 .., ll 5 P. 4 .., .... 3 0 Cl) 2 > Cl) ....I 1 0 -1 -2 -3 196 8 1970 1 97:5 1980 198'.5 19 87 Y e a r s Stande.rd TR I + General I zed TRI Figure 4. 2 The level of the standard and the generalized TRI of Japanese meat imports (1968= 1), 1968-87.

PAGE 130

123 Moreover, the level of the generalized value-index, computed according to the chain-principle, is 28. 7% higher than the level of the standard meat-index throughout the period 1969-87, due to the new-goods adjustment in 1969. As a result, the standard TRI suggests that relative to the initial period, the index level in 1987 is smaller by 35 .5 % while the generalized TRI suggests that the same level is smaller by 17 .1 % In other words although both the standard and the generalized TRI imply higher restrictiveness in the end relative to the initial period, the latter suggests that the magnitude of this restrictiveness is smaller than that suggested by the standard index. It should be stressed nonetheless, that the new traded goods-adjustment does not alter the main conclusions on the pattern of trade protection over time provided by the standard TRI. This is because the introduction of a newly traded good occurred only once, in the beginning of the period under examination, resulting in a uniform adjustment in the size of the standard index throughout the examined period. Nevertheless, such an adjustment may be seen as a suggestive remark on how the magnitude of the standard index can change. The size and the importance of such a change would depend on whether the introduction of new goods occurs more than once, in the middle or in the beginning of the examined period, and the relative expenditure value of the newly traded good(s). Finally, the direction of such a change would depend on the algebraic sign of the total shadow value of distorted trade (as measured by the term (B' QQ B' ,:ir)) at the period wherein the new goods-adjustment takes place.

PAGE 131

CHAPTER 5 SUMMARY AND CONCLUSIONS This study is an attempt to re-evaluate existing measures of trade protection and empirically investigate new ones, in search of measures that satisfactorily aggregate diverse types o f trade distortions and provide a valid estimation of the magnitude of trade restrictiveness across countries and over time. A review of the existing literature on the measurement of trade protection shows that the major protection measures can be classified into four groups: (a) the Average Tariff, (b) the Nominal and Effective Rate of Protection, and the Domestic Resource Cost (DRC), (c) the Producer/Consumer Subsidy Equivalent (PSE/CSE), and (d) the Trade Restrictiveness Index (TRI). Investigating the existing measures of protection reveals that they are not directly comparable; they differ in their definitions, and the sets of policies they cover may include indiscriminately, support policies with weak impact on trade. Additionally, nonprice distortions are either left out of the analysis or only implicitly taken into account (e.g. via shadow evaluation). Focusing on the theoretical robustness of the available protection measures and their ability to cover diverse sets of distorting policies shows that the TRI has some clear advantages over the rest of the measures. These include the robust theoretical derivation of the TRI which enhances the validity of the index, the fact 124

PAGE 132

125 that the index aggregates consistently the level of restrictiveness across a set of commodities and most importantly, the fact that it explicitly incorporates both price and nonprice distortions thus allowing the modelling of a wide array of trade policies. Further this study investigates the TRI empirically, by implementing the index in Japan's agricultural sector. Japan is chosen as the typical example of a country that protects the domestic agricultural sector predominantly by tight border measures along with support programs within the farm sector itself. Although the coverage of trade policies and administrative programs is not comprehensive this study examines a subset of agricultural policies which comprises the most well known trade restrictions in Japan s agricultural trade. In particular the level of trade protection is measured, ceteris paribus, by means of the TRI for the following farm products: ( 1) beef, (2) pork (3) poultry ( 4) wheat, (5) rice, and (6) fresh oranges. By estimating a set of price elasticities of imports and utilizing information on imported quantities, domestic and international prices the combined TRI of the aforementioned products is calculated for the period 1971-87. The computed index indicates a dramatic rise in trade protection in the early 1970s, followed by a slow process of easing price and quantity restrictions so that a slight trade liberalization trend appears towards the end of the examined period. It is worth noting that these conclusions are in contrast with more conventional ad hoc measures of protection. Specifically, a comparison of the TRI approach with some weighted averages of Japan's PSEs/CSEs computed by the USDA over the period 1982-87 shows that while

PAGE 133

126 the PSE/CSE approach implies diminish distortions only in two years (1984-85), the TRI approach suggests diminishing overall distortion throughout the 1982-87 period. Moreover, this study attempts a generalization of the standard TRI by deriving the index in the presence of different sets of traded goods over time. Under the maintained assumption of a C.E.S. welfare function, it is shown that when newly traded goods are introduced over time, the standard TRI is adjusted by a factor that depends on the relative expenditure value of the new goods, and the elasticity of substitution associated with the underlying C.E.S. welfare function. This generalized version of the TRI is then empirically evaluated by calculating a partial TRI pertaining to Japanese meat imports (beef, pork, and poultry) for the period 1969-87. Beef imports in the meat-TRI are considered to be differentiated products distinguished by origin (supplier country). Thus the meat-TRI comprises five types of meat (Australian beef, New Zealand beef, U.S. beef, pork, and poultry). Prior to 1970s U.S. beef imports to Japan were essentially zero. In contrast, U.S. beef imports rose considerably in the 1980s. Thus the calculation of the meat-TRI starts in the absence of U.S. beef imports (1968-69) and the index is adjusted in 1970 for the introduction of a newly traded beef variety (U.S. grain-fed beef). For the computation of the meat-TRI, the utility associated with these five types of meat is assumed to have a C.E.S. functional form with elasticity of substitution greater than unity. This elasticity of substitution is empirically estimated via an instrumental variables technique suggested by Feenstra (1990). Estimation of the elasticity of

PAGE 134

127 substitution makes computable not only the adjustment factor of the generalized TRI but also the demand price elasticities of the goods involved. The calculated index of Japanese meat imports shows considerable variation in the early 1970s (specifically, in 1973-74) but relatively low variation afterwards. The index suggests that the overall trade restrictiveness of Japanese meat imports increased throughout the 1970s and early 1980s followed by a slight trend of trade liberalization towards the end-period (1987). Furthermore, the adjustment of the standard TRI to account for the introduction of U.S. beef imports in the 1970s yields a 28.7% rise in the magnitude of the standard index throughout the examined period. Hence the main conclusions of analyzing the trend of combined protection on meat imports are not altered--except, of course, for the actual size of the index when computed relative to the initial period. This is because the adjustment of the index takes place only once in the beginning of the examined period. More importantly, the empirical exercise of computing the standard and generalized meat-TRI yields some interesting findings about the performance of the index. Specifically, in the case of the meat-TRI, the computed import price elasticities of meat imports suggest high substitutability of all other meat types with respect to pork. This high degree of substitution implies that changes in the price of pork are welfare improving rather than welfare-reducing (due to large cross-commodity effects). This means that in computing the (minus) total welfare cost i.e. the term (B' QQ -B' ,..71") welfare gains may cancel welfare losses, so that this term assumes a positive rather than a

