Is There an Optimal Non-Zero Level of Protectionism? Zagros Madjd-Sadjadi Department of Economics University of the West Indies, Mona & Abdullahi Abdulkadri Department of Economics University of the West Indies, Mona JEL Codes: C7, F10, P16 Abstract: Game theory is used to model actions of countries engaged in trade and examines equilibrium outcomes resulting when each country is faced with the choice of engaging in bilateral free trade or tariff, or unilateral free trade or tariff. The likely outcomes are evaluated under the different scenarios that the players are absolute gains maximizers or relative gains maximizers. The Nash equilibria for these strategic games are then evaluated to determine if they are Pareto superior to other outcomes of the game. Various modifications of this basic model were proposed and analyzed. First we allowed the two countries to allocate different weights to payoffs from being an absolute gains maximizer or relative gains maximizer. We then extended the model to include four trading countries instead of two. The Nash equilibrium of free trade obtained for the base model remained robust for most of the various modifications introduced into the model, although with restrictions on the payoff parameters in some instances. Some of these modifications include adding a negotiation cost for free trade, introducing sequential moves by countries and incomplete information. Thus the model provides testable hypotheses to assist in determining why defection from free trade occurs in the real world. *Canadian Economics Association Annual Meeting, June 4, 2004, Ryerson Polytechnic, Toronto. Please contact the corresponding author at zagros.madjdsadjadi@uwimona.edu.jm Introduction Often times each discipline carries with it only part of the answer to a question that can best be seen as analogous to a jigsaw puzzle. Neoclassical economists look around the world and may wonder why there isn’t more free trade while political scientists, especially of the Realist school of international relations, may wonder why nations trade at all. After all, think political scientists, is not security important and are not nations interested in maximizing power, ensuring their viability as a separate sovereign power by ensuring that they are self-sufficient? At the same time, wonder economists, is not it obvious that the gains from trade are just as evident at the international level as they are at the household level and when did anyone argue that households should be self-sufficient? Yet both disciplines suffer from the acute disease of myopia. Economists often ignore the role that power plays in formulating government policy by assuming that countries seek 1 absolute gains in trade. Such a naïve interpretation cannot explain why countries do not engage 2 in total free trade and may leave one believing that state actors are not utility maximizers. It also makes states little more than entities concerned with the economic well-being of its citizens. Yet countries are not like firms. While stockholders of a corporation can freely dump the shares of a company that is not doing well, citizens of a country cannot change citizenships with the same modicum of time or expense. At the same time, while firm takeovers are part and parcel of the corporate world and their relative merits are determined based upon the benefit to the shareholder, country takeovers are almost always viewed as hostile no matter what the stated intentions of the aggressor. Indeed, it is a staple of political science that countries seek to 1 Though they may find some comfort that there is a group of theorists within international relations called neoliberals who generally take the same perspective. Robert O. Keohane (1988) is the leading exponent of this perspective. 2 Of course, whenever utilities are interdependent, the first welfare theorem may not hold. Thus competitive markets may not bring about Pareto efficient allocations. Examples of this in economics include externalities and public goods. In a sense, free trade agreements are at least partially public goods because they often provide rules of origin that allow for non-partner countries to benefit as well. They, therefore, contribute externalities and if they provide positive externalities, they can be underprovided in the competitive marketplace. In addition, total free trade may not occur because of asymmetries in market power between countries. Most economists, however, treat this, as a special case. Indeed, the classic theory of free trade is almost as close to a “free lunch” as economics can get. 2 maximize power to one extent or another in their dealings with other nations. The reality is that states probably are concerned with both power and plenty. 