The trade restrictiveness index : an application to Mexican agriculture

Policy Research WORKING PAPERS International Trade International Economics Department The World Bank March 1992 WPS 874 The Trade Restrictiveness Index An Application to Mexican Agriculture James E. Anderson and Geoffrey Bannister Applying a new trade restrictiveness index to policy reforn in Mexican agriculture shows substantial trade liberalization be- tween 1987 and 1989, attributable mainly to changes in maize policy. Policy Rcsearch Working Papers disseminate the findings of work in progress and cr.courage the exciange of idcas among Bank staff and ailothers interested in development issues Thsepapers. distributed by the Research Advisory Staff,carry the names oftheauthors. rcflecct only theirvicws,and should be used and cited accordingly. lhe findings, antcrpretations, and conclusion- are theaauthors' own. Thcyshould not be attributed to the World Bank. its Board of )irectors, its management, or any of its memhbr countries.  * i: : t Policy Research International Trade- WPS 874 This paper-a productof the International Trade Division, International Economics Depai ument-is part of a larger effort in the department to contribute to the analysis of trade policies. It is funded by the Dank's Research Support Budget, "The Cost-of-Protection Index," RPO 676-49. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Grace Ilogon, room S8-038, extension 33732 (March 1992, 51 pages). To measure domestic distortions in agriculture, reveal many simultaneous increases and decreases in analysts have used producer and consumer subsidy implicit subsidies or taxes. The trade restrictiveness equivalents (PSEs and CSEs), as well as the familiar index provides a consistcnt aggregation of these trade-weighted averages of tariffs and tariff equiva- policies. From 1985 to 1987, domestic policy on the lents of quotas. All these indices lack a theoretical whole was cquivalent to an increase in trade restric- foundation. tivcness. In the next two years, trade loosened. Anderson and Bannister apply a new concept, the The net effcct ol policies in tradable agricultural trade resrictiveness index, to an evaluation of goods over the five-year period is a significant Mexican agricultural reform. They assess a significant reduction in trade restrictiveness. Restoring the trade reform episode to demonstrate the feasibility of the restrictiveness to its 1985 lcvel requires a uniform 31- method and its advantages over standard techniques. percent trade tax surcharge on 1989 prices. Moreover, Anderson and Bannister set out the theoretical the restrictiveness implied by the 1989 levcls, structure of index numbers for distorted trading compared with free trade, was equivalent to a 17 economies in earlier papers. 1They develop an index percent ad valorem trade tax. Thus, the liberalization number for trade distortirm-: the uniforn tariff, which ol the 1985-89 period carried Mexican agriculture is cquivalent in trade restrictiveness to the actual morc than halfway to free trade. differentiated structure of tariffs and quotas. To One virtue of the index is that the sources of extend the index to domestic distortions, they draw on liberalization can be dctailed. Liberalization is the well-known equivalence between a tariff and an attributable mainly to changes in maize policy, equal level of producer subsidy and consumer tax despite substantial changes in other producer and (when imported and domestically produced goods are consumer price policies. Reducing the subsidy for perfect substitutes). fertilizer use was reiatively unimportanL The trade restrictiveness index for domestic The standard PSE and CSE index methods are distortions is defined as the uniform tariff equivalent not directly comparable to the trade restrictiveness of the consumption and production distortions. It is, index, as they do not aggregate consumer and in turn, a combination of two subindices: the consis- producer distortions. The PSE and CSE indices are, tent producer subsidy equivalent (CPSE) and the however, comparable to the consistent subindices consistent consumer subsidy equivalent (CCSE). CPSE and CCSE. The rates of change of these two These are defined as the uniform subsidy rates that typcs of indices arc only weakly positively associated, are equivalent in trade restrictiveness to the actual diffcr in sign in a quarter of the cases, and in most differentiated subsidy or tax structure. They are cases differ widely in magnitude. counterparts to the PSE and CSE. The differencc The implications of the consistent index of the between the consistent and conventional subindices is change in consumer policy are diametrically opposed in the method of aggregation. Consistent aggregation to the implications of the CSE over the five-year is based on the use of "marginal welfare weights" as period. Using the trade restrictiveness index thus opposed to production and consumption share makes a great practical and a theoretical diffcrence. weights. In Mexico, from 1985 to 1989, the target producer and consumer price policies for major crops ThePolicy RescarchWorking PaperSericsdisseminatcs the findings of work tnder way in tdce ank. Anobjectiveof the series is to get these findings out quickly, even if presentations are less than fully polished. The findings, intcrpretations. and conclusions in these papers do not necessarily represent official Bank policy. Produced by the Policy Research Dissendiiation Center  The Trade Restrictiveness Index: An Application to Mexican Agriculture James E. Anderson* Geoffrey Bannister** Table of Contents L The Trade Restrictiveness Index .............................. 