The Uruguay Round and Welfare in Some Distorted Agricultural Economies

There is widespread concern about the effect of the Uruguay Round policy changes on world agricultural prices and consequently upon the welfare of developing countries. Assessing welfare changes with the standard terms of trade effect calculation can be misleading for distorted economies, since the distortion effect operates in addition or in opposition to the terms of trade effect. This study reveals distortion effects which are many times larger than terms of trade effects in a study of the Uruguay Round’s impact on 9 agricultural economies. In 3 of 9 cases, the distortion effect reverses the impact of the terms of trade effect. In 2 other cases the distortion effect raises a trivial terms of trade effect up to around 1% of national income. JEL Classification : F13, F14, O13, O20 This research was supported by the World Bank, but the usual disclaimer applies. Will Martin's help and the able research assistance of Can Erbil are gratefully acknowledged. Errors and omissions are solely the author's responsibility. The Uruguay Round agreements will reduce agricultural price distortions among major supplying nations. The resulting world price changes are predicted to include increases in basic food prices mixed with price decreases for some important exports such as coffee and cotton (Goldin and van der Mensbrugghe, 1995). There is widespread concern about the effect of these world agricultural price changes upon the welfare of developing countries. It appears that raising food prices paid by food importers must be bad for them. The fall in coffee and cotton prices raises a complementary concern as it appears that exporters of these products in developing countries must be hurt. According to the standard terms of trade effect assessment of welfare changes, each 1% rise in import (export) price adds to the compensation required to maintain real income constant at a rate equal to the import (export) share of base expenditure (income). For example, a 10% rise in the price of an import taking up 10% of national expenditure will cause a compensation requirement equal to 1% of national income. If this is not met, there is a 1% fall in real income. Based on this reasoning and ‘back of the envelope’ calculations, most analysts have concluded that the welfare effects of the forecast price changes will be quite small (e.g., Goldin and van der Mensbrugghe, 1995), even for developing countries. Assessing welfare changes with the standard terms of trade effect calculation can be misleading for distorted economies, as is well known in theory. See Tyers and Falvey (1989) and Alston and Martin (1995) for theoretical analysis of the effect of terms of trade changes when exports are subsidized. Related to this is the theory of the second-best point that liberalizations of some distortions need not be welfare-increasing due to the January 30, 1997 Price Changes and Welfare 2 presence of cross effects with other distortions. This qualification has been shown to have practical bite, as simulation results by Loo and Tower (1990) demonstrate that the benefit of agricultural trade liberalization in the presence of domestic distortions can be substantially offset by cross effects. See also Clarete and Whalley (1988) for a similar exercise with Philippine data. The theoretical qualification is practically important since agricultural exporting countries often have distorted prices in favor of producers, and a fall in agricultural prices can decrease an activity which is too large (its marginal social benefit is below its marginal social cost). Similarly, many food importers have distorted prices in favor of consumers, so that imports are in effect subsidized. In this situation, a rise in food prices causes a reduction of an activity (food importing) which is too large (its marginal social cost is above the willingness-to-pay of consumers). This distortion effect operates in addition or in opposition (as in the examples) to the terms of trade effect of the standard analysis. If in opposition, the offset can in principle be large enough to make the rise in agricultural prices actually desirable for importers, and actually undesirable for exporters. If in addition, the distortion effect can make a small terms of trade effect a big effect in total. The closest antecedent to this paper is Anderson and Tyers (1993). They employ a partial equilibrium model (of 7 food markets for 30 agricultural trading countries) to simulate trade liberalization in a dynamic stochastic setting. Using international price comparisons, they construct domestic distortions of producer prices. In their base simulation, developing countries 1 They model a 'representative' developing country with a single agricultural product and a somewhat more elaborate production structure. The benefits of an agricultural export tax cut depend on the details of assumed import tariff levels, quota rent-seeking and of excise taxation, due to the cross effect of the export tax distortion with the other distortions. January 30, 1997 Price Changes and Welfare 3 lose from the trade liberalization of developed countries, as in the standard story. In their alternative simulation, they parametrically adjust downward the domestic distortion wedge for a subset of the developing countries. With this adjustment, trade liberalization is welfare improving for some of these countries. In contrast to Anderson and Tyers, this paper applies general equilibrium methods to a more detailed set of commodities and a more detailed set of domestic distortions and trade distortions. The results show the practical significance of distortion effects in an analysis of the Uruguay Round’s impact on a group of 9 agricultural developing nations. The