Does the Central American common market benefit its members?
The members of the Central American Common Market (CACM) have recently taken measures to revitalize their union: they have formally pledged to improve regional economic integration and they have reinforced the common external tariff. they took these measures in the belief that a stronger union would improve their economic welfare, a belief that has also led both the United States and the European Community to support a stronger union.
The idea that a regional trading group would improve economic welfare of the Central American countries originated in the early 1950s, when Raul Prebisch and other economists in the United Nations Economic Commission for Latin America (ECLA) argued persuasively that such a union would help these countries develop their economies. To many, the views of the ECLA economists were vindicated when the inceptionof the CACM in 1960 was followed shortly by accelerated economic growth in the region. Further support for their views came from quantitative studies (Cline and Delgado , Willmore , Yotopoulos and Nugent , Nugent , and Wilford  which all concluded that the union provides static welfare gains for its members.
There is, however, room for skepticism about the benefits of the CACM. Much of the argument of the ECLA economists is based on the notion that import substitution is a good way to promote economic development, a notion that has since fallen into disrepute. Nor does the accelerated growth that followed the union's formation provide strong evidence that the CACM provides benefits. The volume of trade between members is too small for the union to have had more than a minor effect on the region's economy; so even if this effect were negative, the region would presumably still have grown rapidly in the mid 1960s.
The evidence from the quantitative studies is also faulty. These studies did not estimate directly the effects of the CACM, but merely imputed these effects by assuming either that market shares or income elasticities would have remained constant without the union. Furthermore, they assumed that supplies of different countries are perfect substitutes for each other, whereas recent studies by Collier , Grossman , and Anderson  have shown that the degree of substitutability in demand between these supplies is important in determining how customs unions and other discriminatory tariff changes affect welfare. It is particularly important to allow for less than perfect substitutability when analyzing the CACM, because member imports from developed countries often have no close substitutes among locally produced goods.
In this paper the static welfare consequences of the CACM are reexamined, with two important advances over the previous studies. First, direct estimates are made of the effects of the union. Second, the shortcomings of the perfect substitutes assumption are overcome by allowing imports from different countries to be imperfect substitutes for each other and for domestic output. The model, which is partial equilibrium in nature, also enjoys two advantages over the computable general equilibrium models such as those Whalley  and Hamilton and Whalley  used to examine the effects of various economic integration schemes: it requires fewer data and its operation is more transparent, so it shows more clearly how introducing imperfect substitutes alters the traditional analysis of a customs union. The model's main limitation is that it applies only to customs unions among small countries.
II. THE MODEL
Figures 1, 2, and 3 show the home market and Figure 4 the export market for country A, a representative member of the CACM. A's residents distinguish between four goods: imports from other members, imports from nonmembers domestic output for home consumption, and exports. Therefore, A's home market is partitioned into three submarkets: a market for imports from nonmembers (Figure 1), a market for imports from other members (Figure 2), and a market for domestic output (Figure 3). It is assumed (reasonably) that no individual member nor the union as a whole is large enough to affect world prices, so the supply curves in Figures 1 and 2 are horizontal. Hick's compensating variation is used to measure welfare changes of A's residents, so all A's demand and supply curves, and the shifts in these curves, are income compensated.
What are the welfare consequences if the members abolish their customs union by imposing external tariffs on imports from each other? This is the way Honduras withdrew from the union in 1970 and it appears more likely that the members would terminate their union in this manner rather than by eliminating all external tariffs.
In order to avoid confounding the effects of the tariff increase with those of an increase in total taxes among the members, it is assumed that the tariff increase is accompanied by a matching reduction in other taxes that leaves total government revenue unchanged. This assumption means that A's residents are automatically compensated for most of the effect of the tariff increase on their real incomes, so the compensated demand and supply curves provide good approximations to the actual changes in prices and quantities. This assumption does not affect the welfare calculations, because these are properly done using A's compensated demand and supply curves regardless of whether compensation actually takes place. It simplifies the presentation, however, by eliminating the need to distinguish between the compensated and actual changes.