PAGE 135

128 negative value. This is indeed the case with Japanese meat imports where pork welfare gains dominate over welfare losses in most of the examined years. Such cases alter the performance of the TRI as follows: changes in the components of the index do not necessarily change according the changes in the policy instruments (domestic prices, quota levels). This is not true for 'normal' cases where all distortions cause welfare losses. It can be verified that in the case of the combined TRI in Chapter 3, the components of the index change in accordance with the changes in the policy instruments. Thus, it must be stressed that for trade regimes involving both tariff and quotas, one should interpret the TRI rriore carefully. If changes in trade distortions result in both gains and losses, the index may register not only the restrictiveness due to changes in trade policies but also implicit restrictiveness arising from cross-commodity effects. A corollary of this analysis is that the adjustment factor in the generalized TRI may raise or reduce the standard index depending on whether the shadow value of distorted trade (B'QQ -B',..1r) is positive or negative. In other words, the size of the standard TRI is increased or decreased, depending on whether the distortions in place are welfare-improving or welfare-reducing, respectively. Concerning the calculations of tariffs and quota rents used in the computation of the TRI it is often argued that the use of Purchasing Power Parity (PPP) rates are preferable than the use of official exchange rates in converting world prices expressed in foreign currency to domestic currency. A recalculation of the combined TRI of Chapter 3 using PPPs changed only the actual values of the index in every period but left

PAGE 136

129 the pattern of change of the index unaffected. Thus the major conclusions drawn from this index remain robust to PPP conversions of the respective international prices. In the case of the meat-TRI however, the use of PPPs altered the pattern of the index. This is because the shadow values of distorted trade turned to be positive (or negative) in different periods when PPP conversions were used. As result, not only the value but also the pattern of change of the meat-index in the various years changed, too. Summarizing, this study concludes that the TRI approach offers a particularly appealing tool in the empirical research to evaluate trade barriers across countries and over time. Nonetheless, it must be noted that the comprehensiveness of the TRI in the e ~ i rical context depends critically on the statistical information available on a country or period of time. Utilization of the flexibility of the index in modeling price restrictions, quantitative constraints, non-traded goods policies, and factor market distortions requires detailed data sets on protection policies that go well beyond the trade data on quantities and values, usually available. Information availability is also vital for the aggregation level of the TRI. Although the index consistently aggregates trade policy distortions across commodities, inclusion of all policies that significantly affect imports(exports) may not be feasible due to the lack of statistical information. Hence surveys on the protection policies pursued by various countries accompanied by detailed statistical information are critical in empirical implementations of overall trade restrictiveness.

PAGE 137

130 Suggestions for Further Research Future empirical work utilizing the TRI approach may focus on an indepth analysis of protection across countries provided that comprehensive data are available to the researcher. This may provide consistent aggregate measures of trade protection that are free from the inadequacies inherent on commodity-based measures, and may allow useful inferences on how the level of trade restrictiveness develops across a set of countries, over time. On theoretical grounds the TRI may be further refined to take into account the effect of exchange rate fluctuations on the international prices used in the computation of the index In addition, research on how to incorporate empirically into the TRI, changes of international prices (thus eliminating the small-country framework) will improve the comprehensiveness of the TRI. Lastly, research on approximating the price elasticities of the TRI with nonparametric estimates will largely enhance the appeal of the index. Turning to the calculation of the TRI in the presence of different sets of traded goods over time, future empirical work could produce particularly interesting results if there is recurrent introduction of high value, newly traded goods over the period under investigation. Finally, operationalizing this generalized TRI while relaxing the maintained assumption of a C.E.S. underlying aggregator function is definitely a desirable task.

PAGE 138

APPENDIX A PROOF OF THE PROPOSITION Given a linearly homogeneous C.E.S. utility function, the minimization problem of consumer expenditure is stated as s. t. u =u(x;M) (A. 1) The first-order condition for the ith good is where represents the Lagrangian multiplier of the expenditure minimization problem (A.l). In principle, the utility function u(x; M) includes all possible goods which may become available to the consumer. Suppose that the range M of all possible goods is split into two sets M = M 0 + M 1 (A. 3) Let only M 0 goods be available to the consumers; then summing (A.2) over M yields 131

PAGE 139

132 (A. 4) Suppose now that M 1 goods become also available to the consumers. Then summing (A.2) over M 1 yields or (A. 5) Dividing equation (A.4) into (A.5), and raising to the power (1/0) gives 1 Ml 8 LPiXi i=l p 1 1 8 1' 1 8 L aixJ = i=l No LaixJ i l u(x;M 1 ) u (x; M 0 ) (A. 6) Q E.D.

PAGE 140

APPENDIX B STOCHASTIC SPECIFICATION OF A C.E.S. AGGREGATOR FUNCTION Following Feenstra (1990), suppose that in general, the set U ={l, ... ,N} of goods may be available to the aggregate consumer. Let the total number of adjacent timeperiods (i.e. t and t-1) be denoted by the set H={2, ... ,T}. The number of adjacent timeperiods for good i is then denoted by Hie {2, ... ,T} and Ti denotes the number of elements in Hi i.e. the number of adjacent periods for which the ith good is available. Furthermore consider a utility function for the aggregate consumer which has a C.E.S. functional form: (B. 1) where u denotes aggregate utility or welfare, xit denotes the demanded quantity of the ith good in period t, G 1 C { 1, ... ,N} denotes the set of goods available to the aggregate consumer in the same period, and a; i = 1, ... ,N are taste parameters. It is convenient to rewrite the utility function as: 1 u (xt, Gt, b) = (!: (bitxit) 6 ) 8 J.ESe (B. 2) Then it can be shown (e.g. Varian, 1984) that the corresponding expenditure function is given by: 133

