BILATERALISM IS GOOD: TRADE BLOCS AND STRATEGIC EXPORT SUBSIDIES

Transcription

1 Oxford Economic Papers 49 (1997), BILATERALISM IS GOOD: TRADE BLOCS AND STRATEGIC EXPORT SUBSIDIES By DAVID R. COLLIE Cardiff Business School, Aberconway Building, Colum Drive, Cardiff CF1 3EU, UK; This paper considers the effect of exogenous trade bloc enlargement in a multi-country version of the Brander-Spencer export subsidy game. In the single-shot game, it is shown that trade bloc enlargement leads to a reduction in the Nash equilibrium export subsidies and thereby increases the welfare of the exporting countries. Although the welfare of the importing countries decreases, world welfare may increase if the export subsidies are financed by distortionary taxation. When the export subsidy game is infinitely repeated, it is shown that trade bloc enlargement reduces the critical discount factor making it easier to sustain free trade. 1. Introduction DESPITE the successful conclusion of the Uruguay Round and the consequent strengthening of the multilateral trade regime, there is still some concern that increased regional integration may result in the division of the world economy into competing trade blocs. This would, it is often argued, lead to an escalation of trade conflict with the result that trade barriers were increased and world welfare reduced. Krugman (1991) has highlighted this possibility using a simple but provocative model in which non-cooperative tariff-setting results in world welfare being minimised when the world economy is divided into three trade blocs. He employs a conventional trade theory model where tariffs are used to improve the terms of trade, however, this is not the only rationale for trade policy. In oligopolistic industries, Brander and Spencer (1985) have shown that countries can use strategic export subsidies to shift profits from foreign to domestic firms. Although export subsidies are prohibited by the World Trade Organisation (WTO), many industrialised countries use export credits and export credit insurance to give their firms an advantage when exporting capital goods to developing countries 1 This paper looks at how the division of the industrialised countries into trade blocs affects their use of strategic export subsidies in oligopolistic industries. The question has some practical relevance as the creation of the Single European Market in 1992 has led the European Commission to call for the harmonisation of the terms on which EC countries give export credit insurance. 2 1 The history of officially supported export credits is examined in Ray (1995), and he also describes the development of international rules governing their use. 2 Exporters in the EC have argued that the proposed reduction in insurance cover would harm their competitiveness against other OECD exporters when telling capital goods in developing countries, see the Financial Times, 31st January Oxford IMrentty Pros 1997

2 D. COLLIE 505 In Krugman (1991) consumers have Dixit-Stiglitz preferences, and each country is endowed with a unique differentiated product. The countries are divided into symmetric trade blocs that pursue a policy of free trade within the trade bloc but impose a common external tariff on imports from the rest of the world. The trade blocs set their tariffs non-cooperatively with each trade bloc setting its tariff to maximise its welfare given the tariffs set by the other trade blocs; the result is a Nash equilibrium in tariffs. An increase in the size of the symmetric trade blocs increases the market power of each trade bloc, and results in an increase in the Nash equilibrium tariffs. As a result of trade diversion and the higher tariffs, world welfare initially declines as the trade blocs become larger until it reaches a minimum with three trade blocs. A further increase in the size of trade blocs increases world welfare and it is maximised when there is one trade bloc which is equivalent to global free trade. Krugman (1993) shows that the reduction in welfare as trade blocs become larger is mainly due to increased trade diversion rather than the higher Nash equilibrium tariffs. In Bond and Syropoulos (1996) it is shown that the results of Krugman (1991) are not robust to changes in the pattern of comparative advantage. If a reduction in the number of trade blocs makes inter-bloc trade less important then the market power of the trade blocs is reduced and the Nash equilibrium tariffs will fall. This implies that world welfare will be minimised when there are more than three trade blocs. Sinclair and Vines (1994) point out that Krugman (1991) assumes that the trade blocs are customs unions with a common external tariff when many trade blocs are actually free trade areas with each country setting its own tariff on imports from outside the free trade area. As a free trade area becomes larger the market power of a country will be reduced as more of its imports come from within the free trade area, and hence the Nash equilibrium tariffs set by countries will fall. Sinclair and Vines (1994) do not consider how world welfare varies with the number of free trade areas, but trade diversion will still occur as with a customs union. The Nash equilibrium tariff rates calculated by Krugman (1991) are much higher than actual tariff rates, and this may be because trade blocs realise that they are engaged in a repeated game rather than a single-shot game. 3 In an infinitely repeated game more co-operative outcome such as free trade can be sustained by the threat of retaliation with the Nash equilibrium tariffs. Bond and Syropoulos (1995) look at how the formation of trade blocs affects the sustainability of free trade in an infinitely repeated version of the Bond and Syropoulos (1996) model. They show that as trade blocs become larger the incentives to deviate from free trade become larger and this makes it more difficult to sustain free trade. Bagwell and Staiger (1994) consider how the announcement of a customs union affects tariff setting in an infinitely repeated game. They show that there will be a honeymoon period with lower tariffs after 3 The tariff rates ralrnint«h by Krugman (1991) are broadly consistent with the Smoot-Hawley tariff rates when trade policy was set non-cooperatively.

