tO PS9 - 23 7Y/ POLICY RESEARCH WORKING PAPER 2374 Should Credit Be Given for As each new round of multilateral trade negotiations Autonomous Liberalization approaches, there is a in M ultilateral Tra e 1demand for a negotiating rule that would give credit for Negotiations? previous unilateral liberalization. The feasibility and desirability of such a rule Aaditya Mattoo depend on when it is Marcelo Olarreaga instituted. The World Bank Development Research Group Trade June 2000a POLICY RESEARCH WORKING PAPER 2374 Summary findings As each new round of multilateral trade negotiations ante assurance that any unilateral liberalization will approaches, there is a demand for a negotiating rule that receive credit in the next round. would give credit for autonomous (unilateral) Such a rule would help induce or enhance liberalization. Mattoo and Olarreaga show that the liberalization in some countries between negotiating feasibility and desirability of such a rule depend on when rounds by reducing the gains from retaining protection it is instituted. as negotiating currency. A credit rule established at the beginning of a round of More strikingly, it could also lead to deeper levels of negotiations has a primarily distributional effect, multilateral liberalization and induce other countries to favoring those who have already undertaken go further than they would in the absence of a rule. liberalization. Implementing such a rule would depend Most important, sucli an ex ante rule would not rely on the generosity of those who have not liberalized. on altruism to be generally acceptable. The authors propose instead establishing a credit rule at the end of a round of negotiations, which creates an ex This paper - a product of Trade, Development Research Group - is part of a larger effort in the group to improve trade policy in goods and services. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Lili Tabada, room MC3-333, telephone 202-473-6896, fax 202-522-1159, email address ltabada@worldbank.org. Policy Research Working Papers are also posted on the Web at www.worldbank.org/research/ workingpapers. The authors may be contacted at amattoo@worldhank.org or molarreaga@worldbank.org. June 2000. (32 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development issues. An objective of the series is to get the findings out quickly, evenif the presentations are less than fully polished. The papers carry the names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countries they represent. Produced by the Policy Research Dissemination Center Should Credit be Given for Autonomous Liberalization in Multilateral Trade Negotiations?* Aaditya MattooI Marcelo Olarreaga" JEL classification: F02, F13, F15 Keywords: WTO, credit for unilateral liberalization. * We would like to thank Simon Evenett, Michael Finger, Carsten Fink, Bernard Hoekmnan, Will Martin, Costas Michalopoulos, Arvind Subramanian , Anthony Venables and participants in a Trade Seminar at the World Bank for very helpful comments and suggestions. We are also grateful to Lili Tabada for excellent assistance. 1Development Research Group (DECRG), World Bank, 1818 H Street, NW, Washington DC 20433; email: amattoo@worldbank.org. " DECRG, World Bank and CEPR, London: email: molarreaga@worldbank.org. Non technical summary As a new round of multilateral negotiations approaches, there is a strong demand for a negotiating rule which would give credit for autonomous liberalization. Developing countries in particular have argued that there should be a permanent mechanism for granting such credit within the WTO. In fact, the Heads of the International Monetary Fund, World Bank and World Trade Organization also expressed interest in the possibility of developing countries obtaining negotiating credit for trade policy reforms introduced under Fund or Bank programmes. However, such demands have produced no significant result. Might it be possible to articulate the proposal in a way that makes it more generally acceptable? We study the simplest form of granting credit for autonomous liberalization: this is to proceed with negotiations as if the unilateral liberalization has not happened; so any agreed reductions are calculated from the pre-liberalization levels of protection rather than the current applied levels. We consider a country in which an internal political change has created an impulse to liberalize, and examine the impact of four different credit rules that have been proposed: i) no-credit rule; ii) an ex-post credit rule, which grants credit for past unilateral liberalization; iii) an ex-ante credit rule, which assures credit for future unilateral liberalization; and iv) a rule that requires immediate reciprocal tariff reductions by other countries in response to a unilateral reduction by any country. The desirable feature of an ex-ante credit rule is that it favourably influences the unilateral decision to liberalize. It does so in three ways. First, the unilateral tariff reduction will be larger than in the absence of such a rule, because the inter-temporally maximising government will see its future terms of trade loss associated with the unilateral liberalization being neutralised by the negotiating credit it will receive. Second, as a consequence of the first reason, certain unilateral tariff reductions that would not have occurred in the absence of this rule, will now be observed. Third, this may induce unilateral tariff reductions by the rest of the world as a reaction to the original unilateral liberalization. 