PAGE 141

1-e P= e 134 (B. 3) where p = {p 1 PN} denotes a vector of prices; /3 is linked to the elasticity of substitution a as /3=1/(a-l). The unit-expenditure function of the ith good is given by: Differentiating and simplifying gives or b. = e -2..!: t Pit (B. 4) (B. 5) (B. 6) where S;t is the expenditure share, owing to the envelope theorem properties of the expenditure function. Logarithmizing through, this expression is written in periods t and t-1 as (B.7) (B.8) Letting denote first-differences, one may write (B. 9)

PAGE 142

135 where di= (e;/e; 1 1 ) denotes the ratio of unit-expenditure for the ith good in t and t-1 while 1 = (lnb; 1 -lnb; 1 1 ) denotes the random change in the taste parameters b in t and t-1. Expression (B. 9) will serve as a reduced form of the demand function for the ith good. At the supply side, the supply curve for the ith good can be written in first differences as (B.10) where r, denotes the inverse supply elasticity and v; 1 is a random error. The quantity term X; 1 in (B. 10) can be replaced by the expenditure share S; 1 as follows: Shephard's lemma yields (B.11) Substituting (B.5) into (B.11), the expenditure-minimizing quantity X; 1 is written as (B.12) and in first-differences the same expression is written as (B.13) recalling that 1/,S=(u-l). Thereafter (B.10) and (B.12) can be used to solve for the price and obtain

PAGE 143

where vit Xu=---( 1 +11 (J) / 136 (B.14) I..= T) (o-1) (B.15) ( 1 +11 (J) Equation (B.14) will serve as a reduced form for the supply function of the ith good. An interesting point about this equation is the coefficient A. This coefficient accounts for the effect that a change in variable E;t has on the supply price. Given that the variable E;t is defined as the change in the taste parameters b; 1 a positive value of E;t implies an outward shift in the demand for good i when a> 1. If A = 0, this means that outward shifts in demand leave the supply price unaffected; in other words the supply curve is horizontal. Thus far a stochastic demand and supply model has been generated consisting of equations (B.9) and (B.14). The estimator set in this model may be viewed as an application of the generalized method of moments (GMM) estimators developed by Hansen (1982). Thus E; 1 and V; 1 are assumed to satisfy the following properties: (i) E; 1 and V; 1 are heteroskedastic and have zero means, (ii) E; 1 and V; 1 are stationary with asymptotic variances plim 1 2 2 H.-oo H .L..J Eit = Of.i .l i tEH1 plim 1 2 2 H.-oo H .L..J vit=avi .l i tEH1 respectively, and (iii) V; 1 is independent of Ejk for all t E g, k E Hj with (B.16)

PAGE 144

137 (B.17) This assumption implies that the error terms in the demand and supply equations are independent for all products i and j so that the model's parameters are identified. A final simplifying step in the specification of the model is the elimination of the terms dt and Kt from (B.9) and (B.14), respectively. In particular, suppose that some good k is available to the aggregate consumer in all time periods T. Subtracting the demand (supply) equations for the kth good from the demand (supply) equations of all other goods i -;c k yields (B.18) (B.19) Multiplying these two equations and then dividing by (1-A) yields (B.21) where (B.22)

PAGE 145

1p2 (1-1)' y 2 = p (21-1) ( 1-1) 138 (B.23) (B.24) (B.25) (B.26) Equation (B.21) holds for all goods i ;ck and for all periods these goods are available i.e. tEHi. Thus an estimable equation system is formed by writing out (B.21) for all the adjacent periods Ti wherein the good i is available and then by stacking these equations for all goods i ;ck. This results in a system Y = Xy + w (B. 27) with total number of observations M=Li;,
PAGE 146

139 z = (B.28) where 1 1 is a T 1 xl vector of ones, 1 2 is a T 2 xl vector of ones etc. for each good i~k. Thereafter the parameter vector (-yi,-y 2 ) can be estimated by using the standard instrumental variables (IV) estimator2 P = Z(Z 1 Z) -lzt z (B.29) The IV estimator however is not the most efficient one. Feenstra (1990) demonstrates that the rank condition for (-yi,-yi) to be identified results in heteroskedasticity in the error term wit The efficient estimator r is obtained (White, 1982) as g = z ( z' z) 1 z'x (B. 30) where W is a MxM diagonal matrix with elements w? repeated times on the diagonal for i=l, ... N, i~k and 20 The variances of the IV estimators corrected for heteroscedasticity are computed as (i'x) 1 x'wi(x'i> 1 where the diagonal matrix Wis shown in (B.30).

PAGE 147

140 (B. 31) i.e. the computed residuals using the initial estimates 'Y Moreover the covariance matrix of the efficient estimator 'Y is estimated as a = ,. and {3. At this point it must be stressed that a prerequisite for obtaining meaningful estimates of >,. and {3 is that the estimated 21 'Y,)0. Then estimates for >,. and {3 are obtained as follows: p = ( 1 l )1 > o 2l 1 2 (B.33) (B.34) Moreover, the variances of {3 and >,. can be computed by taking first order approximations around the expressions (B.33), and (B.34). In particular, it is shown (Feenstra, 1990, Appendix) that 21 If the estimated 'Y,(0 then the estimated>,. and {3 are not in the ranges {3)0 and O(>,.( 1 as expected. In such a case one should conclude that the data do not support the hypothesis of a C.E.S aggregator function with <1) 1.

PAGE 148

141 (B.35) and var~= ( \-A ) 2 var? 2 +( /~ )varA2 -1 (2 -1) 4 -(2?2(1-A))cov(A,?) (2A-1) 3 2 (B.36) where (B.37)

PAGE 149

APPENDIX C DESCRIPTION OF THE DA TA AND THEIR LIMITATIONS Price and quantity data on each of the six goods considered here are reported in Tables C l through C.8 in this Appendix and are discussed below. The relevant data series are compiled from three sources: USDA(1983) for the 1970-78 period OECD(l987) for the 1979-81 period and USDA(l990) for the 1982-87 period. When possible, price definitions, conversion factors etc. used in these studies were also used here. Starting with beef, the respective international price is calculated as the ratio of total import value to total import quantity (import unit value). Then the quota rent is given as the difference between the domestic wholesale price and the import unit value. Finally, the reported quantities of total beef imports are used as the respective quota levels in each period In the case of pork imports, the unit import value is not an adequate international price as importers tend to import high quality and better cuts of pork in order to higher tariff rates (OECD 1987). Consequently, using the actual unit import value may result in underestimation of the respective tariff rates. Alternatively, a possible international price for pork can be computed as the U.S. wholesale pork price adjusted for transportation costs. 142