3 506 BILATERALISM IS GOOD the customs union is announced, but this may be followed by higher tariffs after the customs union is implemented. In all of these models tariffs are used to improve the terms of trade, but the new trade theory has provided a rationale for strategic trade policies in oligopolistic industries. When domestic and foreign firms compete in the domestic market, Brander and Spencer (1984) have shown that a tariff can be used to extract rent from the foreign firms and to shift profits to domestic firms. In Brander and Spencer (1985), where a domestic and a foreign firm compete as Cournot duopolists in a third market, it was shown that the domestic country can gain by using a profit-shifting export subsidy but if both countries use export subsidies then welfare in the Nash equilibrium, assuming symmetry, will be lower than under free trade. In an infinitely repeated version of the export subsidy game, Collie (1993) has shown that free trade can be sustained by the threat of retaliation with the Nash equilibrium export subsidies provided the countries are similar and the discount factor is sufficiently high. There has been no attempt to look at trade bloc formation in models of strategic trade policy apart from the brief analysis in Sinclair and Vines (1994). They extend the Brander and Spencer (1984) model to consider the effect of trade bloc formation on the Nash equilibrium tariffs. For customs unions their results are ambiguous tariffs initially increase then fall as customs unions become larger whereas for free trade areas they unambiguously show that tariffs fall as the free trade areas become larger. Sinclair and Vines (1994) do not consider the welfare effects of trade bloc formation. This paper looks at how the exogenous formation of the industrialised countries into symmetric trade blocs affects their use of strategic export subsidies in oligopolistic industries. The model presented is a multi-country extension of the Brander and Spencer (1985) export subsidy game where the industrialised countries export to countries in the rest of the world. The paper begins by looking at the Nash equilibrium of the single-shot export subsidy game, and then considers the sustainability of free trade in the infinitely repeated export subsidy game. In the single-shot game, the Nash equilibrium export subsidies are derived as functions of the number of countries in each trade bloc, and it is shown that export subsidies are reduced as trade blocs become larger. This is because trade bloc enlargement reduces the effectiveness of export subsidies. The reduction in the Nash equilibrium export subsidies by all the trade blocs leads to an increase in the welfare of all the industrialised countries. Obviously, the terms of trade of the importing countries will deteriorate so reducing their welfare but, surprisingly, world welfare may increase if the export subsidies are financed by distortionary taxation as in Neary (1994). This provides an economic rationale based upon worldwide efficiency for the WTO prohibition of export subsidies. 4 In the infinitely repeated export subsidy 4 An alternative rationale for the WTO prohibition of export subsidies is suggested by Bagwell and Staiger (1996). In their model, export subsidies serve to co-ordinate the entry decisions of firms and the prohibition of export subsidies may increase welfare if the co-ordination provided by export subsidies does more to prevent entry than to promote entry.

4 D. COLLIE 507 game, free trade can be sustained by the threat of retaliation with the Nash equilibrium export subsidies if the discount factor of the trade blocs exceeds the critical value. The critical discount factor is such that the instantaneous gain from unilaterally using an export subsidy is just equal to the present discounted value of the losses from retaliation. As trade blocs become larger the instantaneous gain from unilaterally using an export subsidy decreases because export subsidies become less effective, but the losses from retaliation also decrease because of the increase in Nash equilibrium welfare. It turns out that the overall effect of trade bloc enlargement is to reduce the critical discount factor making it easier to sustain free trade. 2. The model The model is a multi-country version of the Brander and Spencer (1985) analysis of profit-shifting export subsidies based upon Collie (1993), and as in Krugman (1991) some special assumptions will be employed to yield a tractable model. The model is partial equilibrium and only analyses the effect of trade bloc enlargement on the oligopolistic industry located in the industrialised countries. All the output of the oligopolistic industry is assumed to be exported to countries in the rest of the world, hence the model concentrates upon the profit-shifting effect and ignores the usual trade diverting effect of trade bloc enlargement. 5 The model is symmetric, all the industrialised countries are assumed to be identical, and all the trade blocs are of equal size, hence it can be analysed by considering a representative trade bloc. The simplest possible functional forms, linear demand and constant marginal cost, will be used to allow explicit solutions to be obtained for welfare as a function of the number of countries in each trade bloc. This assumption is necessary, in particular to analyse the sustainability of free trade in an infinitely repeated game. Therefore the model is not general; however, the lack of generality can be justified by the possibly counter-intuitive results obtained about trade bloc enlargement. In each of the N industrialised countries there is a single firm with constant marginal cost c that exports its output to the rest of the world where it engages in Cournot competition with the other firms from the industrialised countries. 6 The N industrialised countries are divided into B symmetric trade blocs with M 3 Incorporating demand in the industrialised countries into the model, would introduce the possibility of the trade blocs setting tariffs on imports from other trade blocs and then trade bloc enlargement would affect the Nash equilibrium tariffs as in Sinclair and Vines (1994). However, the assumptions of segmented markets and constant marginal cost allows the export markets in the rest of the world to be analysed independently of the markets in the industrialised countries. The total welfare effect of trade bloc enlargement would include the effect of tariff-setting in the industrialised markets as well as the effect of export-subsidy-setting in the rest of the world. Combining these two effects is beyond the scope of this paper. 'The assumption that there is a singlefirmin each country is not restrictive since a country with n firms is equivalent to a trade bloc of n countries each with a singlefirm;hence, the results would continue to hold if there were n firms in each country.