1 The drawback of any credit rule is that the tariff reduction by the unilaterally liberalizing countries during the multilateral negotiations will be smaller than it would have been under a no-credit rule. How do these opposing considerations affect the inter-temporal tariff schedule of a country that unilaterally liberalizes? The ex-post credit rule is dominated by the ex-ante credit rule, as the latter induces lower protection in the period between negotiating rounds though both lead to identical protection in the negotiating round. The no-credit rule also dominates the ex-post credit rule in that it leads to lower protection in the negotiations though both have no effect on the level of protection chosen in the period between rounds. There is no unambiguous dominance between the no-credit rule and the ex-ante credit rule. The ex-ante credit rule leads to lower protection than the no-credit rule in the period between rounds, but higher protection in the round. Under plausible conditions, for which there is some empirical evidence, an ex-ante credit rule may lead to lower inter-temporal average protection. In a more general framework, where the extent of multilateral liberalization is allowed to depend on the nature of the credit rule, one aspect of the results does not change: both credit rules shift the distribution of protection at the end of multilateral negotiations in favour of those who have unilaterally liberalized. More interestingly, the proportional tariff reduction in the multilateral negotiations is larger in the presence of a credit rule. This, together with the fact that an ex-ante credit rule also induces greater unilateral liberalization in the period between rounds, implies that such a rule is likely to lead to lower intertemporal average levels of protection in both countries. Hence, while the attractiveness of the ex-post credit rule depends entirely on the desirability of its distributive consequences, the ex-ante credit rule has the additional virtue of inducing greater inter-temporal liberalization. A crucial question is whether these rules are feasible. This depends on how the country that does not unilaterally liberalize views the alternative protection profiles of its trading 2 partner. A credit rule that grants automatic reciprocal concessions to unilaterally liberalizing countries is not sustainable because it can inflict significant costs on the other countries that are obliged to reduce their protection - and which may end up worse off than before the original unilateral liberalization. An ex-post credit rule would be resisted by those who have not liberalised unilaterally, because of its adverse distributional consequences. They would be obliged to liberalize more to obtain the same reductions in applied tariffs by their trading partners than in the absence of the rule. A credit rule instituted ex-ante is the only one that may be acceptable to all countries for non-altruistic reasons but it would require an external enforcement mechanism. Ironically, the benefits it provides to countries through inducing greater unilateral liberalization by their trading partners, may make the former content with the status -quo and unwilling to participate in a compensating round of multilateral negotiations. The inability to make a credible commitment to participate in future negotiations on unfavourable terms, may thus deprive all countries of the benefits of an ex-ante credit rule. In this context, the dispute settlement procedures of the WTO may provide a valuable external enforcement mechanism. 