PAGE 150

143 Following the procedure used by the OECD(l987) an international price for pork can be calculated as follows. First, the U.S. wholesale (pork carcass) price is adjusted by a factor of 20% to reflect transportation and insurance costs. In addition the c.i.f. pork price so obtained is adjusted for technical conversion rates which are not identical in Japan and the U.S. In particular the ratio of carcass weight to live weight is 62 % in the U.S.; this ratio is assumed to be 65% in Japan. Hence the (estimated) c.i.f. price for pork is further adjusted by a factor of 95% ((0 62/0.65)=0.95) to reflect the difference in the weight conversion rates. In summary the international pork price h*p11: is calculated by using the formula: h = ( ( 1. 2 o) h%; (YI$) ) ( ) plc ____ ...;..____ 0. 9 5 0.454 (C. 1) where (Y/$) is the respective exchange rate and 1 lb=0.454 kgr. The calculation of the pork tariff is carried out as follows: when the international pork price is above the domestic price, the tariff rate is calculated as 5 % of the international price. For the periods where international pork price is below the domestic price the tariff rate equals the difference between the two prices, as long as this difference is higher than 5 % of the international price. In the case of poultry the international price is assumed to be the domestic wholesale price minus the tariff. Given that a tariff rate t=20% was imposed on poultry imports over the 1970-87 period, the respective tariff can be calculated as

PAGE 151

144 't = h -hh = h -(~) = h (-t ) ch c h c ch 1 + t c h 1 + t ( C. 2) The international price of wheat is computed as the unit value of total wheat imports (i e. ratio of total import value to total import quantity). The Japanese government administers two different resale prices for imported and domestically produced wheat. For the relevant TRI calculations, an average resale price is computed by weighing the two administrative resale prices by the volume of domestic production and imports respectively. Then the implicit consumer (user) subsidy is calculated as the difference between the average government resale price and the import unit value of wheat. Similarly, the implicit producer subsidy is calculated as the difference between the government purchase price and the import unit value. Since no significant rice imports are recorded, the export price of a major rice exporting country is used as an indicator world price In particular the international price of rice is calculated as the unit value (ratio of total export value to total export quantity) of Thailand's rice exports. The consumer (producer) subsidy is then computed as the difference between the government resale (purchase) price of rice and the international price. Furthermore, the subsidy paid to rice growers for diversion of paddy fields to other crops is implicitly calculated as the ratio of the total government outlay on riceland diversion and the resulting actual riceland reduction (in thousands of hectares). In addition, Japan Statistical Yearbook reports annually estimates of the return-to-land along with other major production costs for the principal farm production processes. Thus the

PAGE 152

145 effect of riceland diversion on the balance of trade function is evaluated by considering the difference between the reported return-to-riceland and the respective diversion subsidy. Turning to the imports of fresh oranges, the respective international price is again computed as the unit value of total orange imports. However, computation of the quota rent on fresh oranges requires some average wholesale price of imported oranges which was not available. In the absence of a better alternative, the quota rent of orange imports is computed as the difference between the landed price (import unit value + tariff) and the import unit value of total imports assuming a 40% tariff on orange imports. To the extent of course, that the wholesale price is higher than this estimated landed price, this implies an underestimation of the quota rent.

PAGE 153

146 Table C. l Japan: Beef total beef imports, domestic, and international prices. Total imports Import unit Domestic Quota rent value wholesale Quantity Value carcass pnce Year MT million Y Y/kgr Y/kgr Y/kgr 1969 18624 5678 304.89 813 508 1970 23227 8025 345.51 843 497 1971 41572 16183 389.27 880 491 1972 57609 24448 424.37 930 506 1973 127224 80103 629.63 1419 789 1974 53603 39349 734.07 1445 711 1975 44923 22371 497.98 1640 1142 1976 94234 50052 531.15 1986 1455 1977 84545 36854 435.92 2007 1571 1978 100863 46562 461.63 1938 1476 1979 131792 90139 683.95 1569 885 1980 123953 100488 810.69 1495 684 1981 123646 89021 719.97 1392 672 1982 122694 98064 799.25 1475 676 1983 137542 106198 772.11 1484 712 1984 145084 108109 745.15 1472 727 1985 150207 111042 739.26 1511 772 1986 177948 93470 525.26 1556 1031 1987 216671 li5493 533.04 1546 1013 Sources: USDA (1983), OECD (1987), OECD (1990), LIPC North American Representative Office, Denver Co, (personal communication).

PAGE 154

147 Table C.2 Japan: pork total imports, domestic and international prices. Total imports U.S. Excha Estima Domesti Variable whole nge ted C levy Quantity Value sale rate world wholesal price price e price Year MT million Y cents/ YI$ Y/kgr Y/kgr Y/kgr lb 1968 10483.7 4132.5 55.30 360 499.9 438 24 99 1969 42651.1 18623.8 62.80 360 567.7 481 28.38 1970 17148. 7 7533.5 63.40 360 573.1 397 28.66 1971 27203.9 11219.5 57.00 351 502.4 431 25.12 1972 67932.3 30506.0 71.30 303 542.5 456 27.12 1973 125739 3 63797.7 95.80 272 654.3 493 32.72 1974 41936.0 26837.9 85.50 292 626.9 559 31.34 1975 124512.7 91343.1 115.30 297 859.9 743 42.99 1976 148771.9 124213.9 105.20 297 784.6 747 39.23 1977 108110.0 89848.6 99.00 269 668.7 732 63.00 1978 103516.0 86282.0 107.70 210 567.9 691 123.00 1979 132161.2 111856.4 100.40 219 552.1 615 63.00 1980 108187.6 92168.5 98.00 227 558 6 627 68.00 1981 185747.1 158511.8 106.70 221 592.1 692 100.00 1982 141006.0 135457.7 121.80 249 761.5 656 38.08 1983 165450.8 153033.8 108.90 238 650.8 663 32.54 1984 194464.1 168380.4 110.10 238 658.0 648 32.90 1985 189121.3 165789.2 101.10 239 606.7 535 30.34 1986 206567.8 174826.2 110.90 169 470.6 509 38.00 1987 280003.3 204624.9 113.00 145 411.4 473 62.00 Sources: USDA(l983), OECD(l987), USDA(l990), LIPC North American Representative Office, Denver Co, (personal communication).

