Policy, Reseaieh, and External Affairs WORKING PAPEFR5 Trade Policy Country Economics Department The World Bank August 1990 WPS 480 An Evaluation of Neutral Trade Policy Incentives Under Increasing Returns to Scale Jaime de Melo and David Roland-Holst Under the most plausible scenarios about the entry and exit of finns, a policy of export promotion is likely to be more beneficial than a policy of trade protection for sectors with increasing returns to scale. The Policy, Research, and Extemal Affairs CompIeC disinhutcs PRIE Working Fi4pers todisseminate the findings of work in progress and to encourage the exchange of ideas among FBank staff and all others intcrcsted in dec elopment issucs. 'hese papers carry the names of the authors, rcelcct only thcir %icrs, and sh--uld he uw-d and cited accdmdingls Ihe findings. interpretationc, and conclusions are the authors own TIhiv should not he altnhutcd n thc Lh Wod lia k. is, l[oad of Direuiors. its managenent or anx of us member courutnes. Policy, Research, and Ext, .nal Affairs Trade Policy WPS 480 This paper- a product of the Tradc Policy Division, Country Economics Department -is part of the PRE rcscarch project "Thc Effects of Tradc Regimes on Industrial Competition and Efficiency." Copies are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Rcbecca Sugui, room N 10-031, extension 37951 (15 pages, including tables). Observing the limitations of small domestic be amended to accommodate the welfare effects markets, Bela Balassa has advocated low, of changes in scale efficienicy. uniform, across-the-board tariffs and export subsidies - that is, tariffs of X percent balanced Calculations comparing trade policies that by cxport subsidies of X percent - to overcome achieve neutrality of incentives betwecn sales to the disadvantages of small domestic markets and domcstic and forcign markets suggest that - to permit the exploitation of cconomics of scale under the most plausible scenarios about the through specialization according to comparative cntry and exit of firns - cxport promotion is advantage. likely to bc more beneficial than protection for sectors with increasing retums to scale. Dc Melo and Roland-Holst show analytically and empirically that economics of scale compli- Illustrative calculations of optimal trade catc analysis of the welfare effccts of trade policy packages suggest that the benefits of policy, especially when some sectors have departing from the principle of nondiscrimina- domcstic market power. tion beiween domestic and export sales may be insufficient to justify their higher administrative In particular, the standard distortionary costs costs. of protection under constant rcturns to scalc must The PRE Working Paper Series disseminates the findings of work under way in the Bank's Policy, Rcscarch, and Extcmal AffairsComplex. An objectivc ofthe scries is to get these findings out quickly, even ifpresentations are less than fully polished. The findings, interpretations, and conclusions in these papers do not necessarily represent official Bank policy. Produced by th.e I'RE Disscminaticn Center Table of Contents Page No. Welfare Determinants of Trade Policy Under Increasing Returns 2 Modeling Oligopolistic Domestic Markets 5 A Comparison of Trade Policies Under Constant and Increasing Returns to Scale 7 Evaluation of Protection in Sectors with Scale Economies I 1 Conclusion 13 Notes 14 List of Tables Table 1 Sectoral Features of the Semi-Industiral Economy 8 Table 2 A Comparison of the Welfare Effects of Tariffs and Export Subsidies under Different Pricing Conditions 9 Table 3 Protection and Subsidization of Sectors with Increasing Returns to Scale 11 Table 4 Optimal Export Subsidies for a Given Ten-Percent Import Tariff on All Tradeables 13 An Evaluation of Neutral Trade Policy Incentives Under Increasing Returns to Scale Jaime de Melo and David Roland-Hoist ...in developing countries where protective barriers are high and there is bias against the exportsofmanufactured goods, the limitations of domesticmarketsgenerallypermnitonly the consruction of plants that are below optimum size. By contrast, the disadvantages of small national markets are surmounted in countries where low protective barimers and the lack of bias against exports perm it efficient-scale operations through specialization according to comparative advantage... Balassa (1971, 78-79) New developments in the theory of international trade often suggest, implicitly or explicitly, that in an imperfectly competitive environment, government intervention may be needed to achieve optimality. The most celebrated example in this new literature is the profit-shifting argument of Brander and Spencer (1984). Anotherexample, perhapsmorewidelyapplicable, is the argument developedbyKrugman (1985) showing that protection can serve as an export promotion policy under certain circumstances. These arguments have fostered a literature on strategic trade theory, which deals with conditions of imperfect competition between international trading partners. 1 The trade and development literature, on the other hand, concentrates on the implications of imperfectly competitive domestic markets. In the first of Balassa's (1971) comparative studies on trade policies in developing countries, he argued (in the passage quoted above) that the small size of domestic markets in developing countries was a hindrance to the exploitation of scale economies. He recommended policies to promote exports as a way to break this bottleneck. In his second comparative study of trade policies in semi-industrial countries (Balassa 1982), he ascribed the superior performance of the outward-oriented development strategies in East Asia to the provision of equal incentives to sales on the home and export markets (that is, to the avoidance of home-market bias). Further, in recognition of the learning effects and externalities that accompany the establishment of new industries, Balassa (1975) recommended tempo! ary protection to new activities, which would be gradually scaled down to an across-the-board protection level of about 10 percent. In favorirg market neutrality, Balassa is not only applying the principle of nondiscrimination, but he is also emphasizing trade policy rules, or rules of thumb, that have low administrative costs and do not depend on econometric evidence for their administration. In this paper, we explore the robustness of these strategies in a setting that is representative of semi-industrial