3................... 1.1 The Consumer Price Index .4 L2 The Trade Restrictiveness Index .5 1.3 Domestic Distortions .8 L4 Relation to Standard Methods .10 I.5 Nontraded Goods .12 H. Mexican Agrkcultural Dist-tions .........................................................................13 1I.1 Institutional Descrij Jion ............................ 15 1.2 Data and Elasticities .............................. 20 II. The Trade Restrictiveness of Mexican Agricultural Policy 23 1M1.1 Maize Policy .25 111.2 TIhe Combined TRI ..27 11.3 The Nontraded Goods TRI .30 IV. Perspectives on the Results ...... ............. ...... ...... 31 IV.1 Sensitivity to Elasticities .31 IV.2 Comparison with Other Indices .32 V. Conclusion ............................................................34 References .......................... ............36 Appendix 1.Welfare Cost Accounting in the Mexican Application...................37 Appendix 2: Data ........... . . 41 *Boston College, and ** the University of Texas at Austin. We are grateful to Peter Neary, Refik Erzan, Kala Krishna and D.C. Rao for helpful comments. We also wish to thank Jorge Garcia Garcia for useful discussions and data on Mexican agriculture, as well as Constrnza Valdez and Andres Casco Flores.  1 The Trade Restrictiveness Index: An Application to Mexican Agriculture Agricultural markets in Mexico, as elsewhere, are distorted by domestic subsidies as well as intemationat trade policies. To measure domestic distortions in agriculture and compare them internationaly, analysts have used 'producer subsidy equivalent' (a production-share weighted average of producei subsidies) and 'consumer subsidy equivalent' indices, as well as the familiar trade-weighted averages of tariffs and tariff equivalents of quotas. AU these indices lack a theoretical foundation, as is well-known. Moreover, in the absence of a connection between the three partial indices of consymption, production, and trade distortion, inferences about the trade restrictiveness of all three policies are illegitimate. This paper applies a new con-ept, the trade restrictiveness index MI), to the evaluation of Mexican agricultural reform from 1985 to 1989. It demonstrates the feasibility of the new method and its theoretical and practical advantages over standard techniques, while at the samne time providing an assessment of a particularly significant reform episode. The theoretical structure of index numbers for distorted trading economies is set out in two papers by Anderson and Neary (1991a,b). The basic idea is to form an index number equal to the uniform tariff factor which is equivalent in trade restrictiveness to the differentiated structure. The formerpaper (1991a) defnes the trade restrictiveness index for the case where all distortions are trade taxes or quotas.1 T he latter paper (1991b) extends the index to the case where domestic distortions break the equality between consumer and producer prices. It draws on the well-known equivalence between a tariff on the one hand and an equal rate of production subsidy and consumption tax on the other (for an imported good which is a perfect substitute for a domestically produced good). The trade IJIn our earlier work with quotas, Anderson and Neary (1990), we used the term "coefficient of trade utilization" reflecting the relationship of the index to the work of Debren (1951).  2 restrictiveness index in this case is the uniform tariff factor which is equivalent in trade restrictiveness to the consumption and production distortions. The trade restrictiveness index is in turn a combination of two subindices, the 'consistent producer subsidy equivalent' (CPSE) and the 'consistent consumer subsidy equivalent' (CCSE). These we define as the uniform subsidy rates which yield equivalent trade restrctiveness to the actual differentiated subsidy or tax structure. Thev are counterparts to the producer and consumer subsidy equivalents, PSE and CSE. The difference between our subindices and the conventional ones is in the method of aggregation. Our consistent aggregation is based on the use of 'marginal welfare weights' as opposed to production and consumption share weights. The trade restrictiveness index is comparable across countries