5 508 BILATERALISM IS GOOD countries in each trade bloc; thus M = N/B. Since the model is concerned with strategic interaction between trade blocs there must be at least two trade blocs, and obviously the number of trade blocs cannot exceed the number of countries; hence 2 < B < N and 1 < M < N/2. Each trade bloc is assumed to use an export subsidy to maximise the welfare of the countries in the trade bloc given the export subsidies set by the other trade blocs. With the assumption of symmetry, the model can be analysed by considering a representative trade bloc and treating the other (B - 1) trade blocs as one aggregate trade bloc with (N - M) countries. The representative trade bloc sets an export subsidy e t and its firms each export output x h while the other (B-l) trade blocs set an export subsidy e and their firms each export output x. The price of the exported good in the rest of the world is given by the linear inverse demand function P = a-0x (1) where X Mx t + (N M)x is total sales in the export market. The profits of a firm in the representative trade bloc and the profits of a firm in the other (B 1) trade blocs are, respectively 7T, = (P - C + e^xf n=(p-c + e)x (2) At a Cournot equilibrium the firms independently and simultaneously choose their outputs to maximise profits given the export subsidies set by the trade blocs. Since demand is linear there exists a unique Cournot equilibrium. Assuming that allfirmsexport positive quantities, thefirstorder conditions for a Cournot equilibrium are dni/dx, = P + x,p' -c + e t = 0 &n/dx = P + 5cP'-c + e = 0 (3) Using the inverse demand function (1) and noting that X = Mx t + (N - M)x, these first order conditions can be solved to yield explicit solutions for the Cournot equilibrium quantities x, = {a - c + (N - M + \)e t - (N - M)e}/{N + l)/3 (M+l)e}/(n+l)P (4) The output of a firm in the representative trade bloc is increasing in the export subsidy given by therepresentative trade bloc and decreasing in the export subsidy given by the other trade blocs. The output of a firm in the other trade blocs is decreasing in the export subsidy given by the representative trade bloc and increasing in the export subsidy given by the other trade blocs. Substituting the Cournot equilibrium quantities (4) into the inverse demand function (1) and subtracting marginal cost c yields the price-cost margin P-c = {a-c-me,-{n-m)e}/{n+l) (5)

6 D. COLLIE 509 The price-cost margin of all firms is decreasing in the export subsidies given by the representative trade bloc and all the other trade blocs. Since all output is exported to the rest of the world, the welfare of a country in the representative trade bloc is given by the profits of its firm less the cost of the export subsidy payments. To capture the fact that the government revenue to pay the export subsidy will typically be raised by distortionary taxation, the opportunity cost of government revenue will be allowed to exceed unity as in Neary (1994). Hence, the welfare of a country in the representative trade bloc is given by: W ( iti Xe t x ( = (P c)x t (A \)e t Xi where A is the opportunity cost of government revenue and the term (A 1 )e<jc/ represents the deadweight loss due to the distortionary taxation, which will be zero if lump-sum taxes are feasible, A = I. 7 Thus, using (4) and (5), the welfare of a country in the representative trade bloc is W, = [a-c-{{\- l)(n + 1) + M}e, - (N - M)e) x{a-c+(n-m+ l)et - (N - M)e}/(N + \) 2 0 (6) An export subsidy given by the representative trade bloc has three effects on the welfare of a country in the representative trade bloc. Firstly, the positive profit-shifting effect: the export subsidy shifts profits to the firms in the representative trade bloc by increasing their output and reducing the output of firms in the other trade blocs. This profit-shifting effect is a purely beggar-my-neighbour effect since the firms in the representative trade bloc gain market share at the expense of firms in the other trade blocs. Secondly, the negative terms of trade effect: the export subsidy worsens the terms of trade since it increases total output and hence reduces the price-cost margin of firms in the representative trade bloc. Thirdly, the negative distortionary taxation effect: the deadweight loss from taxation will increase when the export subsidy is financed by distortionary taxation, A > 1. An export subsidy given by other trade blocs has a negative effect on the welfare of a country in the representative trade bloc since it reduces the output and the price-cost margin of firms in the representative trade bloc. 3. Nash equilibrium in export subsidies In this section trade blocs are assumed to behave in a non-cooperative manner when setting export subsidies which leads to an export subsidy war that can be modelled as the Nash equilibrium of the single-shot export subsidy game. Having obtained an expression for the welfare of a country in the representative trade bloc in the previous section, it can now be used to derive the Nash equilibrium of the export subsidy game. Then, the effect of 7 Alternatively, A may reflect purely distributional considerations or the fact that the domestic firm U partly owned by foreign shareholders; see Neary (1994, p. 200).