PAGE 155

148 Table C.3 Japan: Poultry total imports, domestic, and international prices. Total imports Domestic Estimated Tariff wholesale world price Quantity Value carcass price Year MT million Y Y/kgr Y/kgr Y/kgr 1968 16204.4 3969.63 248 207 41 1969 20103.3 5390.03 234 195 39 1970 10686.2 2891.29 234 195 39 1971 27161.8 6549.14 261 218 44 1972 29278.2 6899.89 236 197 39 1973 25887.8 7357.64 268 223 45 1974 25349.7 9279.42 310 258 52 1975 21540.2 8026.69 341 284 57 1976 38274.2 14009.22 353 294 59 1977 47585.2 16979.43 349 291 58 1978 61588.7 18451.22 298 248 50 1979 72285.4 24692 92 277 231 46 1980 72172.3 25420.16 299 249 50 1981 101299.1 37371.18 324 270 54 1982 105532.1 40527.59 305 254 51 1983 104401.4 36470.25 295 246 49 1984 107412.4 41625.36 290 242 48 1985 105292.0 36713.59 264 220 44 1986 180110.3 48359.72 248 207 41 1987 203754.8 50934.32 207 173 35 Sources: USDA(l983), OECD(1987), USDA(l990), LIPC North American Representative Office, Denver Co, (personal communication).

PAGE 156

149 Table C.4 Japan: wheat -total imports, domestic, and international prices. Total imports Import Exchange World price unit value rate Quantity Value Year 1000s MT $10,000 $/MT YI$ Y/kgr 1970 4684.5 31839 68 360 24.468 1971 4871.9 34711 71 351 25.008 1972 5149.3 36151 70 303 21.272 1973 5386.0 65896 122 272 33.278 1974 5376.6 120692 224 292 65.547 1975 5654.2 111709 198 297 58.678 1976 5826.9 105392 181 297 53.719 1977 5675.7 74834 132 269 35.468 1978 5564.1 83743 151 210 31.606 1979 5925.8 109043 184 219 40.299 1980 5682.3 123614 218 227 49.382 1981 5632.6 126976 225 221 49.820 1982 5713.3 111698 196 249 48.681 1983 5816.3 112693 194 238 46.113 1984 5978.3 111402 186 238 44.350 1985 5509.6 99067 180 239 42.974 1986 5619.6 99036 176 169 29.783 1987 5476.2 79347 145 145 21.010 (continued)

PAGE 157

150 Table C.4 continued. Government Government Domestic Average resale price for resale price for production resale price imported wheat domestic wheat Year Y/kgr Y/kgr 1000s MT Y/kgr 1970 34.460 32.330 474 34.264 1971 34.513 32.400 440 34.338 1972 33.900 31.580 284 33.779 1973 37.707 35.610 202 37.631 1974 45.420 43.150 232 45.326 1975 46.553 44.740 241 46.479 1976 58.800 53.630 222 58.610 1977 60 600 54.950 236 60.374 1978 60.600 54.550 367 60.226 1979 60.600 55.590 541 60 181 1980 69.145 60.780 583 68.367 1981 72.450 63.950 587 71.648 1982 73.100 64.400 742 72.100 1983 78.330 69.330 695 77.369 1984 78.217 68.917 741 77.191 1985 78.617 68.917 874 77.289 1986 79.117 68.917 876 77.741 1987 75.310 64.767 864 73.873 ( continued)

PAGE 158

151 Table C.4 continued. Government purchase User subsidy Producer subsidy price for wheat Year Y/kgr Y/kgr Y/kgr 1970 57 180 10 33 1971 61.120 9 36 1972 63 500 13 42 1973 72.420 4 39 1974 92.730 -20 27 1975 102.150 -12 43 1976 109.600 5 56 1977 158.300 25 123 1978 161.500 29 130 1979 165.400 20 125 1980 178.400 19 129 1981 184.120 22 134 1982 184.120 23 135 1983 184.867 31 139 1984 184.867 33 141 1985 182 717 34 140 1986 173.750 48 144 1987 165. 750 53 145 Sources: F.A.O. Trade Yearbook various issues, Japan Statistical Yearbook various issues, USDA (personal communication).

PAGE 159

152 Table C.5 Japan: Rice Domestic and international prices. World Government Government Producer User subsidy price purchase price resale price subsidy (Thailand export unit value) Year Y/kgr Y/kgr Y/kgr Y/kgr Y/kgr 1970 40.95 137.86 124.03 97 83 1971 30.86 142.03 122.95 111 92 1972 30.59 149.23 130.77 119 100 1973 56.49 171.68 130.10 115 74 1974 135.18 227.08 170.93 92 36 1975 91.92 259.50 203.42 168 112 1976 64.83 276.20 224.18 211 159 1977 60.39 287.20 246.18 227 186 1978 66.70 287.52 246.18 221 179 1979 59.79 287.98 256.52 228 197 1980 77.32 294.57 264.85 217 188 1981 85.22 295.93 273.18 211 188 1982 64.42 299.18 283.88 235 219 1983 60.00 304.43 283.88 244 224 1984 56 73 311.13 294.55 254 238 1985 48.82 311.13 305.45 262 257 1986 28.87 311.13 309.97 282 281 1987 28.79 292.61 309.97 264 281 Sources: F.A.O. Trade Yearbook various issues, Australian bureau of Agr. and Res. Econ. (1988).

PAGE 160

153 Table C.6 Japan: rice riceland rents and diversion subsidies. Government outlay Actual land Average diversion on riceland reduction subsidy diversion Year billion Y 1000s ha Y/ha 1970 81.8 351 233048 1971 184.0 541 340111 1972 202.9 566 358481 1973 202.7 562 360676 1974 127.9 313 408626 1975 106.1 264 401894 1976 78.7 194 405670 1977 95 6 212 450943 1978 304.5 438 695205 1979 228.1 472 483263 1980 303.4 585 518632 1981 362.2 668 542216 1982 365.2 672 543452 1983 344.7 639 539437 1984 268.3 620 432742 1985 239.1 594 402525 1986 250.1 616 406006 1987 182.6 6004 304333 (continued)

PAGE 161

154 Table C.6 continued. Paddy Paddy field Planted Rice Paddy field fields land land rent area production land rent rent Year Unit: y Y/MT 1000s ha 1000s MT Y/ha kgr 1970 150 2349 15660 2923 12689 67981 1971 150 2683 17887 2695 10887 72257 1972 150 2867 19113 2640 11889 86075 1973 150 3241 21607 2620 12144 100149 1974 60 1824 30400 2724 12292 137179 1975 60 2280 38000 2764 13165 180995 1976 60 2664 44400 2779 11772 188081 1977 60 2616 43600 2757 13095 207088 1978 60 2931 48850 2548 12589 241355 1979 60 3100 51667 2497 11958 247429 1980 60 3373 56217 2377 9751 230614 1981 60 3540 59000 2278 10259 265707 1982 60 3750 62500 2257 10270 284393 1983 60 3905 65083 2273 10366 296812 1984 60 3430 57167 2315 11878 293316 1985 60 3584 59733 2342 11662 297442 1986 60 3528 58800 2303 11647 297370 1987 60 3626 60433 2146 10627 299266 Estimate. Sources: Japan Statistical Yearbook various issues, Australian Bureau of Agr. and Res. Econ. (1988).