market structures and conduct. We recognize that production in some industrial sectors takes place under increasing returns to scale and that pricing in tradable sectors distinguishes between domestic and export markets. The home country is assumed to be a price-taker in both import and export markets. Thus terms of trade are fixed, and we rule out the possibility of strategic trade policy to exploit monopoly power in international trade (the possibility of using trade policy to shift profits to domestic firms). (By contrast, the strategic trade literature assumes that oligopolistic interactions occur in -2- international markets, so thai trade policy affects the home country's terms of trade -circumstances that are more representative of at . eloped countries than semi-industrial ones.) Thne purpose of the paper is then to reexamine the merits of protection, with and without neutrality of domestic and foreign sales incentives,when some manufacturingsectorsoperate underincreasingreturns toscale anddomesticfirms behave oligopolistically. We first derive analycically the comparativestatics of tariff and subsidypolicyin the setting described above, and then we derive criteria for optimal tariff-subsidy policies. Because of intermediate linkages, the welfare effects of trade policy changes are not generally determinate in this multisectoral, general equilibrium setting. This provides the motivation for our simulation analysis, which is prefaced by a summary of the model and a discussion of alternative specifications of oligopoly behavior. Next, we explore systematically the effects of tariffs and export subsidies on welfare with a computable general equilibrium (CGE) model of a representative semi-industrial country with increasing returns and oligopoly behavior in selected manufacturing activities. Finally, we return to the issues of neutrality and optimal protection, comparing the welfare effects of trade policies that provide only import protection for sectors with scale economies with those of policies that combine tariffs and export subsidies. Welfare Determinants of Trade Policy Under Increasing Returns This section presents basic analytical results on the welfare effects of import tariffs and export subsidies. It extends the work of Dixit (1984) and Rodrik (1988) by encompassing export subsidies, the selling of sectoral output on different markets (domestic and export), and imperfect substitution between domestic and imported goods. These features are reproduced in the model structure of the empirical application in later sections. In this setting, we show that both import tariffs and export subsidies contribute to distortions in domestic demand. On the supply side, our results indicate that tariff protection alone may induce producers to divert output from exports to the domestic market rather than expanding production and realizingscale economics. When protection and export incentives are neutral, however, we show that scale economies can be realized that will offset or even outweigh the welfare costs of distortions in demand. We conclude the section with the derivation of a general expression for the optimal tariff-subsidy combination. The expression takes explicit account of the linkages and cost externalities that arise under increasing returns in a general equilibrium context. The expression also shows how optimal trade policy necessitates a mixture of domestic market protection and export incentives to balance the relative profitability of sales in the two markets. Notational conventions follow Dixit (1984). The economy has k sectors, each consisting of ni identical firms (i -= I ,...,k) producing output (z1) for domestic use (yi) and export (xi). As in the numerical application below. fii m output and sales allocation decisions are separable. Hence the allocation decision along a continuous transformation surface. z, = Fi(xiy,), depends only on relative prices in the producer's domestic and export markets for output.2 Each identical firm has a representative cost function ci(x0yj). Domestic and world prices are k-vectorsp and P, respectively, as are ad valorem import tariffs t and export subsidies s. Sectoral domestic prices are an inverse functionp(q) of domestic demands. qi = M; + niYi, themselves an aggregate of imports and domestic output for domestic use. Toevaluate the welfare effect of import and export distortions, we consider all sectors simultaneously in a general equilibrium framework. In a situation in which the government makes only lump-sum -3 - transfersandcommodity preferences are thoseofasinglerepresentativeconsumer, aggregatewelfarecan be decomposed into three components. The first of these is consumer surplus, q (1) g(q) = fp(t!)du -p(q)q, 0 or the area under sectoral demand curves, net of domestic sales revenues. The second component of domestic welfare is the sum of firm profits across sectors: (2) n 'r = n'[py + P(I+s)x - c(x,y), where a caret expands the vector in question into a diagonal matrix and a prime denotes a trans. osition. This expression accounts for revenues from domestic and export sales (which may be subsidized) and total cost. The third component of domestic welfare is tariff revenue net of export subsicay outlays, [t'PM- s 'fPx]. It reflects the direct change in domestic income due to the imposition of trade- distorting measures when world prices are fixed. The resulting domestic welfare function is then given by (3) W = g(q) + n +- t'PM - s Prr =g(q)+:- i+*&-c(xy)j+t'PM. We are interested primarily in the welfare effects of trade policies in the form of tariffs and export subsidies. Total differentiation of expression 3 gives a decomposition of the welfare effects of trade policy changes: (4) dW = t'qp(pdt + pds) + n '(1- c,) [(yp + xp)pdt + (x5 + y3)ds1 - n'(a - c,) zn-dn. Subscripts denote partial differentiation. So qp is the Jacobian matrix of price derivatives for domestic demand, and y and x5 are matrices of direct supply responses in domestic and export markets. The Jacobian cz is the marginal cost matrix for domestic production, including (off the diagonal) cost externalities that may be conferred by increasing returns sectors. The vector a = zc(x,y) contains sectoral average costs. The first term on the right side of expression 4 measures the distortionary cost in consumption and is negative when the domestic demand curve slopes downward. We have assumed that imports and domestic goods are imperfect substitutes in use and that