7 510 BILATERALISM IS GOOD trade bloc enlargement on the Nash equilibrium export subsidies and the welfare of the industrialised countries can be analysed. In the Nash equilibrium in export subsidies, each trade bloc sets its export subsidy to maximise the welfare of the countries in the trade bloc given the export subsidies set by the other trade blocs. 8 Since all countries are identical there will be no conflict over income distribution within the trade bloc when setting the optimal export subsidy. The representative trade bloc maximises (6) with respect to e t which yields the first order condition l \E{a-c-(N-M)e}-Ge i ]=0 (7) de, (AT+1) 2 /? where E = 2(N - M + 1) - X(N + 1) and G = 2{N-M+ 1){(A - l)(n + 1) + M} > 0 As the model is symmetric it has a symmetric Nash equilibrium where all trade blocs set the same export subsidy e N, where the superscript N denotes the Nash equilibrium. Substituting e t = e e* 1 into (7) and solving for the Nash equilibrium export subsidy yields where e N = (a - c)e/a (8) A = (N - M)E + G = X(N+l) + (N-M+ \){\{N + 1) - 2} > 0 Clearly, the Nash equilibrium export subsidy is positive if E is positive, and this will be the case if the opportunity cost of government revenue is less than the critical value X E = 2(N - M + l)/(n + 1) > 1. This critical value, shown in Fig. 1, always exceeds unity for M ^ N/2 and is linearly decreasing in the number of countries in each trade bloc. Thus, the Nash equilibrium export subsidy is more likely to be positive, the lower is the opportunity cost of government revenue and the fewer countries there are in each trade bloc. 9 There are export subsidies in the Nash equilibrium in regions A and B of Fig. 1 and export taxes in region C. When lump-sum taxes are feasible, A = 1, the Nash equilibrium export subsidy is always positive. 10 Since the purpose of this paper is to analyse profit-shifting export subsidies, it will be assumed henceforth that A < A* (E > 0) so that trade blocs set positive export subsidies in the Nash equilibrium. To see how the Nash equilibrium export subsidy varies with the size of the trade bloc differentiate (8) with s Obviously, if export subsidies are set by the individual countries rather than the trade blocs then trade bloc formation will have no effect on the Nash equilibrium export subsidies. 9 This result is consistent with the optimal export subsidy under oligopoly in Dixit (1984) where lamp-sum taxes are implicitly assumed and with the optimal export subsidy under duopoly in Neary (1994) where distortionary taxation is assumed. 10 Obviously, if there was only one trade bloc consisting of the N industrialised countries then its optimal policy would be an export tax to improve the terms of trade. This would maximise the welfare of the N industrialised exporting countries.

8 D. COLLIE 511 7N tf + \ \ Opportunity cost of government revenue >l \ (^ C: Export taxes y+2 i j A: Export subsidies ^v... *,. \ B-Export subsidies e" >0. -jj^<0 \ l L M Number of countriesra eich trade bloc M Fio. 1. Critical values of the opportunity cost of government revenue respect to the number of countries in each trade bloc M With A < A, the Nash equilibrium export subsidy is positive and decreasing in the number of countries in each trade bloc. 11 This leads to the following proposition: Proposition 1 Trade bloc enlargement (an increase in the number of countries in each symmetric trade bloc) leads to a reduction in the Nash equilibrium export subsidy. The increase in the number of countries in each trade bloc affects the Nash equilibrium export subsidy set by the representative trade bloc in two ways: Firstly, the profit-shifting effect of the export subsidy is reduced since the export subsidy has less effect on the output of the firms in the representative trade bloc; dx t /de t = (N - M + \)/(N + \)0 from (4). This is because there (9) "As the number of countries in each trade bloc increases, it becomes more likely that there will be export taxes in the Nash equilibrium and these export taxes would be increasing in the number of countries in each trade bloc.