3 "We believe that the value of autonomous trade liberalization initiatives should be recognized in the WTO negotiating process. These initiatives contribute to the expansion of world trade, and convey real benefits not only to the countries taking them but also to their trading partners. This should be clearly acknowledged in the forthcoming WTO negotiations, by crediting countries which bind their autonomous trade liberalization under WTO rules." Joint statement by the Heads of the IMF, World Bank and WTO, Seattle, December 1999 1. Introduction As a new round of multilateral trade negotiations approaches, there is a strong demand for a negotiating rule which would give credit for autonomous liberalization. This paper shows that the desirability and feasibility of such a rule depends on when it is instituted. A credit rule instituted at the beginning of a round of negotiations has primarily a distributional effect, favouring those who have already undertaken liberalization. Such a rule is predictably opposed by those who have not liberalized. We propose instead the establishment of a credit rule at the end of a round of a negotiations, which creates an ex- ante assurance that any unilateral liberalization will receive credit in the next round. Such a rule would help induce and/or enhance liberalization between negotiating rounds by reducing the gains from retaining protection as negotiating currency - and leads to lower inter-temporal protection under plausible conditions. Crucially, such an ex-ante rule may be acceptable to all countries. Not surprisingly, the demand for creating a rule to grant credit for autonomous liberalization measures assumes political visibility and support just before or during a round of negotiations.' During the Uruguay Round, discussions on this issue took place in both the Negotiating Group on the Functioning of the GATT System and the ' The possibility of granting credit for unilateral liberalization was first raised by Michalopoulos (1985) and has been linked to the issue of "coherence" between World Bank's effort towards openness in developing countries and the trade negotiations in the WTO. 4 Negotiating Group on Market Access. In the Mid-Term Review of the Uruguay Round, Ministers agreed that participants would receive appropriate recognition for the liberalization measures which they had adopted since 1 June 1986 - i.e. the beginning of the Round. The effect of this decision on the negotiated outcome in industrial goods and services is not easy to estimate, but the consequences for the agriculture negotiations were striking.3 Several countries that had liberalized recently insisted on and secured the choice of a pre-liberalization base period for calculating cuts in protective measures, with the result that there was little change in actual protection in many areas (see Ingco, 1995). Developing countries in particular have argued that a long-term solution should be found for crediting autonomous liberalization measures within the WTO.4 In fact, it was reported that the Heads of the International Monetary Fund, World Bank and World Trade Organization had also expressed interest in the possibility of developing countries obtaining negotiating credit for trade policy reforms introduced under Fund or Bank programmes.5 However, this interest has not as yet manifested itself in the establishment of any enforceable ex-ante rules in this respect, with the exception of a nebulous commitment in the General Agreement on Trade in Services.6 Rather, we see the expression of demands for credit once again at the beginning of a new round of negotiations. 2See "Credit and Recognition for Autonomous Liberalization Measures", WTO Committee on Trade and Development, WT/COMTD/W/4, 29 May 1995. All WTO documents can be accessed electronically at http://www.wto.org/online/ddf.htm. 3 See Finger, et al. (1997) and Fung and Ng (1998) for institutional and empirical details on these issues. 4 One of the clearest articulations of this view is to be found in a "non-paper" submitted by Mexico at the end of 1990, which was co-sponsored by nineteen developing countries. The need for longer term approach is recognized in the Guidelines of the Chairman of the Negotiating Group on Market Access (MTN.GNG/MA/W/13, 19 December 1991) on (a) Credit for tariff bindings and liberalization of NTMs; (b) Recognition for autonomous liberalization measures. 5 See "Ways to Achieve Greater Coherence in Global Policy-Making through Strengthened GATT Relationships with other Relevant International Organizations", paragraph 32, document MTN.GNGING14/W/35, 20 September 1989. 6GATS Article XIX:3 states that "For each round.. .negotiating guidelines shall establish modalities for the treatment of liberalization undertaken autonomously by Members since previous negotiations..." So there exists a prior commitment to take into account unilateral liberalization, but it is not clear how this is to be implemented in concrete terms. More importantly, the development of a clear rule is postponed to the beginning of each round. 