domestic goods are imperfect substitutes in domestic and export sales. These assumptions of product differentiation imply that domestic prices are endogenous and can be affected by either tariffs or export subsidies. So, under the assumption that the aggregate demand curve is downward sloping, the first term on the right side of expression 4 indicates welfare losses from consumption distortions due to a tariff (dt), an export subsidy (ds). or a combination of the two. This term corresponds to the standard welfare costs of protection in the case of constant returns to scale. Scale efficiency effects, which are summarized in the second term on the right side of equation 4, are slightly more complex. Note first tha, .vith normal demand behavior, benefits from protection can arise from expansion of total output in sectors with scale economies. Domestic supply can be expected to rise with domestic prices (yp>O) and exports with the subsidy rate (x3>O). However, the net effect of each of these direct supply responses on total sectoral output depends on the extent of intermarket diversion. Rising domestic prices may induce a diversion from exports to an increasingly lucrative domestic market (xpO). The ultimate effect on output depends on the relative magnitudes of the supply and diversion e,fects (zp yp+ x andz, =XS + Y) and is ultimately an empirical question. What is clear from expression 4, however, is that tariffs and export subsidies can be beneficial if domestic firms' marginal costs are below world prices. Thus, with no firm entry or exit, tariffs and subsidies can be beneficial if the efficiency gains from scale expansion exceed the distortionary costs of protection. The third term in expression 4 represents the effects on welfare of changes in the number of firms. The negative sign indicates that, where there are scale economies, firm entry is detrimental to welfare and that the magnitude of the welfare loss increases with the degree of unexploiUed scale economies. We now pose the questio.,: What would be an optimal choice of tariff and subsidy levels with respect to our domestic welfare function? Given the qualitative symmetry of tariff and subsidy effects, it is unlikely that any policy that implements one without the other could be optimal, but their interplay may be more subtle than simple rules of thumb such ac neutrality (equal rates) would imply. We now derive optimal tariff and subsidy rates in the context of the model already presented. To simplify discussion, we assume that there is no firm entry or exit. To maximize domestic welfare, we form the Lagrangian expression (5) L(t,s) = IV(t,s) +A[z-F(x,y)1. which leads to first-order conditions of the form (6) W, = I 'qp + n '(l - c,) (yp + xp) = A(ZP- Frxp - F,yp) and (7) Ws =t qs + n '(1 - c) (x +ys) = A (Zs -Frs -F,y,). The last two equations can be solved for the vector of optimal tariffs: (8) t n (I-,)yP This expression shows that the optimal tariff depends on the extent of unexploited scale economies and on the elasticities of supplv and demand. Conditions for a nonzero optimal tariff are initial marginal costs below world prices or falling from that lcvel. nonzero elasticity of domestic supply, and finite elasticity of domestic demand. Expression 8 also takes account of interactions across the economy and thus derives consistent optimal policy instruments for all sectors simultaneously. -5- An optimal tariff-subsidy combination wilt tic one that equates the marginal rate of transformation between domestic and export markets with their respective relative prices, that is, one where MRT = Fx I Fy = p / (1 +s) in the one-sector case. More generally, the optimal tariff-subsidy combinationwill be given by (9) (I + s)F, =p~Fy . Assuming that the Jacobians Fx and Fy are diagonal. then the optimal export subsidy would be that which exactly equalizes the value of marginal domestic product between the two markets. In the numerical exercises below, we com?ute the vector of optimal export subsidies for a selected vector of uniform import tariffs. Modeling Oligopolistic Domestic Markets Since the analytical results presented above are ambiguous with respect to the effects of trade policy onwelfare,we use numerical analysisto reveal the relative importance of factors affectingoverallwelfare. First, we describe briefly the structure of the CGE model used for the simulation exercises in the remainder of the paper. As was the case in the analytics of the previous section, the model specifies product ditterentiation between exports and domestic sales and between imports and domestically produced goods in domestic demand. Again, the country is small in international markets. A Leontief technology is specified for intermediate technology. Within sectors, however, domestic and imported inputs are imperfect substitutes. This assumption of product differentiation ,s also maintained for sectors with scale economies. In those se:tors, goods are produced by ni identical firms. Thus all goods produced for domestic sale in the same sector are perfect substitutes, allowing us to aggi 4e sectoral supply across firms. Consumption demand across sectors is dcscribed by a linear expenditure system with nonunitary income elasticities of demand. Finally, value-added is produced by a constant elasticity of substitution technology for two primary factors of production, capital and labor (mobile across sectors), and there is a Leontief technology between aggregate value-added and aggregate intermediates. All final demands arise trom a representative consumer, who also receives net tax revenues as a lump-sum income transfer. As in Harris (1984), fixed costs include capital and labor (equal weight on each). We contrast the case of constant returns to scale (where marginal cost pricing prevails) with two pricing hypotheses in sectors with increasing returns to scale. In the first alternative, we specify an analogue to the case of perfect competition under constant returns to scale. We assume costless entry / exit, so that the threat of entry forces incumbent firms to price at average cost. In this contestable-market scenario (omitting sectoral subscripts), (10) p.