PAGE 162

155 Table C.7 Japan: fresh oranges, total imports. Total imports Import unit Excha Import Estimated value nge unit quota rent rate value Year MT 1000s $ $/kgr Y/kgr Y/kgr (Y/$) 1970 4313 1422 0.329701 360 119 47.48 1971 6896 2515 0.364704 351 128 51.20 1972 13479 4577 0.339565 303 103 41.16 1973 16418 6792 0.413692 272 113 45.01 1974 20437 8479 0.414885 292 121 48.46 1975 22116 11008 0.497739 297 148 59.13 1976 24401 11839 0.485185 297 144 57 64 1977 22676 11728 0.517199 269 139 55.65 1978 51493 36028 0.699668 210 147 58.77 1979 54350 44330 0.815639 219 179 71.45 1980 71814 43184 0.601331 227 137 54.60 1981 75684 64674 0.854527 221 189 75.54 1982 82658 75621 0.914866 249 228 91.12 1983 89489 63159 0.705774 238 168 67 19 1984 89231 82705 0.926864 238 221 88.24 1985 111971 92487 0.825991 239 197 78.96 1986 117661 99283 0.843806 169 143 57.04 1987 123545 121809 0.985948 145 143 57.19 Sources: F.A.O. Trade Yearbook, various issues.

PAGE 163

156 Table C.8 Japanese meat imports: beef, pork, and poultry. Quantity: 1000s MT, Value: $ 1000s. Australian beef New Zealand beef U.S. beef Year Quantity Value Quantity Value Quantity Value 1962 2.796 1455 1.746 833 a 1963 3.390 1635 1.130 549 1964 5.245 3082 0.803 434 1965 7.774 4750 2.569 1608 1966 9.345 7107 3.293 2208 1967 9.938 8797 2.393 2100 1968 10.031 9092 2.298 1884 1969 15.062 12249 3.081 2592 0.097 380 1970 20.123 18070 2.511 2521 0.362 1363 1971 36.959 40625 4.004 4111 0.507 1284 1972 52. 712 71785 3.870 4880 0.597 2100 1973 107.271 240421 9.464 19470 9.527 33150 1974 42.356 101025 2.929 7157 7.712 25827 1975 37.109 51881 3.512 6221 3.545 15824 1976 76.138 116188 4.465 8568 11.267 37793 1977 71.738 102696 3.858 8671 7.264 21328 1978 77.541 152100 7.751 16084 12.745 47209 1979 100.430 289700 3.432 11816 23.534 96514 1980 92.935 306106 3.924 16730 22.437 104877 1981 86.952 258093 6.143 22610 26.464 109833 1982 85.998 237467 3.641 12852 31.57 133990 1983 90.952 272288 7.724 25686 37.714 144471 1984 91.842 268345 7.576 23252 42.238 153035 1985 92.935 248313 6.955 22415 46.514 180111 1986 105.166 281541 6.038 18716 63.389 238930 1987 121.127 364243 7.862 27525 85.292 380939 1288 136 32) 47)762 JO 422 433)5 102 845 637146 (continued)

PAGE 164

157 Table C.8 continued. Pork Poultry Year Quantity Value Quantity Value 1962 0.284 246 1963 6.500 5486 3.471 2550 1964 4 000 3130 5.900 4099 1965 0.070 59 6.135 4602 1966 7.935 6044 1967 8.400 5912 1968 10.484 11480 16.205 11028 1969 42.651 51737 20.103 14973 1970 17.149 20928 10.687 8035 1971 27.204 32458 27.162 18947 1972 67.815 98954 29.278 22425 1973 125. 795 234275 25.888 27003 1974 42.020 92267 25.350 31827 1975 124.549 307920 21.539 27048 1976 148.905 406794 37.613 46470 1977 109.967 332022 47.585 63505 1978 103.327 410085 61.589 88570 1979 131.652 509802 72.285 113118 1980 108.158 407667 72.172 112560 1981 183.629 711445 101.298 169852 1982 141.086 540932 105.532 163246 1983 166.253 644551 104.401 153613 1984 195 611 709971 107.413 175451 1985 190.221 700422 105 292 155078 1986 207.776 1044147 180.110 288756 1987 280.565 1418692 203.754 353179 1288 322 200 1655832 270 600 416522 None or negligible. Sources: U.N. Data Summary, F.A.O. Trade Yearbook various issues.

PAGE 165

158 Table C.9 Japan: consumption levels of beef, pork, poultry, wheat, and rice. (Gross) food: 1000s MT Year Beef Pork Poultry Wheat Rice 1965 220 407 211 3700 12037 1966 163 565 269 4025 11512 1967 168 603 311 4106 11412 1968 185 600 354 4119 11188 1969 246 631 443 4168 10972 1970 292 751 515 4092 10894 1971 332 870 597 4169 10812 1972 371 953 672 4250 10877 1973 371 1099 731 4316 10941 1974 369 1140 782 4409 10950 1975 394 1165 787 4522 10878 1976 387 1205 870 4602 10761 1977 442 1279 824 4655 10487 1978 501 1387 922 4681 10367 1979 576 1604 1169 4749 10227 1980 602 1639 1194 4839 10198 1981 630 1642 1238 4808 10320 1982 679 1647 1302 4845 10837 1983 722 1678 1359 4865 10494 1984 750 1697 1425 4900 9989 1985 772 1804 1466 4920 9962 1986 815 1890 1574 4922 9859 1987 891 1994 1641 4938 9709 Sources: OECD (1981) Food Consumption Statistics 1964-1978, and OECD (1991) Food Consumption Statistics 1979-1988.