5 We consider the simplest form of granting credit for autonomous liberalization: this is to proceed with negotiations as if the unilateral liberalization has not happened; so any agreed reductions are calculated from the pre-liberalization levels of protection rather than the current applied levels. Four alternatives are explored: (i) no-credit is given; (ii) a rule for giving credit is instituted at the beginning of a (new) round of negotiations ; (iii) a rule for giving credit is established at the end of a (past) round of negotiations, and (iv) a rule is set that requires automatic reciprocal tariff concessions to the country that unilaterally liberalizes. Section 2 describes the basic model in which these rules are examined. Section 3 contains the body of the analysis and argues in favour of an ex-ante credit rule (the case of the automatic rule is examined in the appendix). Two sets of simplifying assumptions are made in section 3: that the extent of multilateral trade liberalization and the credit rule are exogenously given, and that the probability of a unilateral tariff reduction being reversed is zero. The implications of relaxing the first set of assumptions are discussed in Section 4, which also demonstrates the incentive problems with each rule. Section 5 examines the implications of relaxing the second assumption and discusses the distinction between applied rates of protection and legally bound rates in this context. Section 6 concludes. 2. A Simple Model Assume a two-country two-good world, where country A imports good 1 and country B imports good 2.7 As in standard endogenous protection theory, the govemrnment's objective function, denoted V, is given by a combination of political-economy factors and social welfare:8 vi(t. ;t2) = P'(t,;t2)+ a'W(t,;t2) (1) 7 Given the two-country assumption, multilateral also implies bilateral. 8 See Hillnan (1981), Grossman and Helpman (1994, 1996). 6 where superscript i = A, B refers to countries, P reflects the contribution of political- economy factors and W the contribution of social welfare to government's objective function; the relative weight of social welfare with respect to political-economy considerations in the govermnent's objective function is given by a; t, is the tariff (or trade restriction) that country A imposes on imports of good 1 and t2 iS the tariff that country B imposes on imports of good 2. For simplicity, we limit the trade instrument to tariffs, but the analysis can easily be extended to non-tariff barriers. Again following the literature we will assume that (i) political-economy considerations always call for a higher domestic tariff, whereas (ii) welfare considerations call for lower tariffs (at the political optimum level).9 More formally: - A>0 and aw <0 ; ap > 0 and aW < 0 (2) At, At, at2 at2 It is convenient to assume that there are three time periods: an initial (past) period T-1 in which a round of negotiations takes place, an intermediate (present) period T in which the government decides on whether to reduce protection unilaterally after a change in its preferences for economic efficiency, and a final (future) period T+1 in which another round of negotiations takes place (motivated by the earlier change in preferences which generates potential gains from multilateral cooperation). Again, for simplicity, we assume that governments do not discount the future (i.e. I dollar today is the same as 1 dollar tomorrow).'0 In period T-1, government i's intertemporal objective function is denoted ' ,. Had the two governments set tariffs non-cooperatively, denoted t* and t, are obtained by solving simultaneously each government first order condition for maximization of its inter-temporal objective: 9 Note that if (i) is true, (ii) needs to be satisfied for the problem in (2) to have an interior solution with a non-negative tariff. '0Assuming a more traditional discount factor between 0.90-0.95 will not modify our qualitative results. 7 t* = argmax 'T A and t4 = argmax T (4) t tB A When governments set tariffs independently, each seeks to influence the terms of trade in its favour. The tariffs set non-cooperatively are inefficient because each government inflicts a negative terms of trade externality on the other. One may challenge the importance of terms-of-trade effects for small developing countries, where credit for unilateral liberalization is probably more relevant. However, there is a significant literature that argues that terms-of-trade are also important in the case of "small" countries. As an example, Argentina and Brazil jointly have less than 1 percent of the world import market, but in some categories of the 6-digit Harmonized Systems of trade classification their import share is above 30 percent. " I As convincingly argued by Bagwell and Staiger (1999) and Levy (1999), in the presence of negative externalities associated with terms-of-trade effects, there will be gains associated with reciprocal tariff reduction accomplished through multilateral trade negotiations. In such negotiations, countries exchange "concessions" in the form of mutually reduced protection. The "I will give you improved access to my market if you give me improved access to your market" form of negotiations can be seen as a way of neutralizing the adverse terms of trade effect (Bagwell and Staiger, 1999).12 Assume that a previous round of negotiations in period T-1, say the Uruguay Round, led to a set of tariffs (4';'t"), such that t,' < t4 and tu 0, will call for a new optimal non-cooperative tariff, t', that is smaller than t;. This is due to the fact that 82WA/aa at, <0 , given that aWA/at, <0 by equation (2). However, it is impossible to say a priori whether t4 t4', then it is clear that a tariff reduction will not occur. A tariff increase is of course ruled out due to the fact that tl' is binding. This ambiguity is shown in Figure 1, where R', is the initial inter-temporal reaction function of country A before the change in preference for economic efficiency and R. is country B's inter-temporal reaction function.'9 Reaction functions after the change in country A's preference for economic efficiency are denoted R and R Aia, e, according to whether the change in preferences is large or small. Figure 1 is drawn so that if the change in preferences is small, then the optimal tariff for country A is higher than t,'. Since it is not legally possible for country A to increase its tariff above the bound level, country A's tariff in period T will remain as its T-l level, t4'. On the other hand if the change in preference is large, then a unilateral tariff reduction will take place and the new tariff will be t' at period T. To summarise, an increase in preference for economic 19 Without loss of generality we assume here for exposition purposes that tariffs are strategic complements, i.e., country's A optimal reaction to an increase in B's tariff is to increase its own tariff. The same arguments could be made assuming strategic substitutes. 11 efficiency may or may not lead to, a unilateral liberalization, depending on the size of this change. 3.2 A rule for crediting unilateral liberalization established ex-post Since the rule is instituted ex-post, i.e. in period T+1, it does not influence country A's decision in period T (see section 3.3 below). Hence, the unilateral choice of tariff in period Twill be the same as the one described in the absence of credit rule in the previous section. 3.3 A rule for crediting unilateral liberalization established ex-ante An ex-ante credit rule implies that when country A makes its decision in period T on whether it should liberalize, it recognizes that any adverse terms of trade movement will not be forever but only for a single time period, since it will be remedied in period T+1. Indeed, country A will know that in the multilateral round it will only have to cut tariffs from its T-1 levels. Therefore, the dampening effect on the incentive for unilateral liberalization is reduced. There are at least three reasons why this ex-ante credit rule for unilateral liberalization may be desirable. First, because the terms-of-trade loss is neutralised in future periods, the optimal tariff reduction that will occur will be larger, leading to a smaller t, . Second, as a consequence of the first, there is a greater likelihood that t' < t", and that countries will liberalize sooner as their preference for free-trade increases. Third, if the unilateral liberalization in A is sufficiently large, liberalization may also be provoked in the rest-of- the-world where there has been no change in government's preference towards free-trade. 12 3.3.] Ex-ante credit leads to larger-unilateral liberalization in period T In the presence of a credit rule the optimal tariff, denoted tN , is such that: tNT= a av av/ V (5') tA atj gp w atA atA And comparing (5) and (5') it is clear that tN < tN as aVTi/lp w apw /lat' > 0 (the terms of trade rationale for high tariffs disappears in period T+1). Thus, introducing the credit rule will lead to a larger tariff reduction. In terms of Figure 1, the change in country A's preferences for economic efficiency in period T will lead to a larger shift of country A's reaction function towards the left, which in turn will implicitly lead to a larger tariff reduction by country A (if t I tu in the absence of the rule, may observe a reversal of this inequality. In terms of Figure 1, this is equivalent to a shift of country A's reaction function from R',4,,, to R4Age, when comparing changes in the reaction function in the presence and absence of the ex-ante credit rule, respectively. Thus unilateral liberalization that would not have taken place in the absence of the credit rule, will be observed. 