= a. for each sector with incrcasing returns to scale, wherep, is the unit price from the constant elasticity of transformation cost function associated with the transformation function describing sales allocation to -6- the domestic and export markets. Here pz is the weighted sum of unit sales prices on the domestic (p) and export (1 +s) markets and a is average costs. In the second alternative, we assume that each (identical) firm behaves in the domestic market as a monopolist facing a downward-sloping demand curve. In equilibrium, each firms equates marginal revenue with marginal costs (ci), that is, - C p ne where E is the endogenous elasticity of demand on domestic sales given by (12) e = eS' + EVS, where F(v) denotes final (intermediate) demand and eF and ev are functions of the parameters describing substitution effects in intermediate and final demand. Because equation 12 is part of the system of equations that must be satisfied in equilibrium, e is endogenous. The variable Q is the representative firm's conjecture about the response of competitors to its output decision with respect to firmj. That is, ifz1 denotes the aggregate output of the rema,ning firms in its sectors, then Q = Azj / Azj. The value of Q is obtained as follows. By choize of units, n is set equal to unity in expression 11. Since the value of f is determined by the parameters and quantities in the model, if one takesp and c, as data, then the value of Q is determined by solving equation 11. We denote by Q the value of the calibrated representative firm's conjecture. We contrast two rules for determining firm entry / exit. Define (13) r-;ry +7, where X is profit per unit of sales and subscriptsy andx denote sales to the domestic and export markets, respectively. In the first alternative, firm entry is determined to ensure that profit per unit of total sales is zero. This assumes that export subsidies allow firms to make a profit on export sales. However, since export subsidies are often justified as a way of defraying the cost of opening new markets, it is reasonable to consider the alternative case in which subsidies to export sales do not give rise to above normal profits. In that second alternative, firm entry is determined to give zero profit on domestic sales. One would expect that the degree of firm collusion would vary with the number of firms. The fewer the number of firms, the more collusive is behavior likely to be. To capture this -ffect, we add the following equation to determine conjectures: (14) Q = n-l, which completes the description of the model. -7- A Comparison of Trade Policies Under Constant and Increasing Returns to Scale We now turn to illustrative numerical calculations based on the model outlined above. All simulations refer to the effects of a departure from free trade in an archetypal semi-industrial economy.5 The structure of the economy in the hypothetical free trade solu.ion is described in table 1. Of the seven sectors, one is nontradable. The data on sectoral structure indicate an open economy with high trade shares in GDP. Sectoral value-added ratios are quite low, indicating the strong interindustry linkages observed in asemi-industrial economy. The three sectorswith increasing returns to scale account for 42 percent of gross output, 73 perccnt of export sales, and 51 percent of import expenses. For the simulations in this section, we assume a low and uniform cost-disadvantage ratio of 10 percent in sectors with economies of scale.'4 Table 2 gives the results of simulations comparing the cffects of tariff protection and export subsidization. All simulations reler to 10-percent tariff ratcs and 10-percent export subsidy rates. We contrast four scenarios: constant returns to scale (CRTS) across all sectors, contestable-market pricing for the three sectors with increasing returns to scale, and Cournot competition with total profit or domestic profit determining firm cntry. The results presented in table 2 are for protection or export subsidization of (1) sectors with constant returns to scale only (primarv, food processing, and traded services); (2) sectors with increasing returns to scale only (consumer goods, producer goods, and heavy industry); or (3) all traded sectors. Two measures of the welfare effects of changes in trade policy are reported in table 2. The equivalent variation measure is derived from the indirect utilitv function associated with the Stone- Geary utility function assumed for final demand. It is an aggregate measure of efficiency gains and losses in production and of efficiency losses in consumption. Equivalent variation measures how much the representative consumer would have to be compensated at the new set of prices to be indifferent to the bundle of goods now available at the initial set of prices. The second measure is the scale efficiency gain or loss from moving along average cost curves.5 Like equivalent variation, scale efficiency evaluates the new output level at old prices, so that the measure controls for shifts in the average cost curve induced by changes in factor and product prices. Now let us examine the results presentcd in tabk. '.. Consider First the results tinder constant returns to scale in the first three columns. In the case of tariff protection, there is a welfare loss from protection regardless of which group of sectors is protected. As expected, the welfare cost of protection increases with the numberofsectors beingprotected. Note that the correspondingwelfare lossestimates forexport subsidizationyieldvery similarorders of magnitude, with the differences dependingon tradevolumes and substitution elasticities. Turn now to the case of contestable-market pricing, which assumes increasing rcturns to scale for the consumer goods. producer goods. and heavy industrysectors. Now protection ofsectors with constant returns to scale is muchi more costlv because a scale efficiency loss rcsults when resources are pulled out of sectors with scale economies. The loss of scale efficiency occurs because firms are forced to produm_ higher up on their avcragc cost curvcs. Bv contrast. protection of sectors w ith increasing returns to scale is much less costly becausc of the scale ei'ficiencv gain. Note, howevei. that ceen though protection is provided across the board for scctorsAwith increasing returns to scale, there is a scale efficiencv loss in one Table 1. Sectoral Features of the Semi-Industriat Economy Share in Imports/ Elasticity of Export Import Cost dis- Domestic price gross Exports/ domestic substitution supply elasticity advantage eLasticity Sector output (X) output (X) sales (X) in production elasticitya of demanda ratio of demand Primary 8.9 4.9 40.4 2.5 0.75 1.8 -- -- Food products 9.6 2.5 6.5 1.5 1.5 2.5 -- -- Consumer goods 14.4 32.5 14.2 1.0 1.5 2.4 0.1 1.6 Producer goods 20.1 16.6 19.2 0.9 1.5 2.2 0.1 1.3 Heavy industry 7.7 31.9 41.0 0.9 1.5 1.9 0.1 1.4 Traded services 13.2 24.4 7.5 1.5 1.5 2.0 -- Nontraded services 26.1 -- 0.9 -- -- -- a. Expenditure-compensated price eLesticities. For imports (exports), expenditures (sales) on constant elasticity of substitution X (transformation) a egate of domestic and import (export) goods held constant. b. The cost disadvantage ratio (difference between average and marginal costs divided by average cost) is a measure of unrealized economies of scale. Table 2. A Comparison of the Vetfare Effects of Tariffs ard Export Smidi.