3 On the other hand, many scholars in political science argue that relative gains 4 maximizers will not engage in free trade. After all, think such realist and neo-realist scholars, cooperation between states that view each other as competitors is inherently nonsensical 5 because every interaction is a zero-sum game. Yet around the world countries appear to be increasingly trading sovereignty for economic and social well-being through the promotion of international organizations and agreements that transfer sovereignty to third parties such as the World Trade Organization, the World Court, the United Nations, the European Community, the Caribbean Community and NAFTA’s trilateral dispute panels. Do Relative Gains Maximizers Cooperate? Although there have been many attempts to model power, the easiest is to build a model of relative gains.6 Under conditions of relative gains, the utility of each country is interdependent. This paper will demonstrate that even relative gains maximizers will find it in their interest to engage in cooperation under certain specified criteria. It begins with the standard economic assumptions about free trade and then progressively alters them. When the gains from trade are equally distributed and decisions are made simultaneously, then the countries can be expected to 3 See Viner (1948) for a viewpoint as to why the accumulation of wealth and power are complementary. 4 The terms “realist” and “neo-realist” are used here to describe the so-called “realist” and “neorealist” schools of international relations scholars. A representative realist is Robert Gilpin (1977), while the primary proponent of neo-realism (also called structural realism) is Kenneth Waltz (1979). The key difference is that realists tend to assume power maximization while neorealists argue that countries seek to balance power. Realists and neo-realists tend to believe that either interpretation would transform cooperation into a zero-sum (or, at best, constant sum) game. 5 For example, Greico (1988) argues that positional concerns dominant international politics so cooperation can be achieved only for a very narrow range of objectives. However, this position does not take into consideration the relative gains that can be made at the expense of those who are not engaged in cooperation. On the other hand, Snidal (1991) uses an iterated prisoner’s dilemma approach to demonstrate that relative gains maximization is not a hindrance to cooperation even in zero-sum situations when the number of countries is large. 6This is not to say that these gains are monetary ones. As Kirshner (1997) has noted, there are “non-economic” reasons for actions, such as security concerns. Later in this note, there will be an allowance for the fact that there may be both absolute and relative components of utility in a free trade regime. 3 be playing the following “game”, where all the assumptions listed follow directly from the neoclassical argument for free trade: A>0 (assumption that free trade provides absolute gains over no free trade) B>0 (assumption that unilateral free trade policy provides absolute gains) A > B > 0 (assumption that total free trade provides more gains than unilateral free trade) B>C (assumption that unilateral free trade is preferred to unilateral protectionism—note that unilateral protectionism may be preferred to bilateral protectionism or vice versa) ABSOLUTE GAINS FREE TRADE TARIFFS FREE TRADE (A,A) (C,B) RELATIVE GAINS TARIFFS (B,C) (0,0) FREE TRADE (0,0) (C-B,B-C) ABSOLUTE GAINS FREE TRADE TARIFFS FREE TRADE (+,+)* (+/-,+) TARIFFS (B-C,C-B) (0,0) RELATIVE GAINS TARIFFS (+,+/-) (0,0) FREE TRADE (0,0)* (-,+) TARIFFS (+,-) (0,0) *Nash Equilibrium Numerical Examples: ABSOLUTE GAINS FREE TRADE TARIFFS FREE TRADE (10,10)* (2,5) RELATIVE GAINS TARIFFS (5,2) (0,0) FREE TRADE (0,0)* (-3,3) TARIFFS (3,-3) (0,0) *Nash Equilibrium Notice that even under relative gains, the general trend is towards free trade. If both countries currently have free trade, neither will choose to move away from this position because to do so will engender a minimum of a one period loss unless the other side also chooses tariffs. 7 To do so would result in less “gain” than a free trade position. On the other hand, there is incentive for countries using tariffs to move to free trade given that they would gain at least a one period advantage (unless the other side also moved to free trade) and would not be penalized at all if both sides moved to free trade. 7 It does not matter at this point if the game is finite or infinite. 