9 512 BILATERALISM IS GOOD are fewer firms outside the trade bloc who will reduce output in response to the export subsidy and more firms inside the trade bloc who will increase output. Secondly, tfie terms of trade effect of the export subsidy is increased since the export subsidy has a larger effect on price; dp/de t = -M/(N + 1) from (5). This is because there are more firms in the trade bloc being given the export subsidy, and hence the price reduction will be larger. Both of these effects reduce the size of the export subsidy set by the representative trade bloc, and since the model is symmetric these effects will be common to all the trade blocs. Thus, the Nash equilibrium export subsidies set by the trade blocs will decrease as the number of countries in each trade bloc increases. As trade blocs become larger and their share of the world market increases, the incentive to use profitshifting export subsidies is reduced. 12 To evaluate how Nash equilibrium welfare of the industrialised (exporting) countries varies with the size of the trade blocs, substitute e t = e = ^ into (6) which yields W N = \ 2 {a - c) 2 G/2(3A 2 (10) Differentiating Nash equilibrium welfare, W N, with respect to the number of countries in each trade bloc, M, yields ^ * 00 This is clearly positive which leads to the following proposition: Proposition 2 Trade bloc enlargement (an increase in the number of countries in each symmetric trade bloc) leads to an increase in the Nash equilibrium welfare of all the industrialised countries. As the number of countries in each trade bloc increases the effectiveness of export subsidies is reduced, and the Nash equilibrium export subsidies set by trade blocs are reduced. Since the model is symmetric there is a reduction of all export subsidies by an equal amount which consequently has no effect on the market shares of firms, but there is an increase in the market price as all the export subsidies are reduced. The welfare of all the industrialised countries is increased since there is an improvement in the terms of trade but no change in the market share of each country's firm. Obviously, the higher price will worsen the terms of trade of the importing countries and reduce their welfare. Assuming that preferences are quasi-linear, the welfare of the importing countries in the rest of the world is given by consumer surplus V = f (a - 0Z) dz - PX = 0X 2 /2 (12) Jo 12 Thi* is the opposite of what happens with the conventional optimum tariffs in Knigman (1991).

10 D. COLLIE 513 Substituting e t e = e^ into (4) and using X Mx ( + (N - M)x yields the welfare of the rest of the world in the Nash equilibrium in export subsidies Differentiating with respect to the number of countries in each trade bloc yields 8V S _-A 3 (a-c) 2 -N 2 (N+l)(N-M+l)<0 (14) DM As the number of countries in each trade bloc increases and Nash equilibrium export subsidies are reduced, the terms of trade of the importing countries are worsened and their welfare falls. To evaluate world-wide efficiency, define world welfare in the Nash equilibrium as the sum of consumer surplus in the rest of the world and the welfare of the N industrialised countries: SI? = V N + NW N. Thus, using (11) and (14), the effect of an increase in the number of countries in each trade bloc on world welfare is The sign of this expression depends upon the term in square brackets and this will be positive if the opportunity cost of government revenue exceeds the critical value A n = N(N -M+ 1)/{N(N - M + 1) - M) > 1 where X n < X E for M < N/2. This critical value, shown in Fig. 1, is greater than unity and increasing in the number of countries in each trade bloc. In region A of Fig. 1 there are export subsidies in the Nash equilibrium and world welfare is increasing in the number of countries in each trade bloc; in region B there are export subsidies and world welfare is decreasing; and in region C there are export taxes and world welfare is increasing. This leads to the following proposition: Proposition 3 Trade bloc enlargement (an increase in the number of countries in each symmetric trade bloc) leads to a reduction in consumer surplus in the rest of the world but will increase world welfare if A ^ A n. When export subsidies are financed by lump-sum taxes, A = 1, trade bloc enlargement leads to lower Nash equilibrium export subsidies which reduces world welfare since the output of the oligopolistic industry contracts increasing the deadweight loss from oligopoly. If export subsidies are financed by distortionary taxation then there is a reduction in the deadweight loss from taxation as export subsidies are reduced as well as the increase in the deadweight loss from oligopoly. When the opportunity cost of government revenue is sufficiently large, A > A n, the reduction in the deadweight loss from taxation outweighs the increase in the deadweight loss from oligopoly and the reduction in export subsidies increases world welfare. Thus, distortionary taxation may (15)