Begin in Figure 2. When A imposes the external tariff on imports from other members, the tariff-inclusive import supply curve shifts upward from S.sub.b to S'.sub.b., the quantity demanded falls from Q.sub.b to Q*.sub.b., and the price rises from P.sub.b to (1 + t)P.sub.b.. This price increase also affects the markets depicted in Figures 1 and 3; it shifts the demand for imports from nonmembers upward from D.sub.w to D*.sub.w in Figure 1 and it shifts the demand for the domestic substitute upward from D.sub.a to D'.sub.a in Figure 3. In Figure 1, the supply remains at the tariff-ridden level S'.sub.w and the price remains at the world price plus the tariff, (1 + t)P.sub.w.. In Figure 3, the price of domestic output for home consumption rises from P.sub.a to P'.sub.a.. This price rise shifts the demand in Figure 2 upward from D.sub.b to D'.sub.b and it shifts the demand in Figure 1 upward farther from D*.sub.w to D'.sub.w.. The new equilibrium demand curves are D'.sub.w., D'.sub.b and D'.sub.a., and the new equilibrium quantities are Q'.sub.w., Q'.sub.b and Q'.sub.a..
The net welfare effect of imposing the external tariff against other members is rectangle abcd in Figure 1 minus triangle abc in Figure 2. This is the gain in tariff revenue in Figures 1 and 2 minus the loss in consumers' surplus in Figure 2. The revenue gain in Figure 1 is a welfare gain to A's residents, because the social cost of their imports from nonmembers is less than the private cost, by the amount of the tariff. there are no net welfare gain or loss in the home market for domestic output in Figure 3, because the gain in producers' surplus is exactly matched by the loss in consumers' surplus. There is a net welfare gain for A if the decline in the value of imports from members is less than twice as great as the increase in the value of imports from nonmembers, causes imports from nonmembers to increase and the less it causes imports from members to decline.
Figure 4 shows the effect on A's exports. There, D.sub.xw is the nonmember demand for A's exports. This curve is drawn as a horizontal line because A's export supply is small compared with nonmember markets. D.sub.xb is the demand for these exports by other members before the customs union is eliminated and S.sub.x is A's export supply. A's exports to other members are Q.sub.xb and A's total exports are Q.sub.xt.. Eliminating the customs union shifts the demand of other members inward from D.sub.xb to D'.sub.xb., causing A's exports to them to fall to Q'.sub.xb.. A's total exports and the price of these exports are unaffected. They would be affected only if A's exports go exclusively to other members before the union is abolished and command a price above the world price as a result of the preferential treatment from other members. Although this possibility seems remote for any product category in which exports of a member are substantial, it was nevertheless investigated. Member exports are apparently limited to other members in only a few product categories, in which the volume of exports is insignificant.
Allowing for imperfect substitutes adds another dimension to the traditional analysis of customs unions. In the traditional perfect substitutes models, the amounts of trade creation and trade diversion depend on the elasticities of the competing supplies, the size of the tariff change, and the elasticity of the domestic demand. With imperfect substitutes, trade creation and trade diversion depend on all these factors plus the degree of substitutability in demand between the various supplies. For example, if the substitutability between domestic output and supplies of other members increases relative to the substitutability between supplies of other members and nonmembers, trade diversion will increase relative to trade creation. This is clear from the diagrams. With an increase in substitutability between domestic output and supplies of other members, the outward shifts in demand in Figures 3 and 2 become greater. With a decrease in the substitutability between supplies of members and nonmembers, the outward shift in demand for imports from nonmembers in Figure 1 becomes smaller.
The introduction of imperfect substitutes also allws the model to incorporate some events that are impossible to include in the perfect substitutes model. For example, even if all import supply curves are horizontal, a member might import the same kind of good from both members and nonmembers simultaneously, both with and without the internal preferences, as shown in the diagrams. Also, it is possible for a member's total imports to rise when the internal preferences are eliminated.
This section provides the equations needed to calculate the welfare effects described in the previous section. Equations (1) through (3) describe the effects of the internal tariff on the quantity demanded in Figures 1, 2, and 3; and equation (4) describes the effect on the quantity supplied in Figure 3. Q'.sub.w./Q.sub.w = (P'.sub.w./P.sub.w.).sup.[xi]ww.(P'.sub.b./P.sub.b.).sup.[xi]wb.(P'. sub.a./P.sub.a.).sup.[xi]wa., (1) Q'.sub.b./Q.sub.b = (P'.sub.w./P.sub.w.).sup.[xi]bw.(P'.sub.b./P.sub.b.).sup.[xi]bb.(P'. sub.a./P.sub.a.).sup.[xi]ba., (2) Q'.sub.A./Q.sub.A = (P'.sub.w./P.sub.w.).sup.[xi]aw.(P'.sub.b./P.sub.b.).sup.[xi]ab.(P'. sub.a./P.sub.a.).sup.[xi]aa., (3) and Q'.sub.a./Q.sub.A = (P'.sub.a./P.sub.a.)e.sub.a., (4) where a prime symbol (') following a variable denotes its equilibrium value after the tariff change, [xi].sub.ij is the elasticity of the income-compensated demand for output of supplier i with respect to the price of the substitute from supplier j, and e.sub.a is the elasticity of the compensated domestic supply. By assumption, the import supplies from both members and nonmembers are perfectly elastic.