-s under Different Pricing Conditions Cournot b Courrot CRTS Contestable3 (total profit) (domestic profit)c Sector CRTS IRTS All CRTS IRTS Ail CRTS IRTS Alt CRTS IRTS All Ten-percent tariff Equivalent variation -7 -9 -12 -50 -2 -46 12 60 78 -33 -30 -57 Scale efficiency (total) -42 6 -34 19 69 88 -27 -22 -47 Producer goods -14 -8 -21 9 33 42 -7 -8 -14 Consumer goods -24 15 -8 8 15 23 -3 -6 -8 Heavy industry -4 -1 -5 2 21 23 -3 -6 -8 Firm entry (+)/exit (-) Producer goods -5 -9 -14 -1 2 1 Consubner goods -5 -1 -5 -1 4 4 Heavy industry -2 -1 -12 0 4 4 Ten-percent export subsidy Equivalent variation -6 -19 -13 -45 52 27 52 -168 -108 18 15 36 Scale efficiency (total) -39 74 42 57 -155 -97 24 25 46 Producer goods -14 38 26 23 -69 -46 8 7 14 Consumer goods -15 22 10 21 -48 -27 12 14 24 Heavy industry -10 14 6 13 -38 -24 4 4 8 Firm entry(+)/exit (-) Producer goods -8 23 15 -4 2 -1 Consumer goods -6 11 6 -4 0 -4 Heavy industry -9 22 13 -5 1 -4 CRTS = constant returns to scale; IRTS = increasing returns to scale; Note: All figures are basis points. Figures for equivalent variation and scale efficiency are basis points of GOP (e.g., -168 is 1.68% of GOP); figures for entry/exit are basis points of initial number of firms. a. Pricing according to equation 10. b. Pricing according to equation 11, with firm entry/exit determined by total profits, so that i = 0. c. Same as note b but with firm entry determined by profits on domestic sales, so that sy = 0. - 10- sector. Finally, protecting all actors results in a larger welfare loss than under the scenario of constant rc urns to scale in all sectors because of scale efficiency losses. o,ow, compare these results with those for export subsidization in thc bottom half of the table. The subsidization results corroborate Balassa's assertion that specialization according tocomparativeenables the disadvantages of small national markets to be surmounted in sectors with unexploited economies of scale. As before, the benefits are greatest when trade policy is confined to sectors with increasing returns to scale. The export subsidization effects dominate the tariff protection effects because of the difficulty of substituting away from imports when incentives are provided to domestic producers, and the ease of expanding sales in international markets when market share is small. When contestable market pricing is replaced by Cournot competition. the welfare effects of trade policy are affected by three additional adjustment mechanisms: firm cntry / exit (the mechanism that achieves zero profits in long-run equilibrium), the endogeneity of firm collusion, and-generally less significant-the anti-competitive cffect of protection. which lowers the elasticity of domestic demand, e.6 The most important of these mechanisms in influencing (hc welfare effects of trade policy under Cournot competition is thc pattern of firm entry or exit. Take the case of tariff protection, which raises the profitability of domestic sales and lowers the profitability o; export sales because of induced appreciation in the real exchange rate. If firm entry or exit depends on the joint profitability of sales in both markets, then there is firm exit beca dse sectors with increasing returns to scale happen to have high export shares in our numerical example. Firm exit allows the remaining firms to move down their average cost curves, thereby reaping the benefits of more efficient scale. If, however, one assumes that firm entry is governed by profits from sales in the domestic market alone, then there is firm entry and protection results in a welfare loss because of the loss in scale efficiency. By symmetry, a rpolicy of subsidizing exports has opposite effects. Export subsidization leads to crowding-in if the decision to enter depends on total profits because export subsidies lead to large profits on export sales. For the caseof export subsidizatior- of sectorswith increasing returns to scale, the welfare loss amounts to 1.7 percent of GDP. If. on the other hand, one assumes that cxport subsidies do not give rise to abnormal profits but rather contribute to defraying the costs (and risks) of selling in new markets, there is a small welfare gain. Interestingly, in the case of export subsidization in sectors with scale economies there is a scale efficiency gain despite some firm entry. The results presented in table 2 clearly show thit if Cournot competition is a reasonable representation of behavior in sectors with increasing rcturns to scale, lirm cntry and exit are crucial in dctermining the sign and magnitude of the effects of trade policy interventions. For the illustrative trade policy interventions reported in table 2. one could argue that entry behavior based on total profits is the more reasonable assumption. However, one can interpret a policy of protection more broadly as one that produces home-markct bias because it usuallv involves quotas and nontariff barriers that create harriers to entry as competition from abroad is suppressed. Then a sheltered domestic market is likely to lead to cxcessive firm entry because of high profits.7 On the other hand, the experience of successful East Asian exporters suggests that it was the provision of export incentives that put domestic producers on an equal looting with their foreign Competitors. As argued byvFrischtak ct al. (1 989.1 (0-l l). exporters nced support to make the commitment to riskicr activities that have a long lead time and sunk costs for identifing suitable markets and setting up distributionchannels. Underthisintcrpretation orthccostsofestablishingsucccssfulexportactivities. - 11 - subsidies (or incentives that increase the relative profitability of exports) are not likely to give rise to abnormally high profits and hence to induce excessive entry. An altemative interpretation would emphasize that the