PAGE 166

REFERENCES Anderson J.E. (1991). 'The Coefficient of Trade Utilization: The Cheese Case. in R.E. Baldwin (ed) Empirical Studies of Commercial Policy NBER, Chicago: University of Chicago Press. Anderson J.E. (1992). 'Tariff Index Theory.' mimeo, Boston College. Anderson J.E. and G. Bannister (1991). The Trade Restrictiveness Index: An Application to Mexican Agriculture.' mimeo, World Bank. Anderson J.E. and J.P. Neary (1991a). 'A New Approach to Evaluating Trade Policy.' mimeo, Boston College. Anderson J.E. and J.P. Neary (1991b). 'The Coefficient of Trade Utilization: Back to the Baldwin theorem.' in Jones R.W. and A.O. Krueger (eds) The Political Economy of International Trade: Essays in Honor of Robert E. Baldwin, New York: Basil Blackwell. Anderson J.E. and J.P. Neary (1992a). 'Trade Reform with Quotas, Partial Rent Retention, and Tariffs' Econometrica, 60(1) pp. 57-76. Anderson J.E. and J.P. Neary (1992b). 'Domestic Distortions and International Trade.' mimeo, Boston College. Armington P.S. (1969). 'A Theory of Demand for Products Distinguished By Place of Origin.' IMF Staff Papers, 21, pp. 159-77. Australian Bureau of Agricultural and Resource Economics (1988). Japanese Agricultural Policies: A Time for Change, Canberra: Australian Government Publishing Service. Balassa Bela (1965). 'Tariff Protection In Industrial Countries: An Evaluation.' Journal of Political Economy, 73(6), pp. 573-94. Barber C.L. (1955). 'Canadian Tariff Policy Canadian Journal of Economics and Political Science, 21, pp. 81-7. 159

PAGE 167

160 Barten A.P. (1964). 'Consumer Demand Functions Under Conditions Of Almost Additive Preferences.' Econometrica, 32, pp. 1-38. Basevi G. (1966). 'The United States Tariff Structure: Estimates of Effective Rates of Protection of United States Industries and Industrial labor.' Review of &onomics and Statistics, 48, pp. 147-60. Bieda K. (1963). 'Trade Restrictions in the Western Community.' American Economic Review, 53(1), pp. 130-32. Bowden R.J. and D.A. Turkington (1984). Instrumental variables, New York: Cambridge University Press. Bruno M. (1972). 'Domestic Resource Cost and Effective Protection: Clarification and Synthesis.' Journal of Political Economy, 80, pp. 16-23. Chambers R.G. (1988). Applied Production Analysis: A Dual Approach, Cambridge University Press. Corden W.M. (1963). 'The Tariff' in A. Hunter (ed.) The Economics of Australian Industry. Melbourne: Melbourne University Press, pp. 174-214. Corden W.M. (1966). 'The Effective Protective Rate, the Uniform Tariff Equivalent and the Average Tariff Economic Record, 42 pp.200-16. Corden W.M. (1971). The Theory of Protection, Oxford: Clarendon Press. Corden W.M. (1985). Protection, Growth and Trade: Essays in International Economics, Oxford, England: Basil Blackwell. Crawford G.J. (1934). 'Tariff Level Indices' Economic Record, 10, pp. 213-21. Deaton A.S. (1979). The Distance Function in Consumer Behavior with Applications to Index Numbers and Optimal Taxation, Review of Economic Studies, 46, pp. 391405. Debreu G. (1951). The Coefficient of Resource Utilization, Econometrica, 19, pp. 273292. Diewert W.E. (1980). 'Aggregation Problems in the Measurement of Capital' in D. Usher (ed) The Measurement of Capital, Chicago: University of Chicago Press. Dixit A. K. and V. Norman (1980). Theory of International Trade: A Dual, General Equilibrium Approach. London: Cambridge University Press.

PAGE 168

161 Feenstra R.C. (1988). Empirical Methods for International Trade, London: MIT Press. Feenstra R.C. (1990). 'New goods and Index Numbers: U.S. Import prices.' mimeo, University of California, Davis. Feenstra R.C. and J.R. Markusen (1991). 'Accounting for Growth with New Inputs.' Working Paper #380, Dept. of Economics, University of California, Davis. Fitchett D.A. (1988). 'Agricultural Trade Protectionism in Japan: A Survey.' World Bank Discussion Papers, #28. Food and Agriculture Organization-(FAO) (1973). Agricultural Protection: Domestic Policy and International Trade, FAQ Conf. Doc. C73/LIM/9, Rome. Food and Agriculture Qrganization-(FAQ) (1975). Agricultural Protection and Stabilization Policies: A Framework of Measurement in the Context of Agricultural Adjustment, FAQ Conf. Doc. C75/LIM/2, Rome. Griliches Z. (1961). 'Hedonic Price Indexes for Automobiles: An Econometric Analysis of Quality Change' in, The Price Statistics of the Federal Government, New York: National Bureau of Economic Research. Haberler G. (1936). The Theory of International Trade With its Applications to Commercial Policy, London: William Hodge. Hansen L.P. (1982). 'Large Sample Properties of Generalized Method of Moments Estimators', Econometrica, 50(4), pp.1029-54. Hausman J.A. (1978). 'Specification Tests in Econometrics', Econometrica, 46, pp.125172. Humphrey D.D. (1962). The United States and the Common Market, New York: Praeger. Josling T.E. (1980). Developed Country Agricultural Policies and Developing Country Supplies: The Case of Wheat, Report No. 14, International Food Policy Research Institute: Washington DC. Johnson H.G. (1960). 'The Cost of Protection and the Scientific Tariff.' Journal of Political Economy, 58, pp. 327-45. Krueger 0. A. (1972). 'Evaluating Restrictionist Trade Regimes: Theory and Measurement.' Journal of Political Economy, 80, pp.48-62.