3.3.3 An ex-ante credit rule may induce unilateral liberalization in the rest-of-the world This follows from the assumption that tariffs are strategic complements, i.e., the best reaction of country A to a tariff increase in country B is to increase its own tariff in the non-cooperative equilibria. If the immediate tariff reduction in country A, due to a move 13 in preferences towards economic efficiency, is sufficiently large, then it may induce country B to also reduce its own tariffs. In the absence of the credit rule country B will be facing the potential adverse terms of trade movement and it is not clear a priori that it will liberalize. Given the effect discussed above, a tariff reduction in country B is more likely and will be larger than it would had been in the absence of the credit rule.20 Such a reduction would not happen in Figure I as the unilaterally optimal tariff for country B in that figure remains above the bound level, tu . However as shown in Figure 2, if the shift in country A's reaction function is sufficiently large, this will induce country B to unilaterally decrease its tariffs, which in turn will give incentives to country A to further reduce its tariff and so on, until the new non-cooperative equilibrium is reached (point "*" in Figure 2). Thus, credit for unilateral liberalization not only brings earlier and larger liberalization in the country that experiences a move in preferences towards economic efficiency, but it may also lead to earlier and larger liberalization in the rest-of-the-world. 20 This relies on the assumption of strategic complements. Had we assumed strategic substitutes, then country B will be willing to increase its tariffs; but this is impossible given that t4' is binding. Thus the assumption is not crucial for the "weak" result: the opposite can never happen. 14 3.4 A comparison of tariff profiles under alternative credit rules Assume that the round of negotiations in period T+1 will lead to a percentage multilateral tariff reduction of (1 - a), which is exogenously given. 21 Then, as previously discussed, imagine that the government of a particular country faces an increase in its preference for economic efficiency and decides to unilaterally liberalize between rounds. Figure 3 shows the profile of tariffs through time under three different types of rules. The black columns represent the evolution of tariffs under a no-credit rule; the grey column under an ex-post credit rule and the white column under the ex-ante credit rule. If the objective is to achieve a greater extent of trade liberalization, then as shown by Figure 3, the ex-post credit rule is dominated by the ex-ante credit rule, as the latter induces lower protection in the period between negotiating rounds, tv < tl, though both lead to identical protection in the negotiating round. The no-credit rule is also superior to the ex-post credit rule in that it leads to lower protection in the negotiations, a tN < t, , though both have no effect on the level of protection chosen in the period between rounds. It is impossible to determine a priori whether the no-credit rule leads to higher inter- temporal protection than the ex-ante credit rule. The ex-ante credit rule leads to lower protection than the no-credit rule in the period between rounds, but higher protection in the round in period T+1. The question then, is whether the average inter-temporal level of protection is lower under an ex-ante credit rule. Let us first consider the case where the ex-ante credit rule has induced liberalization, whereas in the absence of the credit rule, there would have been no unilateral liberalization. It is then clear that the inter-temporal average tariffs is lower under the ex- 21 We will relax the assumption of ( - a) being exogenously given in section 4, where countries A and B will negotiate a multilateral deal that is incentive compatible for both countries (in section 3, incentive compatibility is assumed). 15 ante credit rule, i.e., t, + a t' < ' + a t'. In the case where the tariff reduction would have occurred in the absence of the ex-ante credit rule, as depicted in Figure 3, the inter- temporal tariff is smaller if tN* + a t4' < tN + a t' . This inequality can be rewritten as: - tN1 a N (6) Inequality (6) implies that the inter-temporal average tariff is smaller under the credit rule if the additional unilateral tariff reduction induced in period T by the ex-ante credit rule as a ratio of the reduction that would have taken place in its absence, is larger than the proportionate reduction in tariffs agreed during the negotiations in period T+1. The likelihood of inequality (6) being satisfied in the real world is an empirical question and depends in the importance of terms-of-trade effects in the determination of tariffs. A study by Olarreaga, Soloaga and Winters (1999) has shown that even in the case of 6"small" countries such as Brazil and Argentina with a share of world trade below 1 percent, "terms-of-trade" effects can explain up to 30 percent of the tariff variation.22 This in turn implies that for inequality (6) to hold a < 30/70 = .43. Rieealling that in the Uruguay Round, an average tariff reduction between 25 and 33 percent was achieved (see Finger et al, 1999), it seems reasonable to assume that the inter-temporal average tariff will be lower after the introduction of an ex-ante credit rule. 4. Endogenous multilateral tariff reduction and credit rules We have so far taken the extent of multilateral liberalization as given, and also examined each rule as if it were exogenously given. Consider now the implications of relaxing these assumptions. 