ruutAincj <? or TlC^ tID28 .M414 ALFRED P. WORKING PAPER SLOAN SCHOOL OF MANAGEMENT /, QUOTAS AND THE STABILITY OF IMPLICIT COLLUSION by Julio J. Rotemberg and Garth Saloner" WP#1 778-86 May 1986 MASSACHUSETTS INSTITUTE OF TECHNOLOGY 50 MEMORIAL DRIVE CAMBRIDGE, MASSACHUSETTS 02139 '"quotas and the stability of implicit collusion by Julio J. Rotemberg and Garth Saloner-WP#1 778-86 May 1986 "Sloan School of Management and Department of Economics, M.l.T. respectively. We would like to thank Robert Gibbons and Paul Krugman for helpful comments and the National Science Foundation (grants SES-8209266 and IST-8510162 respectively) for financial support :? -^o, ^- i AUB 2 7 Ills'" 1986 | !o I Introduction This paper shows that the imposition of an import quota by one country can protection can reduce the lead to increased international competitiveness: price in the country that imposes the quota, the foreign country, or both. somewhat paradoxical result emerges from a model of implicit collusion will ensue if collusion that can be sustained Our strategic variables. variables Since a quota reduces the punish a deviating domestic firm, the amount ability of the foreign firms to We study both the case The more powerful the any firm deviates. threat, the more collusion that can be sustained. In such by the threat that a setting the firms in an industry sustain collusive prices more competitive pricing . in is This of correspondingly diminished. which sales and the case results are strongest when which prices are the in prices are the strategic quotas always make monopolization of the domestic market more : difficult in that case. This is in sharp contrast to the results in static imperfect competition models in which quotas tend to make domestic firm to act as a monopolist at home. It is it easier for the also different from the impact of tariffs which, in a dynamic setting like the one studied here, do not necessarily make Whether it tariffs more difficult for the firms to sustain collusive make coDusion more difficult to sustain or not outcomes. depends on the severity of the punishments that the firms can reasonably be expected to inflict on their cheating rivals. The maximal punishments developed by Abreu (1982) involve an outcome This in . can, willing to tolerate the ensuing losses, it if it makes result, is it is which the domestic firm earns because even with a very large zero profits of the style tariff, the foreign firm charge a price so low that impossible for the domestic firm to earn profits at home. tariffs do not affect the abUity of the duopoly to maintain As a monopolistic outcomes. There the maximum punishment the foreign firm Contrast this with a quota. can still inflict on the domestic firm is to sell its entire quota. Thus yields positive profits for the domestic firm. firm faces a lower punishment, monopolistic outcome. fact that, it as pointed out has a larger incentive to deviate from the by Davidson (1984), trade restrictions country to share of the domestic market. Thus the domestic favor and thereby giving also make more it In this case of maximal punishments, firm's incentive to abide to lose if it deviates. therefore, our results have the In his classic paper he showed that a single domestic producer who faced a competitive foreign a single domestic producer and than with a quota. tariff by by altering the market opposite implication of those of Bhagwati (1965). would act more competitively with a it the domestic country have a larger let the collusive scheme can at least be partiaUy restored its since the domestic This general intuition must be qualified somewhat by the less costly to the foreign shares in This generally a single foreign market When we consider producer, the opposite result emerges. If the punishments are not as grim as this, but instead involve reverting to the equilibrium from the one-shot game in each period, then large tariffs (which are anologous to small quotas) make collusion more difficult In that case, to sustain. then, the consequences of tariffs and quotas are similar. As mentioned previously, our results are strongest for the case prices are the strategic variables. When quantities are the strategic variables we we analyze the case in in which which find that only small quotas break the discipline of the oligopoly while large quotas, if anything, strengthen the incentives to go along with the monopolistic outcome. Thus our case parallel closely those of Davidson (1984) for tariffs. results for this He uses quantities as the strategic variables in a model very similar to ours and shows that small promote collusion, while the opposite tariffs true for large tariffs. is After considering the effects on monopolization in the country that imposes the quota we study the foreign repercussions we used that the arguments We of quotas. notice immediately show that more competition emerges to country in the that imposes the quota also imply that quotas lead to increased competition abroad the the domestic firm if is We model the capacity capacity constrained. constraint in a simple way, assuming that the firm can produce at constant marginal cost up to some quota is limit and cannot exceed that imposed when the domestic firm the monopoly output is to a capacity constrained in this way. If be sold in the domestic market, the presence of a quota requires that the domestic firm a limited capacity to is Suppose that limit. sell more at supply the foreign market. home. Therefore, it has only In that case its capacity constraint has the same effect abroad as the quota does on the