11 514 BILATERALISM IS GOOD provide a rationale based upon worldwide efficiency for the multilateral WTO agreement that prohibits export subsidies. Rather than assume Cournot competition, it could be assumed that all the firms in each trade bloc collude so that there is effectively only one firm in each trade bloc. Then, in contrast to proposition one, trade bloc enlargement leads to an increase in the Nash equilibrium export subsidies until there are only a few trade blocs when they start to decrease. However, as in Proposition 2, trade bloc enlargement results in an increase in the welfare of the industrialised countries since total industry output contracts despite the higher export subsidies. Hence, there is a reduction in the welfare of the importing countries and also world welfare since export subsidies increase as industry output contracts. 4. The sustainabillty of free trade In the single-shot game of the previous section, free trade is not sustainable and the only possible outcome is the Nash equilibrium in which all trade blocs give export subsidies to their firms. To avoid such subsidy wars, in the case of export credits, the industrialised countries of the OECD have reached an agreement which aims 'to provide the institutional framework for an orderly export credit market and thus to prevent an export credit race in which exporting countries compete on the basis of who grants the most favourable financing'. 13 In game theory, it is well known that if a game is infinitely repeated then more co-operative outcomes can be sustained. The perfect folk theorem states that, if agents are sufficiently patient, then any individually rational outcome can be a perfect equilibrium of an infinitely repeated game. Since the purpose of the OECD agreement is to limit export subsidies rather than attempting to maximise the welfare of the OECD countries by imposing export taxes, it seems reasonable to look at free trade as a possible co-operative outcome. Collie (1993) analyses the sustainabilty of free trade in the Brander and Spencer (1985) model of profit-shifting export subsidies using the trigger strategy suggested by Friedman (1971). This section will look at possibility of sustaining free trade as the perfect equilibrium of an infinitely repeated version of the single-shot game analysed in the previous section using trigger strategies as in Collie (1993). The strategy of each trade bloc is to play free trade until any other trade bloc deviates by using an export subsidy, and thereafter to revert to using the Nash equilibrium export subsidy. If all trade blocs pursue a policy of free trade, e/ = e = 0, then from (6) the welfare of a country in the representative trade bloc is W F = (a-c) 2 /l3(n+l) 2 (16) 13 Quoted from the OECD (1990, p. 7) which details the Arrangement on Guidelines for Officially Supported Export Credits. Most OECD countries participate in this Arrangement which came into force in 1978 and sets minimum interest rates for export credits thus limiting the implicit subsidy. Participants are allowed to match any deviations from the Arrangement by other countries.

12 D. COLLIE 515 where the superscript F denotes free trade. Given the symmetry of the model, welfare is higher under free trade than in the Nash equilibrium as can be seen by subtracting (10) from (16) which yields 0 (17) where H X(N + 1) 2 > 0. This is clearly positive since the assumption that the Nash equilibrium export subsidies are positive implies that E is positive. 14 When all other trade blocs pursue a policy of free trade, the representative trade bloc will have an incentive to deviate from free trade and unilaterally use a profit-shifting export subsidy. The optimal unilateral export subsidy is given by maximising welfare (6), with e = 0, which yields the first order condition ^P = T- [E(a - c) - Ge,\ =0 (18) de ( (N+\) 2 0 [ K J ' Hence, solving for the optimal export subsidy when the representative trade bloc unilaterally deviates from free trade yields e = (a - c)e/g > 0 (19) where the superscript D denotes deviation from free trade. Differentiating (19) it can be shown that the optimal unilateral export subsidy for the trade bloc is decreasing in the number of countries in each trade bloc, de D /dm < 0, since export subsidies become less effective as the number of countries in each trade bloc becomes larger. Substituting the optimal export subsidy into welfare (6) yields the welfare of a country in the representative trade bloc when it deviates from free trade W D = X 2 (a - c) 2 /20G (20) Differentiating (20) it can be shown that the welfare of a country in a trade bloc that deviates from free trade is decreasing in the number of countries in each trade bloc, dw D /8M < 0. Obviously, welfare is higher when the representative trade bloc deviates from free trade than under free trade, as can be seen by subtracting (16) from (20). Also, welfare is higher when the representative trade bloc deviates from free trade than in the Nash equilibrium, as can be seen by subtracting (10) from (20). These subtractions yield 2p(N+l) 2 G 14 If there were export taxes in the Nash equilibrium then welfare in the Nash equilibrium would be higher than welfare under free trade so attempting to sustain free trade would be pointless and impossible.