PAGE 169

1 62 League of Nations (1927). Repon on Tariff Indices 1927, (Doc. C.E.I. 37) Geneva: League of Nations. Lerdau E. (1957). 'On the Measurement of Tariffs: The U.S. Over Forty Years.' Economia Internazionale 10, pp. 232-47 Lloyd P. J. and Schweinberger A.G. (1988). Trade Expenditure Functions and the Gains from Trade, Journal of International Economics, 25, pp. 275-98. Lloyd P. J. and Schweinberger A. G. (1990). Distortions, True Trade Price and Quantity Indices and Welfare, Scandinavian Journal of Economics, 92(4), pp. 541-57. Longworth J. W. (1983). Beef in Japan, St. Lucia: University of Queenland Press, Australia. Loveday A. (1929). 'The Measurement of Tariff Levels.' Journal of the Royal Statistical Society, 92, pp. 487-516. Loveday A. (1930). 'The Australian Tariff: A Criticism' Economic Record, 6, pp. 272-8. Mori Hiroshi and W.D. Gorman (1985). 'Issues, Facts, and Opportunities for Exports of U.S Beef to Japan', Agribusiness, 1 (2) pp.211-18. Mori Hiroshi and W.D. Gorman (1986). 'U.S. Grain-fed Beef in the Japanese Market', in W.D. Gorman, G.A. Welsh and J.S. Hillman (ed) Research Opponunities in Beef Expon Markets: United States and Pacific Rim countries, Tucson, Arizona: University of Arizona. Neary J.P. (1988). Tariffs, Quotas, and Voluntary Export Restraints with and without Internationally Mobile Capital' Canadian Journal of Economics, 21, pp. 714-735. Neary J.P. (1989). 'Trade Liberalization and Shadow Prices in the Presence of Tariffs and Quotas.' Working Paper # 89/4, Centre for Economic Research, University College Dublin. Neary J.P. and K.W.S. Roberts (1980). 'The Theory of Household Behavior Under Rationing' European Economic Review, 21, pp. 25-42. Organization for Economic Cooperation and Development (OECD) (1981). Food Consumption Statistics 1964-1978, Paris: OECD. Organization for Economic Cooperation and Development (OECD) (1987). National Policies and Agricultural Trade. Country Study: Japan Paris: OECD.

PAGE 170

163 Organization for Economic Cooperation and Development (OECD) (1991). Food Consumption Statistics 1979-1988, Paris: OECD. Roningen V.O. and P. M. Dixit (1991). 'A Simple Measure for Agricultural Trade Distortion.' Working Paper #91-10, International Agricultural Trade Research Consonium (I.A. T.R. C.). Research and Policy Committee of the Committee for Economic Development (1964). 'The Height of United States and EEC Tariffs' in Trade Negotiations for Better Free World Economy: A Statement on National Policy, New York: The Committee foe Economic Development. Schwartz N.E. and S.Parker (1988). 'Measuring Government Intervention in Agriculture for the GATT Negotiations' American Journal of Agricultural Economics, 70(5) pp. 1137-45. Tangermann S., T.E. Josling and S.Pearson (1987). 'Multinational Negotiations on Farm-support Levels' World Economy, 10, pp. 265-81. Theil H. (1967). Economics and Information Theory, Amsterdam: North-Holland Publishing Company. Theil H. (1980). The System-Wise Approach to Microeconomics, Chicago: The University of Chicago Press. Towle L.W. (1956). International Trade and Commercial Policy, New York: Harper and Brothers. U.S. Department of Agriculture (USDA) (1983). 'Japan's Feed-Livestock Economy: Prospects for the 1980s' Econ. Res. Serv. FAER no. 177, Washington DC. U.S. Department of Agriculture (USDA) (1984). The 1984 U.S. -Japan Beef and Citrus Understanding: An Evaluation, Econ. Res. Serv. FAER no. 222, Washington DC. U.S. Department of Agriculture (USDA) (1987). Government Intervention in Agriculture: Measurement, Evaluation, and Implications for Trade Negotiations, Econ. Res. Serv. FAER No. 229, Washington DC. U.S. Department of Agriculture (USDA) (1988). Estimates of Producer and Consumer Subsidy Equivalents: Government Intervention in Agriculture 1982-86, Econ Res. Serv. Staff Report AGES880127, Washington DC.

PAGE 171

164 U.S. Department of Agriculture (USDA) (1989a). Agriculture in the Uruguay Round: Analysis of Government Suppon, Econ. Res. Serv. Staff Report AGES880802, Washington DC. U.S. Department of Agriculture (USDA) (1989b). 'Elasticities in the Trade Liberalization Database' Econ. Res, Serv. AGES 89-20, Washington DC. U.S. Department of Agriculture (USDA) (1990). Estimates of Producer and Consumer Subsidy Equivalents: Government Intervention in Agriculture 1982-87, Econ Res. Serv. Statistical Bulletin No 803, Washington DC. Weatherspoon D.D. and Seale J.L.Jr. (1992). 'Do the Japanese Discriminate Against Australian Beef Imports? Evidence from the Differential Approach' Paper presented at the American Agricultural Economics Association Annual meeting, Baltimore, Maryland August 9-12, 1992. White H. (1982). 'Instrumental Variables Regression With Independent Observations', Econometrica, 50(2), pp.483-500. White H. and I. Domowitz (1984). 'Nonlinear Regression With Dependent Observations' ,Econometrica, 52(1), pp.143-61.

PAGE 172

BIOGRAPHICAL SKETCH Christos Pantzios was born on October 22, 1963, in Almiros Volou, Greece. After graduating from high school, he enrolled at the Economics Department of The Graduate Industrial School of Thessaloniki, Greece, and graduated with a bachelor's degree in economics in April, 1986. In August 1987, he started postgraduate studies in agricultural economics, at the Mediterranean Agronomic Institute of Rania, Greece. He received the Postgraduate Diploma of I.C.A.M.A.S. (International Center for Advanced Mediterranean Agronomic Studies) in August 1988, and the Master of Science (M.S.) degree of I.C.A.M.A.S. in March, 1990. While finishing his M.S. program, Mr Pantzios was accepted in the doctoral program of the Food and Resource Economics (F.R.E.) Department, at the University of Florida. He begun his Ph.D. studies at the F.R.E. in August 1989, with primary field of specialization the area of agricultural production economics and productivity and secondary field the area of agricultural trade. His doctorate was awarded in December 1993. 165

PAGE 173

I certify that I have read this study and that in my opm1on 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 // / / /4 _,,H ; Timothy G. T lor Chair 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. ~~li t : ~ ~ ~ ~~~ge ~!M l ~i k 1-:< 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. a::fxt:/~ 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. Az.. ..... -;;,: J~esL Seale, Jr. Associate Professor of Food and Resource Economics

PAGE 174

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. Elias DjJl0P0111os Professor of Economics This dissertation was submitted to the Graduate Faculty of the College of Agriculture and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. December 1993 Dean, l :; of:~kre Dean, Graduate School