4 Yet such a scenario is simply not realistic. Countries are interested in both relative and absolute gains8 and the restrictive assumptions about the relative rankings of options make free trade a foregone conclusion. In liberalizing these assumptions, suppose that the relative and absolute gains division depends on a factor, q, which represents the weight given to absolute gains. Then 1-q will be the weight given to relative gains. This leads to the following: 1 >= q >= 0 Gains from free trade with the other side having free trade: qA + (1-q)0 = qA > 0 Gains from trade under free trade with other side having protectionism: (q)B + (1-q)(B-C) = qB + B - qB + qC - C = B + qC - C = B-(1-q)C > 0 Gains from trade under protectionism with other side having free trade: (q)C + (1-q)(C-B) = qC + C - qC - B + qB = C + qB - B = C-(1-q)B ABSOLUTE/RELATIVE GAINS FREE TRADE TARIFFS FREE TRADE (qA,qA) (C-[1-q]B,B-[1-q]C) TARIFFS (B-[1-q]C,C-[1-q]B) (0,0) Now if we drop the assumptions about the benefits or costs of free trade and the relative 9 rankings, we find that a Nash equilibrium free trade regime exists for all A > (C-[1-q]B)/q where q ≠ 0 and q ≠ 1. This equilibrium is unique under the additional condition that B-[1-q]C > 0. Or equivalently, if solving for q: qA > C-(1-q)B qA > C-B+qB q(A-B) > C-B B-C+Cq > 0 Cq > C-B Where 0 ≤ q ≤ 1 : q > (C-B)/(A-B) where (A-B) ≠ 0 and q > (C-B)/C where C ≠ 0 8 See Powell (1991) for a discussion on why both relative and absolute gains are important to states. 9 If we fail to relax these assumptions we find that countries that value both relative and absolute gains generate same results as before with total free trade being the only pure Nash solution. One important aspect of relaxing these conditions is that any possible rank order is now possible. As such, this represents the general case and no normative implications are present. 5 This generates the necessary and sufficient conditions for Pure Nash equilibria in free trade regimes. A Pure Nash equilibrium free trade regime exists for all relative gains maximizers (q = 0) where B > C (unilateral free trade is preferred to protectionism). A Pure Nash equilibrium free trade regime exists for all absolute gains maximizers (q = 1) where A > C (universal free trade is preferred to unilateral protectionism) and this equilibrium is unique where B > C. Negotiating Costs The next modification is to add a discounted payoff over perpetuity, along with a 10 negotiating cost that is paid up front, n. Given a known interest rate, r, then the discounted payoff of a perpetuity is always benefit/r. Under this condition, free trade is not always an optimal policy. Under non-Nash equilibrium free trade conditions, the present value payoff from free trade must be greater than the present value payoff from defection. To this end, the payoff determination depends crucially on what the other country is doing (engaging in free trade or protectionism): If the other country is engaging in protectionism, then the benefit is from unilateral free trade: free trade benefit = (B-[1-q]C)/r free trade “cost” = n (negotiating cost) So if (B-[1-q]C)/r > n, the country will implement free trade. This solution depends on the interest rate, so interest rates must be “low enough” in order to have free trade. The “free trade” interest rate will be: r < (B-[1-q]C)/n 10 This negotiating cost could be the cost of negotiating a free trade agreement or it could be the cost of negotiating side agreements with domestic interests. One question that may arise is why the benefit is discounted but not the cost. The way to solve this is to think of whether or not you should invest $1000 or spend it. The cost is paid up front but the benefit is generated far into the future. In some cases, the benefits from free trade may not be immediately realizable but instead accrue only in the future. In fact, it may also be the case that free trade actually lowers utility for the country in the short run (the 30,000 people who lose their jobs in sugar manufacturing make a lot more noise politically than the masses who just saw the cost of their sugar decrease by 2 cents per pound, even though the economic benefit of cost reduction outweighs the economic cost. 6 This means that if a country is a relative gains maximizer (q =0), we get r < B-C/n. If the country is an absolute gains maximizer (q= 1), we get r < B/n. On the other hand, if the other country is committed to free trade, the gain from trade is between the benefit from bilateral free trade and unilateral protectionism: Free trade benefit = (qA – (C-[1-q]B))/r Free trade “cost” = n (negotiating cost) So if qA/r – (C-[1-q]B) > n then the country will implement free trade. This solution depends on