13 516 BILATERALISM IS GOOD Number of countriej in each trade bloc M Fio. 2. Incentives to deviate from free trade Once the representative trade bloc has deviated from free trade, the other trade blocs will retaliate by setting the Nash equilibrium export subsidy, e", forever afterwards. Faced with the Nash equilibrium export subsidies being set by the other trade blocs, the optimal response of therepresentativetrade bloc is to use its Nash equilibrium export subsidy, e N, and hence the welfare of a country in the representative trade bloc will be as in the Nash equilibrium, W N, forever afterwards. Figure 2 shows W F, W N, and W D a& functions of the number of countries in each trade bloc. Assuming that the other trade blocs use the trigger strategy, if the representative trade bloc deviates from free trade by using an export subsidy e then the welfare of a country in the representative trade bloc in that period will be W D, but thereafter the other trade blocs will retaliate and the welfare of a country in the representative trade bloc will be W N forever afterwards. Whereas, if it does not deviate and pursues a policy of free trade then the welfare of a country in the representative trade bloc will be W F in every period. Therefore, using the trigger strategy will be an optimal response for the representative trade bloc to the trigger strategy of the other trade blocs if the present discounted value of the welfare of a country in the representative trade bloc from free trade exceeds that from deviation and the resulting retaliation

14 D. COLLIE 517 -L-W F^W +-^-zw N =>(W D -W r )^-^-JW F -W N ) (22) 1 o 1 o 1 6 where 6 is the discount factor which is between zero and one, 0 < 5 < 1, and will be assumed to be the same for all trade blocs. As the model is symmetric, all trade blocs are in the same situation. The trigger strategy will be optimal for all trade blocs if the discount factor exceeds the critical value, 6*, where the instantaneous gain from unilaterally using an export subsidy is just equal to the present discounted value of the losses from retaliation. Using (21) in (22), the critical discount factor is W D -W F EA 2 (23) where tf = {N-M)(N+l) 2 (A + G) > 0. The critical discount factor is clearly positive, but by subtracting the critical discount factor from unity it can be shown that 0 (24) Hence, the critical discount factor is between zero and one, 0 < S* < 1. Free trade is sustainable as a perfect equilibrium of the infinitely repeated game if the discount factor of the trade blocs is greater than the critical value, 6^6*. Obviously, the lower is the critical discount factor then the easier it is to sustain free trade. To see how the critical discount factor varies as the number of countries in each trade bloc changes, differentiate (23) with respect to M which yields &?_ = A(rt + 1) ^ _ M) {2GH + 2(N-M)A + E} + HA(A + G)}<0 As the denominator is clearly positive, the sign of the derivative depends upon the terms in square brackets which are all positive; hence, the derivative is negative which leads to the following proposition: Proposition 4 Trade bloc enlargement (an increase in the number of countries in each symmetric trade bloc) leads to a reduction in the critical discount factor making it easier to sustain free trade. Looking at Fig. 2 it can be seen that trade bloc enlargement has two effects on the critical discount factor firstly, the instantaneous gain from unilaterally using an export subsidy is reduced since the welfare of a country in a trade bloc that deviates from free trade is decreasing in the number of countries in each trade bloc, dw D /dm < 0. This will reduce the incentive to deviate from free trade and make the critical discount factor smaller. Secondly, the losses from retaliation are reduced since Nash equilibrium welfare is increasing in the number of firms in each trade bloc, dw N /dm > 0. This will increase the (25)

15 518 BILATERALISM IS GOOD incentive to deviate from free trade and make the critical discount factor larger. It turns out that the first effect dominates the second effect so that the incentives to deviate from free trade become smaller as trade blocs become larger; hence, the critical discount factor becomes smaller and it becomes easier to sustain free trade. 15 Rather than an agreement to sustain free trade, the trade blocs could agree to set export taxes to maximise their joint welfare. Provided the discount factor was sufficiently high, such an agreement could be sustained by the threat of retaliation with the Nash equilibrium export subsidies. Obviously, it will be more difficult to sustain the joint welfare maximising export taxes rather than free trade as the critical discount factor will be higher. As above, it can be shown that trade bloc enlargement leads to a reduction in the critical discount factor making it easier to sustain the joint welfare maximising export taxes. 5. Conclusions This paper has analysed the effect of exogenous trade bloc enlargement in a multi-country extension of the Brander and Spencer (1985) model of profitshifting export subsidies. In the single-shot export subsidy game, trade bloc enlargement was shown to reduce the effectiveness of export subsidies with the result'that trade blocs set lower export subsidies in the Nash equilibrium. The lower export subsidies set by the trade blocs increased the welfare of the industrialised (exporting) countries while reducing the welfare of the importing countries. Surprisingly, it was shown that trade bloc enlargement may result in an increase in world welfare if export subsidies are financed by distortionary taxation. In the infinitely repeated game, trade bloc enlargement was shown to reduce the critical discount factor making it easier to sustain free trade (or the joint welfare maximising export taxes). Thus, in contrast to Krugman (1991), bilateralism is good for the industrialised countries and may be good for world welfare. In conventional models where trade policy is used to improve the terms of trade, such as Krugman (1991), trade bloc enlargement increases the market power of trade blocs thereby increasing the incentive to use trade policy; whereas in this strategic trade policy model, trade bloc enlargement reduces the incentive to use profit-shifting trade policy. Thus, trade bloc enlargement provides an example where the new trade theory yields the opposite conclusion to that obtained from conventional trade theory. This analysis also suggests that small trade blocs (or countries) will have the greatest incentive to use strategic trade policy whereas large trade blocs (or countries) will have the greatest incentive to use trade policy to improve their terms of trade. This implies that small countries, such as Korea and Taiwan, are more likely to use strategic trade policy than a large country like the United States. "This is the opposite of what happens in Bond and Syropoulos (1995) in a conventional model where trade blocs use optimum tariffs to improve their terms of trade.