imports of the foreign firm. in the Thus, for the reasons discussed above, more competition results foreign market. It is possible for quotas to lead to increased competition abroad even in the absence of capacity constraints, however. levels of demand fluctuate over the two markets. In that case, time and in either Suppose, for example, that the are not perfectly correlated across market taken alone, unable to sustain the monopoly level of prices in and Saloner (1906), when demand is outweigh the punishment which meted out is demand. states. sufficiently high, As Bernheim and Whinston (1986) show, sustain higher profits all in later, in the firms As in Rotemberg (which is may the incentive to cheat "more normal", periods. such settings the firms can when they share two markets with imperfectly In the simplest case, may be also the one that correlated we study), the markets are of equal size and have perfectly negatively correlated demand. In that case the two markets taken together are perfectly stable over time and hence, since the firms' incentive to cheat doesn't fluctuate over time, they may be able to sustain monopoly profits each period. Imposing a quota on one of the markets, however, means that the firms lose the countervailing force that smooths the incentive to cheat when demand Instead, the foreign market is high in the foreign market. transformed into a market with fluctuating demand is and monopoly profits may no longer be sustainable there. Thus the market or imposition of a quota can increase competition in the foreign domestic market. in the The natural question is whether the separate analyses discussed above can be combined to show that the imposition of a quota can contemporaneously increase competition in both the domestic and foreign markets. We show The presence that this is indeed the case. of implicit collusion is not the only reason result in increased competition in both markets. model developed in Dixit why quotas may Indeed we demonstrate that the and Kyle (1985) can be modified to show that potential entry can have the same effect. The paper is organized as follows: We show that quotas can increase competition at home with prices as the strategic variables in Section with quantities as the strategic variables in Section HI. use the analysis developed in increased competition abroad show that in is if the domestic firm demand In Section IV, is we in We capacity constrained. that this can occur even not capacity constrained in Section V. show that increased competition can result market. and these sections to show that quotas can lead to the presence of fluctuating the domestic firm II In Section VI when we both the domestic and the foreign Section VII concludes the paper with a discussion of the effect of potential entry. II Quotas and price competition at home We consider an There are two countries, domestic and foreign. oligopolistic industry with one domestic the simplest case restriction in is we assume both countries. Marginal cost is buy the good so that consumers can only slightly contradictory assumption P = a where P is constant and equal to c is own country in their it made mainly 2 . for tractability Thus the foreign Yet, and this does at home. by the two firms are viewed as perfect substitutes is This sales abroad. However, the markets are segmented firm can charge a different price abroad than Demand start with Foreign firms do not face any transportation costs from shipping the good to the domestic country. sold makes no that the domestic firm relaxed in Section IV. To and one foreign firm. , the goods the domestic market. in given by: - bQ (1) the industry price and Q is the sum of the amounts sold by the domestic and foreign firms. In this section price is This means that if one supplies the entire market. If the strategic variable. firm quotes a price lower than the other, it two firms quote the same price they share the market assumptions about the way gets half the market. feasible in which this sharing The second is the presence of quotas the foreign firm case. Thus the There are two possible done. The first is that that any possible market division and that market shares are also the market. is . implicitly agreed upon. may be forced the to each is However, in have less than half of latter assumption is the only one that makes sense in this We start by analyzing equilibrium under free trade. This equilibrium 3 the standard duopoly equilibrium in a repeated setting. is We assume that the firms try to sustain the monopoly price (a + c)/2 and that they each serve half the market. If 4 Then, 2 if either firm deviates, market so that neither firm deviates, each earns (a-c) /4b per period. it undercuts the price slightly and captures the entire However, after earns twice this amount. it a deviation the firms are assumed to revert forever to the noncooperative equilibrium for the corresponding one-period game which has a price equal This punishment which gives each firm profits of zero possible punishment that can be conceived industry 5 It . is if to the is marginal cost c. also the strongest the firms are free to leave the then impossible to make them earn negative profits. With this punishment, each firm will be deterred from deviating as long as: (a-c)^/4b < 6(a-c)^/4b(l-6) where the RHS of this equation represents the future profits that are given by cheating and long as 6 6 is the rate at which future profits are discounted. equals at least 1/2, the monopoly price We now consider the effect of a quota. allowed to import be E(a-c)/b so that sold Similarly, of if e is 