The net welfare gain from imposing the tariff on member imports, keeping the tariff on imports from nonmembers constant (rectangle abcd in Figure 3 minus triangle abc in Figure 2), is given as dW = tP.sub.w.(Q'.sub.w - Q.sub.w.) - (1/2)tP.sub.b.(Q'.sub.b - Q.sub.b.). (5)
Noting that P'.sub.b./P.sub.b = (1 + t) and P'.sub.w./P.sub.w = 1 for this case, and substituting from equations (1) through (4) into (5) yields (after some tedious algebra) dW = tP.sub.w.Q.sub.w.[(1 + t).sup.[xi]wb + .sup.[xi]wa[xi]ab./(e.sub.a - [xi].sub.aa.[ - 1] - (1/2)tP.sub.b.Q.sub.b.[(1 + t).sup.[xi]bb + [xi]ba[xi]ab./(e.sub.a - [xi].sub.aa.) - 1]. (6)
III. APPLYING THE MODEL
Application of equation (6) requires estimates of the five compensated demand elasticities ([xi].sub.bb., [xi].sub.ba., [xi].sub.ab., [xi].sub.wb., and [xi].sub.wa and the excess supply elasticity for the competing domestic good (e.sub.a - [xi].sub.aa.) for each CACM member. Since data on domestic shipments generally are not available for the Central American countries, it is not possible to estimate the supply or demand curves for domestic output.
Therefore, in what follows, ordinary demand curves are estimated, from which are calculated four of the needed compensated demand elasticities ([xi].sub.bb., [xi].sub.ba., [xi].sub.wa and [xi].subwb.) for each member; then corresponding values for [xi].sub.ab are imputed. The welfare effects given by (6) are calculated using these demand elasticities and a range of values for the excess supply elasticity, e.sub.a - [xi].sub.aa..
Estimating the Parameters
Two import demand equations are estimated for each member country; one is for the member's imports from other members and the other is for member imports from nonmembers. Two data constraints dictate the form of the regressions. First, data for domestic prices are not available for disaggregate commodity classifications, so only aggregate demand elasticities can be estimated. Second, the sample is restricted to the period between 1971 and 1979 to avoid distortions caused by armed conflicts in Central America. This limits the number of annual observations and means that it is necessary to pool across member countries to obtain enough degrees of freedom.
In the regression explaining imports from other members, Chow tests showed that the price elasticities and the income elasticities do not vary significantly across members. Therefore, the following specification was used for this import demand. 1n(Q.sub.b.) = [alpha] + [eta].sub.bb.ln(P.sub.b.) + [eta].sub.bw.ln(P.sub.w.) + [eta].sub.ba.ln(P.sub.a.) + [mu].sub.b.ln(Y.sub.a.) + [alpha].sub.E.DE + [alpha].sub.G.DG + [alpha].sub.N.DN + [alpha.sub.HTDTH + U.sub.b., (7) where [alpha] is the constant of the regression; Y.sub.a is real income in the importing country; DE, DG, DH and DN are dummy variables for four of the five CACM members (El Salvador, Guatemala, Honduras, and Nicaragua); DTH is an additional dummy for Honduras to cover its imports during 1971 to 1973, a transition period following the border war with El Salvador; and U.sub.b is an error term. The [eta].]s denote uncompensated price elasticities and [mu] denotes the income elasticity.