distortion effects are generally much larger than the terms of trade effects, sometimes with the same sign and sometimes with opposite sign. In several cases the distortion effect is large enough to push the money metric welfare effect up from a trivial fraction to nearly one percent of national income. Dramatically, in 3 of the 9 cases the distortion effect offsets the terms of trade effect by enough to reverse the direction of welfare impacts. Overall, the correlation of the money metric welfare effects and the standard terms of trade effects across the 9 countries is equal to .63. The calculations analyze the welfare effect of external agricultural price changes for 15 commodities in 9 agricultural trading countries with a variety of distortions of trade and of domestic production and consumption through subsidy and tax policies. The countries are chosen on the basis of availability of detailed trade distortions data and of detailed agricultural production and distortion data. All these countries are net exporters of agricultural products though some of them are importers of key foods. The commodities are chosen 2 The discount on the distortion wedge is 15% for one group and 30% for another group of developing countries. The adjustments are based on Krueger et al. (1988), who calculate a net disincentive to agriculture of policies which protect nonagricultural activities for a set of developing countries. January 30, 1997 Price Changes and Welfare 4 to fit the structure of the RUNS model (Goldin and van der Mensbrugghe, 1995), which is the source of the forecast price changes for the 15 commodities. Sensitivity analysis is conducted with other price change estimates. The impact of simultaneous removal of domestic agricultural distortions and the external price changes is also calculated for the subset of countries for which the domestic distortions data is available. This simulation combined with the first again reveals that distortion effects arising through domestic distortions can be larger than terms of trade effects. The welfare measurement is done using a very simple Computable General Equilibrium (CGE) model designed to focus on the details of terms of trade changes and trade distortions while suppressing much detail of the production structure. The suppression is for convenience, to allow analysis where detailed production structure information is missing. Related work with this type of CGE model (see Anderson and Neary, 1994, for a summary) shows that the details of the trade and domestic distortion structure are very important to the type of inference which may be drawn. Sensitivity analysis reveals that all qualitative conclusions are robust to variations in assumed values of world price changes, to possible errors in domestic distortion measures and to elasticity values. However, the numerical results for individual countries are sensitive to domestic distortion levels and to some extent to variations in assumed world price changes. The results depend more fundamentally on aspects of model structure which are difficult to probe for sensitivity. Besides the limitations of the simple production structure of the model, the analysis is conducted with a representative consumer/producer. This assumption ignores the distributional concerns which motivate much agricultural policy in all January 30, 1997 Price Changes and Welfare 5 countries. As usual, the welfare results may be taken as indicating a potential surplus which could in principle be used to achieve distributional goals. The results taken together emphasize that reliable conclusions about the welfare effects of world agricultural price changes must be based on general equilibrium models incorporating the effects of both trade distortions and domestic agricultural distortions. Due to data limitations, only a small group of countries is analyzed. In future work, effort should be devoted to developing the detailed trade distortions and domestic distortions data necessary to apply the method used here to other poor net agricultural importing countries. Section I reviews the basic theory of terms of trade changes and welfare. Section II is a technical description of the computable general equilibrium model used to simulate the welfare changes due to world price changes. Section III presents the results. 1. Agricultural Price Changes and Welfare Changes in agricultural prices affect welfare through the usual terms of trade effect and through several channels of distortions. In subsection A, the welfare effect of world price changes on a small country with tariffs is analyzed. In subsection B, the analysis is extended to allow for domestic distortions and for quotas. 1.1 World Price Changes and Welfare with Tariffs It is well known from the theory of the second best that in the presence of distortions, perverse results are possible. Actual demonstrations of perverse results are, however, rather rare. External rises in the price of imported food when imports (consumption) is subsidized provide an interesting example. January 30, 1997 Price Changes and Welfare 6 To see the technical essentials, suppose that a vector of traded goods are purchased or sold at external price vector p* and trade domestically at a price vector p, the difference p-p* being equal to a specific tax/subsidy vector. The balance of trade function for the representative agent economy is equal to b(p,p*,u) = e(p,u) (p-p*)'ep(p,u). Here, e is the trade expenditure function and ep is the trade vector, by Shephard's Lemma. It is also convenient to denote the trade vector as m, with positive elements of m for imports and negative elements of m for exports. Here and below, a subscript denotes differentiation, save when index variables i, j or s are used. Given p* and a fixed policy p-p*, the