1/2, it demanded can is sell So as sustainable. Let the amount the foreign firm is scaled by the amount that would be A quota where under perfect competition. firm to supply the amount it is up e is 1 would allow the foreign at a price equal to the amount demanded marginal cost. at the monopoly price (a+c)/2, and so on. Notice as an aside that any quota which is binding equilibrium, i.e. which reduces the amount Imported, measures of domestic welfare. This is at the original raises the standard so because, even if the price remains at (a+c)/2, the domestic firm having higher sales, now earns higher profits. c Domestic welfare is only increased further the national identity of firms if the price actually falls however, we focus mainly a slippery concept, is Since . on the competition-enhancing affects of quotas. We begin studying the equilibrium with quotas by analyzing the punishments for deviating from the standard practice (1971), in understanding. implicitly collusive As is models of repeated oligopoly (see for example, Friedman Rotemberg and Saloner (1986) and Brock and Scheinkman (1986)) we assume that firms revert to the Nash equilibrium in the deviates from the collusive understanding. who focuses instead on While, as one -shot game This if in contrast to is any firm Abreu (1982) the maximal feasible punishment for any deviating firm. we show below, our results are not very sensitive to our assumption of one-shot Nash punishments, there are three reasons why these may be preferable even when they are lower than the maximal punishments. the firm can place is sell of the firm to signal, to post prices other signalling is if the owners of the firm forward, the only equilibrium until the sale takes the one-shot Nash equilibrium. managers First, 7 Second, it is in the interest of the once punishments begin, that they are unwUling than those of the one-shot Nash equilibrium. credible maximal punishments as in If this Abreu (1982) become Q infeasible. Third, the one- shot Nash equilibrium easier to characterize is than the maximal punishments. The static one-shot game to which firms revert when they are punishing each other has no pure strategy equilibrium. Kreps and Scheinkman (1983) analyze the mixed strategy equilibrium, and compute the expected profits of the firm with larger capacity. They also exhibit one such mixed strategy equilibrium (for which they compute the entire distribution of prices) show that this equilibrium is the . It is straightforward to unique mixed strategy equilibrium and thus to also compute the expected profits We now list of the foreign firm at this equilibrium. The arguments behind the salient features of this equilibrium. these assertions are contained in the appendix: The highest price charged by both firms (i) The lowest price charged (ii) (a+c)/2 = and (iii) e) 2 it is [a+c-E (a-c) ]/2. is is (a-c)(2E-E^)^^^/2 - (2) charged by both firms with probability zero In equilibrium, the domestic firm has expected profits of (a-c) /4b per period while those of the foreign firm equal e 2 (1- (o-c) (a-c)/b. Notice that this static equilibrium has the features of the differentiated products model of Krishna (1986). A higher quota highest and lowest price charged. Kreps and Scheinkman (1984) show that the entire distribution of prices is (a stochastically dominated higher e) lowers both the by that with a lower quota. There comment. c) 2 is one more feature of this punishment equilibrium that deserves At this equilibrium the domestic firm earns a present value of (a- 2 What must be noted (1-e) /4b(l-6). is that the domestic firm can never earn less than this present discounted value of profits at The reason one involving maximal punishments. firm can gviarantee for itself that is for this 2 it is that the domestic 2 wUl earn (a-c) (1-e) /4b per period by posting a price equal to [a + c-E (a-c) ]/2b. the domestic firm any equUibrum; even the This what gives quotas their limit on the punishability of ability to break the monopoly price With the value of the punishments in hand we now turn to the analysis of the repeated game. The price preferred by the domestic (a + c)/2 while the foreign firm, higher price. which is firm continues to be subject to a quota, naturally prefers a Yet we concentrate on the question of whether the duopoly can We do sustain the "monopoly" price of (a + c)/2. First, this for two related reasons. higher prices are more difficult to sustain so that sustain (a+c)/2 they cannot sustain any higher price. whether the introduction in if the firms cannot Second, we are interested of a quota lowers equilibrium prices from their free trade level of (a+c)/2. The firm is sustainability of the price (a + c)/2 expected to seU allowed to sell its have no incentive deviate since is expected it at this price. entire quota, to deviate. earns more foreign firm the foreign firm In this case, game. Similarly, 2 Instead, /2b. if to sell y(a-c)/b (0<y<l) 1/2, in which case it sells its entire 1/2, in which case it sells Consider 2 c) /2b. (£ first the It will which the in the collusive arrangement. its profits it in which t is - quota, and the case in which it By + E - 6e(2e - 2 E )^/2 exceeds deviating, the foreign firm earns E(a- e(2e - E^)^/^]/2(l-6) (3) or: V > sells (a-c)/2b. former case. E are y(a- smaller than or equal to thus choose to deviate unless: y)/2 < 6[p - in entire quota at a price essentially identical to its We analyze separately the case (a+c)/2. the foreign firm deviates by undercutting the price slightly, it demand or either total if the domestic firm never deviates but the