In the regression explaining member imports from nonmembers, Chow tests indicated that only the own-price elasticity ([eta].sub.ww.) and the cross elasticity with respect to the price of imports from members ([eta].sub.wb.) were the same for all members. The cross elasticity with respect to the domestic price ([eta].sub.wa.) and the income elasticity ([mu].sub.w.) were found to vary significantly among members. Therefore, the coefficients of the following regression were used. 1n(Q.sub.w.) = [beta] + [eta].sub.wb.ln(P.sub.) + [eta].sub.ww.ln(P.sub.w.) + [sigma].sub.wai.D.sub.i.ln(P.sub.a.) + [sigma].sub.i.[mu].sub.wi.D.sub.i.ln(Y.sub.a.) + beta].sub.E.DE + [beta].sub.G.DG + [beta].sub.H.DH + [beta].sub.NDN + [beta].sub.HT.DHT + U.sub.w., (8) where i denotes individual members of the CACM, [beta] is the constant of the regression, D.sub.i is a dummy variable that equal one for the ith country and zero otherwise, and U.sub.w is an error term. It was not possible to pool estimates for the income and domestic price elasticities, probably because imports from nonmembers are less homogeneous than imports from members. Each member has a wide spectrum of commodities to choose from when buying from outside the region, whereas choices are more limited when buying from other members.
Table I presents the results from equations (7) and (8). Both equations performed well. The adjusted correlation coefficients are high and, except for the coefficient of the domestic price term for El Salvador in equation (8), all the parameters have the expected sign. The generalized Durbin-Watson statistic (GDW) suggested by Bhargava et al.  was used to test for serial correlation, since the ordinary Durbin-Watson statistic is not valid for pooled cross-section and time-series data. The GDW for (7) is 2.28, which shows that serial correlation is not rpesent. The GDW for (8) is 1.51, which indicates that serial correlation may be present, but that it probably does not pose a serious problem. No attempt was made to correct for this correlation because, given the size of the sample, such an attempt would probably not improve the efficiency of the estimates.
Imputing Demand Elasticities
The estimates of uncompensated price elasticities and income elasticities of demand in equation (7) and (8) can be used to calculate the compensated demand elasticities [xi].sup.bb, [xi].sup.ba, [xi].sup.wb and [xi].sup.wa for each member. The elasticity of the member's compensated demand for output from i with respect to the price of output supplied by j is [xi].sub.ij = [eta].sub.ij + [mu].sub.i [theta].sub.j, (9) where [theta].sub.j denotes the share of the member's total expenditures that is supplied by j. The elasticity [xi].sub.ab for each member was imputed from the equation [xi].sub.ab = [xi].sub.ba[theta].sub.b/[theta].sub.a. (10)
Equation (9) follows from the Slutsky equation (10) is from the symmetry of substitution effects. Table II reports the results from these equations.
The Welfare Estimates
The welfare effects of abolishing the CACM were calculated using 1979 trade data, and ranges of values for the external tariff t and the excess supply elasticity e.sub.a - [xi].sub.aa. 1979 was chosen because it is the most recent year that reflects a "normal" pattern of trade, one that would prevail in the absence of armed conflict and widespread political unrest in the region. A range of values for t were used because estimates of this rate vary widely and the statutory rates themselves are frequently changed.
Table III gives the estimated welfare effects as percents of total imports and total gross domestic product (GDP) of members in 1979. The results indicate that eliminating the internal preferences would have improved the economic welfare of every member, the amount of the gain increasing with the level of the external tariff. The estimates are not very sensitive to changes in the elasticity of the domestic excess supply.
IV. SUMMARY AND CONCLUSIONS
This paper develops and applies a partial equilibrium model to estimate the static welfare consequences of the CACM. The results suggest that the CACM imposes static welfare losses on each of the member countries. This contradicts the results of previous studies, which all concluded that the CACM provides static welfare benefits.
The static effects are not the sole economic criteria for determining whether the CACM is desirable for its members. In fact, some have argued that the dynamic effects of customs unions are more important (for example, the Inter-American Development Bank  and Todaro ). However, a recent study by Brada and Mendez  found that over the period from 1961 to 1977 the CACM provided dynamic benefits at an average annual rate of only 0.36 percent of the region's gross domestic product (or 0.35 percent if these benefits were cumulative), and this is smaller than most of the annual static welfare effects shown in Table III. Furthermore, our estimates tend to understate the static losses imposed by the internal preferences, because they do not account for the deadweight loss of the alternative tax that would be needed to replace the tariff revenue. This consideration is likely to be important because, as Corden  notes, trade taxes are likely to be efficient compared with the alternative sources of revenue in developing countries.
|Printer friendly Cite/link Email Feedback|
|Author:||Mendez, Jose A.; Rousslang, Donald J.|
|Date:||Jul 1, 1989|
|Previous Article:||Production externalities and long-run equilibria: bargaining and Pigovian taxation.|
|Next Article:||Supply shocks and the interest rate.|