the interest rate so interest rates must be “low enough” in order to have free trade. The ‘free trade’ interest rate will be: qA > r(n + C-[1-q]B) r < qA/((n + C-[1-q]B)) At the limit, if the other country is committed to free trade: where q = 0, then r < 0 and where q = 1, then r < A/(n + C) Thus if the gains from trade are equally divided and the other country is committed to free trade, relative gains maximizers will institute free trade only when real interest rates are negative, while absolute gains maximizers will institute free trade only when they are sufficient low, i.e., the gains from trade dominate over the cost associated with the transition. An important implication is that even absolute gains maximizers will not implement free trade when real interest rates are “too high.” Unequal Terms of Trade Now the assumption will be that the gains from trade differ between countries, such that the second country’s gains are multiplied by a factor k to arrive at the first country’s gains. Furthermore, the preference for absolute versus relative gains may differ between countries such that country 1 has a factor q but country 2 has a factor p. This results in the following: ABSOLUTE GAINS FREE TRADE TARIFFS FREE TRADE (kA,A) (kC,B) RELATIVE GAINS TARIFFS (kB,C) (0,0) FREE TRADE (kA-A,A-kA) (kC-B,B-kC) 7 TARIFFS (kB-C,C-kB) (0,0) ABSOLUTE/RELATIVE GAINS Free Trade (country 1) under free trade (country 2)= q(kA)+(1-q)(kA-A)= qkA+kA-qkA-A+qA = kA+qA-A = (k+q-1)A Free Trade (country 2) under free trade (country 1)= p(A)+(1-p)(A-kA) = pA+A-pA+pkA-kA = A+pkA-kA = (pk+1-k)A Free Trade (country 1) under non-free trade (country 2) = q(kB)+(1-q)(kB-C) = qkB + kB - C -qkB + qC = kB-C+qC Non-Free Trade (country 2) under free trade (country 1) = p(C)+(1-p)(C-kB) = pC+C-kB-pC+pkB = C+pkB-kB Non-Free Trade (country 1) under free trade (country 2) = q(kC)+(1-q)(kC-B) = qkC+kC-B-qkC+qB = kC-B+qB Free Trade (country 2) under Non-Free Trade (country 1) = p(B)+(1-p)(B-kC) = pB+B-kC-pB+pkC = B-kC+pkC Non-Free Trade (country 1) under non-free trade (country 2) = 0 Non-Free Trade (country 2) under non-free trade (country 1) = 0 FREE TRADE TARIFFS FREE TRADE ([k+q-1]A,[pk+1-k]A) (kC-B+qB,B-kC+pkC) TARIFFS (kB-C+qC,C-kB+pkB) (0,0) In this case, free trade will be a unique Nash equilibrium only when when: kA+ (q-1)A > kC + (q-1)B kB + (q-1)C > 0 B + (p-1)kC > 0 and A + (p-1)kA > C + (p-1)kB This will be true when: k >((1-q)A)/(A-C) k > ((1-q)C)/B k > B/(1-p)C and k > (C-A)/(A+(1-p)B) Using the first country as the test case: For absolute gains maximizers (q=1), this means that free trade will be preferred whenever: k > 0 (there are gains to trade) For relative gains maximizers (q=0), free trade is preferred whenever: k> A/(A-C) and k > C/B For such countries, assuming the gains from trade are positive (k>0), the more they receive gains from trade relative to the other country, the more likely that they will accept free trade. 8 Thus the terms of trade matter to the relative gains maximizer while they do not to the absolute gains maximizer. But, more critically, the terms not only have to be in favour of the relative gains maximizer but even if the terms of trade are favourable, the relative gains maximizer may not always institute free trade. For example, the gains from free trade must be greater by at least the factor k than any perceived gains under unilateral protectionism. Similarly, the gains from free trade must be greater than the ratio of the gains from unilateral protectionism divided by the gains 11 from unilateral free trade. Of course, if the gains from unilateral protectionism are negative and the gains from unilateral free trade are positive, relative gains maximizers will institute free trade as well unless they are not receiving “enough” of the terms of trade. In such a case, a relative gains maximizer will still not institute free trade unless k> A/(A-C), meaning that the less the harm from engaging in protectionism, the greater the terms of trade must tend towards equality for the relative gains maximizer to engage in trade. In other words, enemies will only trade if the terms of trade are relatively balanced but allies will trade much more frequently. Thus, the United States, which has little perceived benefit from trade with Cuba would require a trade agreement that provides it with very close to an even split in the terms of trade (or a favourable terms of trade agreement) but it would not require such an advantage vis a vis Canada, where the desire for relative gains is probably much less. Similarly, Cuba will likely demand a very even split in the terms of trade vis a vis the USA but not with Canada. This helps to