and time in the same way (and with the same qualifications) that inflation rates or real growth rates are internationaUy or intertemporally comparabte. By reducing many ditortions to a single uniform trade tax equivalent, analysts can legitimately compare the international t implications of the domestic agricultural policies of a set of countries. Thus it achieves the goal of tlhe PSE and CSE measures as they have been used in GAIT negotiations. The application of the TRI to Mexican agriculture reveals the usefulness of our method. From 1985 to 1989, the target producer and consumer price policies for major crops reveal many simultaneous increases and decreases in the implicit subsidies or taxes. The trade restrictiveness index is a consistent aggregator of these policies. From 1985 to 1987, the changes in domestic policies were equivalent to an increase in trade restrictiveness, followed by a decrase in trade restrictiveness in the next two years. The net effect of policies in tradable agricultural goods over the 5 year period is a significant reduction in trade restrictiveness. Restoring the trade restrictiveness to its 1985 level requires a 31 pr cent uniform trade tax surcharge on the 1989 prices. Moreover, the restrictiveness implied by the 1989 levels compared to free trade is equivalent to a 17 per cent uniform ad valorem trade tax. Thus the liberalization of the 1985-89 period carried the Mexican agricultural  3 sector about two thirds of the way to free trade. Another virtue of the index is that the sources of this development are teasily detailed. It is mainly accounted for by maize policy, despite substantial changes in other producer and consumer price policies. The reduction in the subsidy to fertilizer use, in particular, accounted for very little. Our method provides results which differ widely from those of the standard PSE and CSE index methods. The PSE and CSE indices are comparable to the consistent subindices of the TRI, the CPSE and CCSE. The rates of change of the subsidy equivalents and their consistent counterparts are only weakly positively correlated, differ in sign in one quarter of the cases, and differ widely in magnitude in most cases. Dramatically, the consistent index of the change in consumer policy gives implications diametrically opposite to the CSE over the 5 year period. Use of the TRI thus makes a great practical as well as theoretical difference. The outline of the paper is as follows. Section I defines the trade restrictiveness index for an economy with domestic distortions, following Anderson and Neary (1991b). Section I also defines the consistent versions of the PSE and CSE, the CPSE and CCSE, and compares the latter with the former. The trade restrictiveness index (I) is shown to be a weighted average of the CPSE and CCSE. Section II sets out a description of the Mexican agricultural sector and the data used for this study. Section m presents our results covering the period from 1985 to 1989. Section IV provides perspective on the results. We compare our results with those generated by standard methods, and we test the sensitivity of our results to elasticity parameters and to measurement problems. Section V concludes. I. The Trade Restrictiveness Index Changes in index numbers are generally weighted averages of changes in the comnponents of the index. To be consistent with economic theory, the weights must arise  4 from a fundamental economic structure. It is helpful to begin with reviewing the consumer price index, or CPI where the weights are familiar. Subsection I.1 derives the consumer price index based on the consumers expenditure ,unction. In Subsection 1.2, the trade restrictiveness index, or TRI, is derived based on the economy's trade balance function. The latter is defined in Anderson and Neary. Here, the weights are less familiar, but the same logic girds the construction of the index. Subsection 1.3 extends the analysis to the case of domestic distortions. Subsection 1.4 relates the TRI to the PSE and CSE. Subsection 1.5 extends the index to cover the trade restrictiveness effect of distortions in purely nontraded goods. 1.1 The Consumer Price Index A set of consumer prices changes. The consumer price index (CPI) in rates of change measures the uniforn rate of change in all prices which produces an equivalent rise in the expenditure required to maintain welfanr. -he formal basis for the CPI is the consumer's expenditure function, e(q,u), where q is the vector of prices and u is the reference level of utility. e is the minimum level of income required to achieve u when the consumer faces q. The CPI is derived as follows. The derivative of e with respect to q is equal to the vector of the consumer's demands, X. The effect of an arbitrary set of price clNtnges on the level of income required to support u is then X'dq. The effect of a wtUbform proportonal set of changes is X'qOdoc, where a is the (scalar) proportionality factor (q = aq