16 D. COLLIE 519 This model also indirectly casts some light on multilateral agreements to limit or prohibit export subsidies. It was shown that reducing export subsidies will increase the welfare of the industrialised countries and this obviously provides a rationale for the OECD agreement to limit export credits given by the industrialised countries on exports to the developing countries. However, intuition suggests that reducing export subsidies in an oligopolistic industry will increase the deadweight loss from oligopoly and reduce world welfare since the gain to the exporting countries is outweighed by the loss to the importing countries; hence, it is argued that strategic trade policy does not provide a rationale for worldwide agreements to prohibit export subsidies. In this model, it was shown that reducing export subsidies may increase world welfare if export subsidies are financed by distortionary taxation since then the increase in the deadweight loss from oligopoly may be outweighed by the reduction in the deadweight loss from taxation. Thus, distortionary taxation provides a rationale based upon worldwide efficiency for the multilateral WTO agreement that prohibits export subsidies. ACKNOWLEDGEMENTS Earlier versions of this paper were presented to seminars at Cardiff Business School and the University of Warwick, and at the 1996 European Economic Association Congress in Istanbul. I would like to thank the participants for their comments. It has been greatly improved by the insightful comments and suggestions made by two anonymous referees for which I am extremely grateful. I am solely responsible for any errors that remain. REFERENCES BAGWELL, K. and STAIGER, R. W. (1994). 'Multilateral Tariff Co-operation During the Formation of Customs Unions', NBER Working Paper No. 4543, Cambridge, MA. BAGWELL, K. and STAIGER, R. W. (1996). 'Strategic Export Subsidies and Reciprocal Trade Agreements: the Natural Monopoly Case', NBER Working Paper No. 5574, Cambridge, MA. BRANDER, J. A. and SPENCER, B. J. (1984). Tariff Protection and Imperfect Competition', in H. Kicrzkowski (ed.), Monopolistic Competition and International Trade, Oxford University Press, Oxford. BRANDER, J. A. and SPENCER, B. J. (1985). 'Export Subsidies and International Market Share Rivalry', Journal of International Economics, 18, BOND, E. W. and SYROPOULOS, C. (1995). Trading Blocs and the Sustainabflity of Inter-Regional Co-operation', in M. Canzoneri, W. Ethier, and V. Grilli (eds), The New Transatlantic Economy, Cambridge University Press, Cambridge. BOND, E. W. and SYROPOULOS, C. (1996). The Size of Trading Blocs: Market Power and World Welfare Effects', Journal of International Economics, 40, COLLIE, D. R. (1993). 'Profit-Shifting Export Subsidies and the Sustainability of Free Trade', Scottish Journal of Political Economy, 40, DDCTT, A. K. (1984). 'International Trade Policy for Oligopolistic Industries', Economic Journal, 94 (supplement), FRIEDMAN, J. W. (1971). 'A Non-Cooperative Equilibrium for Supergames', Review of Economic Studies, 38, KRUGMAN, P. R. (1991). 'Is Bilateralism BadT in E. Helpman and A. Razin (eds), International Trade and Trade Policy, MIT Press, Cambridge, MA.

17 520 BILATERALISM IS GOOD KJLUOHAN, P. R. (1993). 'Regionalism versus Multilateralism: Analytical Notes' in J. Dc Melo and A. Panagariya (eds), New Dimensions in Regional Integration, Cambridge University Press, Cambridge. NBARY, J. P. (1994). 'Cost Asymmetries in International Subsidy Games: Should Governments Help Winners or Losers?*, Journal of International Economics, 37, OECD (1990). The Export Credit Financing Systems in OECD Member Countries, Fourth Edition, Organisation for Economic Cooperation and Development, Paris. RAY, J. E. (1995). Managing Official Export Credits: the Quest for a Global Regime, Institute for International Economics, Washington, DC. SINCLAIR, P. J. N. and VINES, D. (1994). 'Do Fewer, Larger Trade Blocs Imply Greater Protection?: The good news and bad news about regional trading blocs'. International Economics Study Group Meeting, London.