22 Note that for larger countries the terms-of-trade effects may represent a larger share of the tariff variation. Note that in the case where the terms-of-trade effect represent 50 percent of the tariff variation inequality (6) is always satisfied as the right-handvside goes to 1. 16 What is the outcome of multilateral negotiations likely to be? It seems plausible to assume that the negotiated outcome will be efficient and lie on the contract curve, which is the locus of points of tangency of the iso-value curves. 23 We continue to assume that there are equiproportionate reductions in protection from an initial level that depends on the choice of credit rule.24 These two assumptions together yield a determinate outcome to the negotiations. Geometrically, the outcome is the point where a line joining the initial protection pair to the origin (representing equi-proportionate reductions in protection) intersects the contract curve (see Figure 4). The tariff pair (till; t41) can be seen as lying on the contract curve (CC) that corresponded to the governments' objective function in period T-1. Note that under reasonable assumptions, the contract curve necessarily has a negative slope.25 The change in preferences in country A causes a shift in A's isovalue curves and hence in the contract curve.26 Since the change in preferences (i.e. the weight a) is in favour of greater efficiency, the new contract curve, C'C', will lie closer to the origin, implying that there is scope for mutually beneficial reduction in levels of protection.27 The precise outcome on the contract curve will depend on the credit rule: a no-credit rule will result in an outcome denoted by point X, whereas a full-credit rule will produce point Y as outcome. It is evident that a credit rule favours the country that has unilaterally liberalized, at the expense of the one that has not. More interestingly, the proportional tariff reduction in the multilateral negotiations, a, will be larger in the presence of a credit rule than without one. To see this, note that in the 23 As Bagwell and Staiger (1999) have demonstrated, when countries are not Symmetric, the mutual benefits from reciprocal liberalization may terminate before the efficiency locus is remcbed. Our assumption would hold if the bargaining solution could be supported by side-payments. 24 This allow us to abstract from the problems associated 'with imbalances in bargaining power as in Maggi (1999). 25 This can be shown by deriving the contract curve as in Bagwell and Staiger (1999), solving for the tariff of country A and then taking the derivative with respect to the tariff of country B. 26 There is no change in the position of the contract curve between period T (when A's preferences change) and period T+ I when the negotiations take place. 17 presence of a credit rule, a = (t2' - t} )/t = as, whereas in the absence of a credit rule, a = (tu 2 )/t2 'aN(. And a' > a N, given that ty t[ - f> tl - tI ). How are the results of the previous section affected? There is a greater likelihood that an ex-ante credit rule will lead to lower intertemporal average levels of protection in both countries. In the case of country B, this is straightforward and unambiguous, since the proportional multilateral tariff reduction is larger in the presence of a credit rule and country B's tariff in period T is unaffected. In the case of country A, there is again an ambiguity, an ex-ante credit rule implies lower protection in period T and higher protection in period T+1 than with a no credit rule. But since there are higher proportional cuts in period T+1 with a credit rule, there is a greater likelihood of a lower intertemporal average. Thus, endogenizing the extent of agreed multilateral liberalization only strengthens the presumption in favour of an ex-ante credit rule. We have assumed so far that credit rules are exogenously given and imposed on WTO members. If the type of rule were not given, which one would be endogenously chosen by WTO members? To address this question, we use Figure 5 which presents a plausible set of B's iso-value contours passing through the different pairs of equilibrium tariffs identified in Figure 4. Consider first a situation with no ex-ante rule, in which A has already liberalized, so that at the beginning of period T+1 we are at point N in Figure 5. Negotiations based on granting A full-credit would imply a decline in B's value function (from B2 to B.), which 27 Again, this can be proved following Bagwell and Staiger (1999) derivation of the contract curve and then showing that after an increase in "a", the new contract curve will imply a lower tariff for country A for any given tariff of country B. 18 as drawn in Figure 5, could bring country B below its pre-unilateral liberalization level (i.e., Bo