equilibrium level of utility u is determined by the balanced trade requirement b(.) = 0. Now consider the effect of a rise in the external price vector of traded goods p*. The policy vector p-p* is assumed to be unchanged, so the rise in p* passes fully through to p and the vector of changes in foreign exchange required to support u with the new price vector p* is equal to bp* = m' (p-p*)'mp. Using the symmetry of cross effects, the ith element of the row vector b' p* is equal to bi = mi (p-p*)'∂mi/∂p. Based on this formula, a local approximation to the welfare change of a set of price changes expressed as a proportion of national income y is: 1 y b p m p y p p m y p i i i i i i i i j j j j ji i ˆ ˆ { } ˆ , = + ∑ ∑ ∑ τ ε where τj denotes the ad valorem tax on good j, (pj p*j)/pj and ε ji denotes the elasticity of compensated import demand j with respect to price i. The equation says that the percentage change in real income due to the external January 30, 1997 Price Changes and Welfare 7 price changes is equal to the terms of trade effect, the first term, plus the distortion effect, the second term. The terms of trade effect is equal to the sum of the trade shares times the percentage changes in external prices. It is readily calculated. The second term is the familiar dead weight loss term when the p̂’s are due to tax changes. More intuition can be drawn out by focusing on a special case. For the Constant Elasticity of Substitution (CES) import demand system (identified with CES final demand and Ricardian production for example) the elasticity ε ij is equal to σ(-δij + sj), where σ is the elasticity of substitution, δij is the Kronecker delta and sj is the expenditure share of good j. The marginal welfare cost of a rise in external price i in the CES setup reduces to mi[1 + στi σΣτjsj]. If good i is subsidized while trade on average is taxed (Στjsj is positive), the square bracket term is less than one, the cost of a rise in the external price is lower than in the absence of distortions. If τi is large in absolute value relative to 'average' taxes, or if σ is large, a rise in external food prices has a negative welfare cost; i.e., it is welfare improving. Plausible parameter values can produce the 'perverse' result. (τi is equal to (pi-p*i)/pi. A 40% subsidy defined by p=0.6p* implies a value of τi equal to 0.67, and an elasticity in excess of 1.5 suffices for the 'perverse' result, even if the 'average' tax is equal to zero.) This marginal analysis is informative in showing possibilities. It is also instructive because it shows that the perversity is due to the inefficient given policy. However, it can be misleading because most general equilibrium production structures with imported inputs necessarily have some complementarity in their excess demand systems (Lopez and Panagariya, 1992) rather than the universal substitutability of the example. The CGE January 30, 1997 Price Changes and Welfare 8 model used in this paper exhibits complementarity between exports and imported inputs. 1.2 Domestic Distortions and Quotas More realistic complications of general equilibrium complicate the analysis in an interesting and practically important way. First, agriculture has significant consumption and production distortions in addition to trade distortions. See Anderson, Bannister and Neary (1995) for a discussion of related issues. Subsidized consumption acts in part like subsidized imports (or taxed exports) while subsidized production acts in part like taxed imports (or subsidized exports). Second, the presence of quotas leads to cross effects in which changes in external prices have effects on the amount of rent lost to foreigners. These can lead to ‘perverse’ results. As for nontraded nonagricultural goods distortions, while undoubtedly important, they are omitted from the model. 2. Description of the Model The model is fully Walrasian: all nontraded goods (and all quotaconstrained) markets clear with flexible prices. The closure rule is that net external borrowing must be no more than in the base year of the model. This is appropriate for a medium term model of a credit constrained economy, but it obviously ignores potentially important intertemporal linkages. (Investment expenditure is subsumed into ‘expenditure’, which shuts down potentially powerful linkages from policy changes to investment plans to changes in levels of distorted activities.) The simple CGE model used in this study has a CES/CET production structure and a CES final demand structure. Imported inputs and a single nonproduced and nontraded domestic input are used to produce domestic January 30, 1997 Price Changes and Welfare 9 'activity'. Domestic activity is jointly allocated to a composite nontraded good and agricultural production. The agricultural production is in turn divided into a set of home (nontraded) agricultural goods and exports. The elasticity of transformation between home and exported agricultural goods is thus allowed to be higher than that between agricultural goods and nonagricultural goods. The nontraded goods (composite and agricultural) are produced and consumed domestically. For the purpose of this study, all agricultural imports are treated as final goods. On the demand side, CES preferences determine expenditure between an agricultural aggregate, the nontraded composite good, and other final imports. Within agriculture, a CES aggregator divides expenditure among the home and imported agricultural goods. Thus, the elasticity of substitution in consumption between domestic and imported food can be greater than the elasticity of substitution between food and any other type of good. The chief limitation of this nested CES/CET setup is that agricultural production and nonagricultural production are assumed to have identical joint production