explain why the United States is the predominant country imposing trade sanctions against “enemies” and vice versa because the relative gains from trade matter to these countries. Free Trade Agreements Now, an assumption will be made that there are four countries in the world and two of them are considering a free trade agreement. A further supposition will be made that there is an equilization of trade among all of the countries. Now what happens? Well, in this case, each of the absolute gains will be divided in half and the relative gains, instead of being simply the 11 It should be pointed out there can be gains from unilateral protectionism. After all, if one country is engaged in free trade then both countries can benefit, although the benefit may not be as great as if both countries had free trade policies. All this means is unilateral protectionism is preferred to bilateral protectionism (tariffs by both parties). 9 subtraction of the other side’s gains from the current gain will also account for the absolute gain against the other two non-participating countries, according to the following formula: relative gains = gains by country 1 - gains by country 2 + absolute gains by country 1 x 2 (for the relative gains against the other two countries)/3 (the number of other countries). ABSOLUTE GAINS FTA NO FTA FTA (5,5) (-5,2.5) NO FTA (2.5,-5) (0,0) RELATIVE GAINS FTA (3 1/3,3 1/3) (-10 5/6, 9 1/6) NO FTA (9 1/6, -10 5/6) (0,0) In this case, under both absolute and relative gains, the Free Trade Agreement is beneficial and produces a Nash equilibrium. However, while absolute gains and losses are halved in such a situation, the relative gains payoffs will actually increase for those in the free trade agreement while the benefits from unilateral free trade and unilateral protectionism will decrease. Thus, if gains from trade are no longer divided equally but are instead distributed unequally, there will be greater cooperation as the number of states increases. Thus, while countries outside of the free trade zone will always want in, countries in the free trade zone may not always want them to join. Instead, the number of countries allowed to join the free trade zone will increase until a maximum is reached where the benefits from cooperation with some partners are exactly balanced by the benefits from exclusion of others. In general, where A would be the total gain generated if all countries were free traders and where there are a total of n countries and with p countries being the number outside the free trade zone and assuming that all countries are equal in size and that there are equally divided gains from trade, the countries inside the free trade zone will allow other countries to join as long as: A((n-p+1)/n)((p-1)/n-1) > A(n-p)/n)(p/n-1) This reduces to: p > (n-1)/2 Thus, even relative gains maximizers will seek free trade up to a certain optimal point when the payoffs are evenly distributed among countries. Of course, if there are great differences in payoffs between countries, cooperation may still not occur. This does not mean that worldwide agreements will fail. To the contrary, if countries cannot exclude other countries 10 from joining (such as the WTO), we might see a “snowball” effect once the threshold is reached. That is because the greater the number of countries that are party to a cooperative agreement, the greater the value of that agreement for those who are not part of it in a realist world. Indeed one of the great ironies of the realist argument is that in seeking relative gains, countries will also not want to be put in a relative disadvantage. Thus, international cooperation may actually increase as the number of countries already committed to a policy increases. This effect will be seen only under conditions of relative gains, not absolute gains, because under the former countries care about the relative positioning of themselves vis a vis other countries while under absolute gains, they do not. Returning to the case of differences existing between the preferences for relative and absolute gains by country but including the idea that each country has the same payoff functions, the following results are generated: ABSOLUTE/RELATIVE GAINS FREE TRADE CURRENT SYSTEM FREE TRADE (qA,pA) (C-[1-q]B,B-[1-p]C) CURRENT SYSTEM (B-[1-q]C,C-[1-p]B) (0,0) A Nash equilibrium free trade solution exists for only a subset of the results in the first absolute/relative gains maximization problem. Thus, if preferences differ, a Nash equilibrium free trade solution exists in all cases where: Where 0 ≤ q ≤ 1 : q > (C-B)/(A-B) where (A-B) ≠ 0 and q > (C-B)/C where C ≠ 0 and Where 0 ≤ p ≤ 1 : p > (C-B)/(A-B) where (A-B) ≠ 0 and p > (C-B)/C where C ≠ 0 If one side is a relative gains maximizer (q = 0) and the other side is an absolute gains maximizer (p = 1), Nash equilibrium free trade regimes exist if and only if B > C and A > C, and the equilibrium is unique. 