appropriate policy in a setting of increasing returns to scale is to promote competition in domestic markets. This logic recognizes that imperfect competition in domestic markets can act as an export barrier by increasing the relative profitability of domestic operations. Ideally, industrial policy would be coordinated with trade policy to encourage the exploitation of efficient scale, promoting exports while avoiding excessive entry.8 Evaluation of Protection in Sectors with Scale Economies We return to the issues raised in the introduction: are there welfare gains from protecting sectors with increasing returns to scale and how does import protection compare with neutral incentives (for example, with tariffs and export subsidies at equal rates)? To answer these questions, we report on simulations in which we contrast across-the-board tariffs of 15 percent with across-the-board export subsidies of 15 percent, both in sectors with scale economies. Protection and export subsidies are confined to the consumer, producer, and capital goods sectors. Nowwe assume a cost disadvantage ratio of 20 percent, a value more in line with the unexploited economics of scale in the manufacturing sector of a typical semi-industrial country. The results of these simulations appear in table 3. In the constant-returns-to-scale benchmark case, there is, as before, a welfare loss from protection alone or from export subsidization alone. Neutrality, however, is less costly because the distortion introduced by the export subsidy partly offsets the distortion introduced by the tariff. Table 3. Protection and Subsidization of Sectors with Increasing Returns to Scale (cost-disadvantage ratio of 20 percent) Proct-d-O-bidy CRTS Contetbea (dOte3dC ,oit) 15-percent tariff Equivalent variation -18 12 -106 Scale efficiencv 0 31 -88 15-percent export subsidy Equivalent vanation -42 274 86 Scale efficiencv 0 336 110 15-percent tariff and export subsidv Equivalcnt vanation -31 281 4 Scale efficiencv 0 322 27 CRTS is constant returns to scale. Note: AMI figures are basis points of GDP (e.g., -106 is -1.06 percent of GDP). a. Pncing according to equation 10. b. Pricing according to equation 11. vith firm entrv and exit determined bv profits on domestic sales. so Ithat .7rT = 0 -12- Thc same pattem of welfare estimates emnerges under the assumption of contestable markets. I lowever, because we have assumed a greater degree of unexploited economies of scale, the magnitudes are larger than in table 2. There is a welfare gain of 2.7 percent of base year GDP to be reaped from subsidizing export sales of sectors with increasing returns to scale. Note the superiority of export subsidization over import protection, which springs from our assumption that exporters face a perfectly elastic foreign demand whereas dorrmestically produced goods face a downward sloping domestic demand curvc. 1-ence thc incentives created by export subsidization are more direct than those created by protection fordomestic sales. V/hik' the export demand specification deserves furtherscrutiny, it appears to correspond to the expericnce of countries that have followed an export-led development strategy. In the contestable-market scenario, neutrality produces the largest welfare gains from trade incentives to sectors with increasing returns to scale and sustains the recommendations of Balassa (1975, 1989). In the case of Cournot competition, under the assumptions about firm entry, subsidization of exporls dominates the alternativeof providing equal incentives to domestic and exportsales. This occurs becausewc have assumed that subsidies to exportsdo not give rise to profits (and hence do not inducefirm entry) vhereas protection on the domestic market gives rise to profits and induces firmentry. Aswesaw above, tirm entry results in scalc efficiencv losses, an effect that comesout clearly in the caseof protection to domestic salcs. In that case, tariff protection results in a welfare loss that exceeds 1 percent of GDP. It is obvious that the results under Cournot competition are quite sensitive to the determinants of the number of firms-about which little is known. In the simulations reported here, we have attempted to portray the stylized facts suggested by the comparative studies of foreign trade regimes in developing countrics. These studies reveal that countries that have followed import-substitution industrialization stratcgies have often tended to provide made-to-measure protection for all domestic activities. This protection has, in turn, tended to create excess profit opportunities from domestic sales. When pushed to the cxtr, me, exccssivc across-the-board protcction of industrial activities has been shown to result in excessive firm entry. We concludc by comparing ncutrality of incentives with optimal trade policy. The "optimal" trade policv package is obtained by maximizing the value of the utility function for the representative consumer, taking tariffsas givcen and export subsidies as endogenous policy instruments. To facilitate the comparison uith the rcsults in table 2, we fi-x all import tariffs at 10 percent for all sectors. The results of the calculation of these "optimal" trade policy packages appear in table 4. Note first that undcr the assumption of constant rcturns to scale in all sectors, the numerical calculations confirm the well-known rcsults predictcd bv Lcrncr (1936), namely that across-the-board tariff and export subsidics at the same rates arc cquivalent to free trade.9 Note also that the equivalent variation measure achicvcs a maximum of zero in this case because departure from free trade cannot be beneficial under constant returns to scale. Under increasing rcturns to scale. Lcrner svmmetrv still holds: across-the-board tariffs and export subsidies at the samc rate are equivalent to free trade. But, as the pattern of export subsidy figures shows, neutrality is no longer optimal. Two results stand out in the contestable-market case. First, as expected, optimalitv rcquircsthatgrcaterincentivesbeprovidedtosectorswithincreasingreturnstoscale. Second, the diffcrcncc in wclfare henef its is small between optimal trade policy and the rules of thumb advocated by Balassa (1975. 