11 What if the gains from trade are not equally distributed? In that case, we have the following game: ABSOLUTE GAINS FREE TRADE TARIFFS FREE TRADE (a1, a2) (c1,b2) RELATIVE GAINS TARIFFS (b1,c2) (0,0) FREE TRADE TARIFFS (a1-a2, a2-a1) (b1-c2,c2-b1) (c1-b2,b2-c1) (0,0) For the absolute gains maximizer, we still have the case where if the gains from total free trade dominate, the country will choose free trade. However, things have gotten a little more complicated for the relative gains maximizer. In this case, the more the relative differences between the two countries, the less likely that they will agree to free trade. This is because for a Nash equilibrium free trade regime to exist, it must be true that: a1-a2 > c1-b2 and a2-a1 > c2-b1 Rearranging terms, this implies that: a1- c1 > a2-b2 and a2-c2 > a1-b1 Therefore, the gains that each country makes from total free trade over unilaterial protectionism must be greater than the gains that the other country gets from going from unilateral free trade to total free trade. However, this does not guarantee a unique Nash equilibrium. A unique Nash equilibrium also requires the following condition be true: b1 - c2 > 0 and b2 - c1 > 0 Therefore, b1 > c2 and b2 > c1 . This implies that the gains from unilateral free trade for each country must be greater than the gains from unilateral protectionism for the other country. Now, going back to our original equations, we know that, for a free trade regime to exist: a1 - a2 > c 1 – b2 a2 - a1 > c2 - b1 Suppose that the country that is relatively better off is country 1. Then a1 - a2 > 0 and a2 - a1 < 0. If a2 - a1 < 0 and a2 - a1 > c2 - b1, then c2 - b1 < 0. Therefore b1 > c2. So one of the conditions is true. Therefore, the only question is whether the other condition is true. In that case: 12 a1 - a2 > c 1 – b2 However, if c1 – b2 is greater than 0 (implying that c1 > b2 which means that the gains from unilateral protectionism for one country exceed the gains from unilateral free trade for the other country) we do not have a unique pure Nash Equilibrium. What about the mixed strategy case? In that case, the payoff for Row is c(a1 - a2) + (1-c) (b1c2) + c(c1 – b2)=ca1-ca2+b1-c2-cb1+cc2=cc1–cb2. Therefore: c(a1-a2-b1+c2-c1+b2)= c2-b1 Therefore: c = (c2 - b1)/(( a1 - a2) + (b2 - b1) + (c2 - c1)) This implies that we have a mixed strategy Nash equilibrium whenever: (c2 - b1) > 0 and (a1 - a2)- (c1 – b2) > 0 OR (c2 - b1) < 0 and (a1 - a2)- (c1 – b2) < 0 This implies that for column to have a mixed strategy, it must be that free trade is preferred to tariffs by row when column chooses free trade and that column receives more benefit than row when row has unilateral free trade OR if column receives less benefit than row when row has unilateral free trade, then tariffs must be preferred to free trade by row when column chooses free trade. Sequential Moves by Countries Until now, we have only considered cases in which countries take decisions on whether to engage in free trade or to impose tariff without knowing what their trading partner or competitor has decided to do. In reality, such decisions are made sequentially and the outcome of a trade negotiation may be markedly altered when one of the countries has committed, a priori, to either a free trade or to a tariff regime. There may also be a first-mover advantage or disadvantage, as the case may be, as one country realizes a higher (in the case of first-mover advantage) or lower 13 (in the case of first-mover disadvantage) payoff when it first commits to a regime regardless of what the other country does. For absolute gains maximizers, a sub-game perfect equilibrium with a bilateral free trade outcome will always be attained under the following conditions whenever country 1 moves first: a1>c1 a2>c2 and b2>0. These require that Country 1’s gains from bilateral free trade be higher than its gains from unilateral tariff (a1>c1), and country 2’s gains from bilateral free trade be higher than its gains from unilateral tariff (a2>c2), in addition to country 2 having strictly positive gains from unilateral free trade (b2>0). The gains to country 1 from unilateral free trade (b1) has no effect on this equilibrium whenever country one makes the first move. However, when we consider the case of relative gains maximizers, a more interesting result emerges. A sub-game perfect equilibrium with a bilateral free trade outcome will only exist under the following conditions whenever country 1 moves first: a1-a2 > c1-b2 a2-a1 > c2-b1 and b2-c1>0. These require that country 1’s gains from switching from unilateral tariff to bilateral free trade be higher than that of country 2’s gains from switching from unilateral free trade to bilateral free trade (a1-c1>a2-b2) and country 2’s gains from switching from unilateral tariff to bilateral free trade be higher than country 1’s gains from switching from unilateral free trade to bilateral free trade (a2-c2>a1-b1). Also, country 2‘s gains from unilateral free trade must be higher than country 1’s gains from unilateral tariff (b2>c1). The implication of these conditions for the sub-game perfect equilibrium bilateral free trade regime is that relative gains maximizers will always have 14 zero payoff whenever their gains from free trade are identical, otherwise one country derives a strictly positive payoff while the other derives a strictly negative payoff, although the country with a negative payoff will still be better off under bilateral free trade than any other outcome. Incomplete Information and International Cooperation The final modification adds incomplete information to the mix. We assume that countries can correctly estimate the gains they will receive but imperfectly estimate the gains made by others. In this case, the actual benefit is listed using lower case but the perceived benefits are listed using upper case: ABSOLUTE GAINS FREE TRADE TARIFFS RELATIVE GAINS FREE TRADE TARIFFS FREE TRADE (a1 , a2) (b1 , c2) (a1 - A2, a2 – A1) (b1 - C2, c2 – B1) TARIFFS (0,0) (c1 - B2, b2 – C1) (0,0) (c1 , b2) When will we institute free trade in this case? Now the results become even more cloudy. In general, there is a single unique Nash equilibrium of free trade when: a1 - A2 > c1 - B2 and b1 - C2 > 0 and b2 – C1 > 0 and a2 – A1 > c2 – B1 In this case, perceptions form reality. In cases where countries believe that they are gaining relative to other countries, they will institute free trade but when they believe that they are doing worse relative to other countries, they will not. This will only harm cooperation when countries overestimate the gains from trade accruing to others. Of course, as the number of countries increases, relative gains must be measured against both countries within and outside the free trade area. Conclusion The models provided in this article improve upon the existing literature by providing the general form solutions to the game-theoretic problem of relative gains maximizers and the ability to cooperate on the international level. It shows that when rank ordering of options is relaxed, relative gains maximization can result in a lack of free trade. Additionally, when there are transaction costs associated with implementing trade agreements, there may be little reason to 15 pursue free trade. Indeed, one could argue that as trade agreements multiply, the gains from concluding additional agreements will decrease and the costs will increase. This is because one would likely reach agreements on those issues that are most beneficial to both sides first and would defer negotiation on more difficult matters that will likely cost more in terms of negotiating costs. Thus, since the marginal cost of implementing trade agreements is rising as more trade agreements are reached and since the marginal benefit from those trade agreements is falling, this implies that there is some optimal level of protectionism that is not zero (and, by extension, an optimal level of free trade that is not complete). This should not be a surprise conclusion to economists who are used to seeing arguments in favour of optimal non-zero levels of pollution and crime. Bibliography Gilpin, Robert (1987). The Political Economy of International Relations. Princeton, NJ: Princeton University Press. Grieco, Joseph (1988) “Anarchy and the Limits of Cooperation: A Realist Critique of the Newest Liberal Institutionalism.” International Organization, 42: 485-507 Keohane, Robert O. (1988). International Institutions: Two Approaches,” International Studies Quarterly 32(4): 379-396. Kirshner, (1997). “The Microfoundations of Economic Sanctions,” Security Studies, 6(3), Spring 1997, pp. 32–64 Powell, Robert (1991) “Absolute and Relative Gains in International Relations Theory.” American Political Science Review 85: 1303-1320. Snidal, Duncan (1991). “International Cooperation Among Relative Gains Maximizers,” International Studies Quarterly 35: 387-402. Viner, Jacob (1948). “Power versus Plenty as Objectives of Foreign Policy,” World Politics 1(1), October: 1-29. Waltz, Kenneth (1979). Theory of International Politics. New York: McGraw-Hill. 16