1989)- across-the-board protection (withequal incentivestoexports) formanufacturing activities. Here.optimalitvdominatestherulcofthumbofincentiveneutralitybylessthanlObasispoints. - 13- Given the notorious lack of the precise elasticity estimates needed to calculate optimal incentivcs, the illustrative calculations here do not support a departure from the rule of thumb advocated by Balassa. Table 4. Optimal Expor. Subsidies for a Given Ten-Percent Import Tariff on All Tradeables (cost-disadvantage ratio of 10 percent) Cmn CTRlS Cmttwabk (do crpflt)6 Equivalent variation ° 66 (5g)b 55 (-6)b Scale eMckttcy (total) 128 102 Producer goods 54 32 Consumer goods 49 52 Heavy industry 24 18 Firm entry Producer goods -4 Consumer goods -12 Heavy industry -11 Export subsidy Primary products 10 -6 62 Food processing 10 -I 36 Producer goods 10 24 33 Consumer goods 10 27 25 Heavy industry 10 25 41 Traded services 10 - 1 41 Note: The subsidy is in percentage points. Other figures are in basis points. a. Pricing according to equations 11, with firm entry determined by profits on domestic sales. so that.7, = 0. b.Corresponding equivalent variation figure under neutrality, that is. from combining a 10- percent import tariffwith a 10-percent export subsidy in sectors with increasing returns to scale. In the case of Cournot competition. however, the optimal pattern of export subsidies dcparts further from neutrality. Under thisscenario, an optimal policywould encourage firm exit to reapscale economies. As the figures in the last column of table4 indicate, firm exit would be achieved by providinghigherexport subsidies to sectors with constant returns to scale.10 Now departure from a simple rule of thumb yields larger welfare benefits. However, the discussion of table 2 suggested that the results under Cournot competition areverysensitiveto thedeterminantsof firmentry, so these resultsshould be interpretedwith care. Conclusion This paper set out to test the robustness of Balassa's recommendation of neutral incentives to domestic and export sales in a setting where some sectors have domestic market power. Wc have shown analytically that the welfare effects of trade policy are more complex than they are in a setting of across- the-board constant returns to scale. In particular, we have shown, analytically and numerically, that the standard distortionary costs of protection emphasized under conditions of constant returns to scale must be amended to accommodate, among other things, the welfare effects of changes in scale efficicncy. - 14- be amended to accommodate, among other things, the welfare effects of changes in scale efficiency. Illustrative numerical calculations also show that the magnitude of the welfare gains or losses from trade policy intervention are sensitive to the determinants of firm entry and exit. Calculations comparing trade policies that achieve neutralityof incentives between sales to domestic and those to foreign markets found such policies to be generally superior to policies creating non-neutral incentives. Numerical results also suggest that export promotion is likely to be more beneficial than protection for sectors with increasing returns to scale. Finally, illustrative calculations of optimal trade policy packages suggest that the benefits of departing from the principle of neutrality, or nondiscrimination between domestic and export sales, may be insufficient to justify their higher administrative costs. Notes The research reported here is part of the World Bank research project "Industrial Competition, Productive Efficiency and Their Relation to Trade Regimes." RPO 674-46. The numerical work is based on a model developed in de Melo and Tarr (forthcoming). The views expressed here are those of the authors and should not be attributed to their affiliated institutions. 1. See Harris (1989) and Helpman and Krugman (1989) for surveys of this work. 2. Imperfect substitutability in the allocation of sales implies that F; / Fyi varies along a convex transformation frontier. Lower case letters indicate partial derivatives. 3. The archetypal economy was obtained from the free trade solution of a seven-sector CGE model calibrated to the Korean economy for the year 1982. For a description of the data set and parameters values, see de Melo and Roland-Holst (forthcoming). 4. The cost disadvantage ratio is the difference between average and marginal costs, divided by average costs. It is a measure of unrealized economies of scale. 5. The aggregate scale efficiency measure is calculated by usiig currcnt outputs as weights. For further discussion, sce de Melo and Roland-Hoist (forthcoming). 6. The magnitude of this effect is small for the functional forms specified here and is not reported. For a discussion of its magnitude, see de Melo and Roland-Holst (forthcoming). Also, see Devarajan and Rodrik (1989). 7. Frischtak et al. (1989) document the pervasive barriers to competition in the manufacturing sectors of developing countries. Eastman and Stykolt (1962) is an early example of a model in which protection leads to firm entry. The typical example is the automobile industry in Latin America (see Baranson 1968). 8. In this regard, the Korean experience during the 1970s is instructive. An activist industrial policywas successful in promoting the growth of large conglomerates and reaping the benefits of scale economies. While exports benefited from this policy, oligopolistic marketsdeveloped, and avigorous antitrust policvwas established in theearly v980sto promote greatercompetition in domestic markets. For further discussion, see Lee, Urata, and Choi (1988) and World Bank (1987). 9. Since there is no guarantee that the optimal vector of subsidies is unique, numerical verification of Lerner symmetry is a useful computational check. 10. While the results of these optimal calculations appear reasonable, there is no guarantee that the comnputed optima are global optima rather than local optima. Hence these results should be viewed as suggestive and subject to further scrutiny. - 15- Reffnces Balassa, B. 1975. "Reforming the System of Incentives in Developing Countries." World Development (June): 365-81. Balassa, B. 1989. 'Tariff Policy and Taxation in Developing Countries." PPR Working Paper No. 281. Washington, D.C.: Policy, Planning, and Research, World Bank. Balassa, B., and associates. 1971. The Structure of Protection in Developing Countries. Baltimore, Md.: Johns Hopkins University Press. Balassa, B., and associates. 1982. Trade Strategies for Semi-Industrial Countries. Baltimore: Johns Hopkins University Press. Baranson, J. 1968. The Automotive Industry in Latin America. Praeger: New York. Bergsman, J. 1974. "Commercial Policy, Allocative Efficiency and X- Efficiency." Quarterly Joumal of Economics 409-33. Brander, J., and B. Spencer. 1984. "Tariff Protection and Imperfect Competition." In H. Kierzkowski, ed.. Monopolistic Competition in International Trade. Oxford: Oxford University Press. Chenery, H. 1975. "The Striicturalist Approach to Development Policv." American Economic Review. Corden, M. 1967. "Monopoly, Tariffs and Subsidies." Economica 50-8. Devarajan, S., and D. Rodrik. 1989. "Pro-competitive Effects of Tariff Reform. ' Working Paper, J.F. Kennedy School of Government, Harvard University, Cambridge, Mass. Dixit, AK. 1984. "International Trade Policy for Oligopoly Industries." Economic Joumal 94: 1-16. Eastman, H., and S. Stykolt. 1962. "A Model for the Study of Protected Industries." Economic Journal 70:336-47. Frischtak, C., with B. Hadjimichael, and U. Zachan. 1989. Competition Policies for Industrializing Countries. Policy and Research Series No. 7. Washington, D.C.: World Bank. Harris, R.G. 1984. "Applied General Equilibrium Analyses of Small Open Economies with Scale Economies and Imperfect Competition."American Economic Review 74:1016-33. Harris, R.G. 1989. "The New Protectionism Revisited." Canadian Joumal of Economics 24, no. 4: 751-78. Helpman, E., anc P. Krugman. 1989. Trade Policy and Market Structure. Cambridge, Mass.: MIT Press. Horstmann, I.. and J. Markusen. 1986. "Up the Average Cost Curve: Inefficient Entry and the New Protectionism." Joumal of International Economics, 225-48. Krugman, P. 1984. "Import Protection as Export Promotion." In H. Kierzkowski, ed.. Mfonopolistic Competition in Intemational Trade. Oxford: Oxford University Press. Lee, K., S. Urata, and 1. Choi. 1988. "Recent Developments in Industrial Organization in Korea." World Bank, Washington, D.C. Lerner, A 1936. "The Symmetry Between Import and Export Taxes." Economica 3: 306-13. Melo, J. de. and D.W. Roland-Hoist. Forthcoming. "Industrial Organization and Trade Liberalization: Evidence from Korea" in R. E. Baldwin. ed., Empirical Studies of Commercial Policy Chicago: University of Chicago Press. Melo, J. de. and D. Tarr. Forthcoming. A General Equilibrium Analysis of US Foreign Trade Policv Cambridge, MIT Press. Rodrik. D. 1988. "Imperfect Competition. Scale Economies, and Trade Policy in Developing Countries." In R.E. Baldwin, ed., Trade Policv Issues and EmpiricalAnalysis. Chicago: University of Chicago Prcss and Cambridge: National Bureau of Economic Rcscarch. Venables, A 1985. "Trade and Trade Policy with Imperfect Competition: The Case of Identical Products and Free Entr'." Joumal of Intemational Economics 1-19. World Bank. 1987. Korea: ,fanaging the Industrial Transition. Washington. D.C.: World Bank. ERE W kin PaDer Series Contact ide AuAhor for pape WPS455 A Formal Fstirnation of the, Effect Junichi Goto June 1990 M. T. Sanchez of tile MFA on Clothing Exports 33731 from LDCs WPS456 Improving the Supply and Use of S. D. Foster June 1990 Z. Vania Essential D)rugs in Sub-Saharan Africa 33664 WPS457 i-inancing Health Services in Africa: Germano Mwabu June 1990 Z. Vania An Assessment of Alternative 33664 Approaches WPS 58 Does Japanese Direct Foreign Kenji Takeuchi June 1990 J. Epps Investment Promote Japanese Imports 33710 from Developing Countries? WPS459 Policies for Economic Development Stanley Fischer June 1990 WDR Office Vinod Thomas 31393 WPS460 Does Food Aid Depress Food Victor Lavy July 1990 A. Murphy Production? The Disincentive 33750 Dilemma in the African Context WPS461 Labor Market Participation, Returns Shahidur R. Khandker July 1990 B. Smith to Education, and Male-Female 35108 Wage Differences in Peru WPS462 An Aiternative View of Tax Anwar Shah July 1990 A. Bhalla Incidence Analysis for Developing John Whalley 37699 Countries WPS463 Redefining Government's Role in Odin Knudsen August 1990 K. Cabana Agriculture in the Nineties John Nash 37946 WPS464 Does A Woman's Education Affect Shoshana Neuman August 1990 V. Charles Her Husband's Earnings? Results Adrian Ziderman 33651 for Israel in A Dual Labor Market WPS465 How Integrated Are Tropical Timber Panos Varangis August 1990 D. Gustafson Markets? 33714 WPS466 Is There An Intra-Household Lawrence Haddad August 1990 J. Sweeney Kuznets Curve? Ravi Kanbur 31021 WPS467 Structural Adjustment and Living Nanak Kakwani August 1990 B. Rosa Conditions in Developing Countries Elene Makonnen 33751 Jacques van der Gaag WPS468 Does the Structure of Production Indermit Gill August 1990 M. Abundo Affect Demand for Schooling in Peru? 36820 PRE Working Paper Series Contact TA6 Auhor for paper WPS469 Modeling Economic Behavior in J. Barry Smith August 1990 M. Abundo Peru's Informal Urban Retail Sector Morton Stelcner 36820 WPS470 What Do Alternative Measures of Alexander J. Yeats August 1990 J. Epps Comparative Advantage Reveal 33710 About the Composition of Developing Countries' Exports? WPS471 The Determinants of Farm Gershon Feder August 1990 C. Spooner Investment and Residential Lawrence J. Lau 30464 Construction in Post-Reform China Justin Lin Xiaopeng Luo WPS472 Gains in the Education of Peruvian Elizabeth M. King August 1990 C. Cristobal Women, 1940 to 1980 Rosemary Bellew 33640 WPS473 Adjustment, Investment, and the Riccardo Faini August 1990 R. Sugui Real Exchange Rate in Developing Jaime de Melo 37951 Countries WPS474 Methods for Measuring the Effect Anne Maasland August 1990 P. Dixon of Adjustment Policies on Income 39175 Distribution WPS475 Does Divestiture Matter? A Ahmed Galal August1990 G. Orraca-Tette Framework for Learning from Experience 37646 WPS476 Health Insurance in Sub-Saharan Ronald J. Vogel August 1990 K. Brown Africa: A Survey and Analysis 35073 WPS477 Private Participation in the Delivery Thelma A. Triche August 1990 M. Dhokai of Guinea's Water Supply Services 33970 WPS478 Interrelations Among Child Mortality, John Marcotte August 1990 S. Cochrane Breastfeeding, and Fertility in John B. Casterline 33222 Egypt, 1975-80 WPS479 Conversion Factors: A Discussion Michael Hee August 1990 E. Zamora of Alternate Rates and Corresponding 33706 Weights WPS480 An Evaluatior of Neutral Trade Policy Jaime de Melo August 1990 R. Sugui Incentives Under Increasing Returns David Roland-Holst 37951 to Scale WPS481 The Effects of Trade Reforms on James Tybout August 1990 R. Sugui Scale and Technical Efficiency: Jaime de Melo 37951 New Evidence from Chile Vittorio Corbo WPS482 Membership in the CFA Zone: Shantayanan Devarajan August 1990 R. Sugui Odyssean Journey or Trojan Horse? Jaime de Melo 37951