foreign Then, by going along with the collusive arrangement, c) is the domestic firm will always we consider the intermediate case So, supposed is for instance, E(a-c)/b, at this price the foreign firm will in the static to sell nothing, firm always deviates. If, depends on the amount the foreign 1 . (4) 11 Note that for small e, must essentially equal y that the price will roughly equal (a+c)/2 punished 9 Thus at the is Note also that the higher is the firms' discount factor, it is monopoly price On that price. is being sell essentially its y/2 if it goes along cheats it sells it can 2 the other hand - If it while of (a + c)/2, Thus the domestic y/2 < 6[E/2 allowed to knows firm 6, entire the less willing to sell without cheating. consider the domestic firm. cheating. goes along or it the foreign firm Now it deviates unless . capacity. whether The foreign e. sell (a-c)(l/2 - vi)/b (a-c)/2b at 2 earns only (a-c) (1-e) /4b per period after it firm is deterred from price-undercutting if: e^/4]/(l-6) - or: y < 6e - 6e^/2. (5) Equation (5), which y must be relatively small since higher levels of y is if valid also when punishments the domestic firm is to are maximal, shows that be deterred from cheating make cheating more attractive without increasing its cost to the firm. If (4) and To sustainable. (5) contradict one another the monopoly price see under what conditions this occurs is not we combine the conditions and obtain: e{ 1 - 6 - 6[(2e - e^)^'^^ - E/2]l < 0. (6) For small E enough we can neglect the term second order) and the condition clearly violated. is saw that the foreign firm must be allowed But the domestic firm always requires that strictly smaller than e for y equal to ; in e it square brackets (which When e small is is of enough we to sell essentially its entire quota. y be smaller than 6e which is has too strong an incentive to undercut the monopoly price For brackets increasing in is up e entire quota when .62 for (6) cheats to the term in brackets 1 z until it cheats is square only this the foreign firm cannot sell its e , 6 must exceed about y)/2 < 6[y - E exceeds 1/2 so that when the foreign firm e Then the foreign earns (a-c) /4b. - Yet this analysis in be satisfied. consider the case in which (1/2 The term positive. For this maximal applicable . 2 it is reaches .553. e beyond to 1/2 since relevant for Now and between E + e(2e - firm will cheat unless: e^)^''^]/2(1-6) or: y > (l-6)/2 + - 6e 6£(2e - t'^)^^^ (7) . which must now be satisfied together with (5) for monopolization to be feasible. So we substitute (7) (holding with equality) in (5) and obtain: (l-6)/2 < 6e(2£ For demanded E - £^)^/^ equal one, that is - 6e^/2. when at the competitive price, (8) the foreign firm can sell the entire quantity (8) requires that 6 exceed 1/2 as it did 13 under free trade. Since the (and thus of of 6/(1-6) 6) RHS of (8) is strictly increasing in t, required to make (8) hold, the level when falls strictly e rises. To summarize the results of this section, more restrictive quotas (starting quota which allows the foreign firm to at a sell demanded the entire amount at the competitive price) monotonically reduce the ability to monopolize the Note that, for a given market. domestic market is 6, the quota that gives the minimum price in the one which satisfies strictly smaller than the When these equations hold with with equality. For lower values of sustainable. equality, monopoly However, the price falls. t, (or (8)) (6) is just the quota if is reduced significantly more, the price starts rising again as the prices charged even In the one-shot game rise. For equal to zero, the monopoly price t is reestablished. It must be pointed out again that the increased competition brought about by quotas is not sensitive to the use of one-shot Nash punishments. instance even deviations by the foreign firm lead this firm to earn zero if profits from then on (which for low e is a much harsher punishment than maximal punishment) the foreign firm wUl require that so as to refrain from deviating. 1/2 and any positive This is y briefly compare the results analyzing tariffs in a similar setting. of the foreign firm are higher, domestic firm. 6 equal to e. in this setting is in direct contrast to the conclusions of the static analysis reported We now the equal at least (1-6)e inconsistent with (5) for The increased competition brought about by quotas (1985). For by the The repeated game in A to those by Krishna one would obtain when tariff simply implies that the costs tariff rate, than the costs of the which the two firms have different costs has been analyzed by Bernheim and Whinston (1986), who consider optimal punishments firms in the style of Abreu (1982). Then, both the domestic and foreign earn zero profits when they are being punished still . This means that the incentives to participate in the implicit agreement that requires that the price (a + c)/2 be charged, do not change as a result of the tariff. Moreover, Bernheim and Whinston (1986) show that the equilibrium that Involves the highest profits for the duopoly as a whole now has a price higher than (a + c)/2. This occurs because the profit maximizing price from the point of view of the foreign firm is now higher. So a tariff has the potential for increasing the domestic price above the monopoly price. Thus, in the case of maximal punishments the classic results of Bhagwati (1965) about competition between a foreign and a domestic firm, are precisely reversed. A quota, because punished effectively, makes consequence. it makes it impossible for the domestic firm to be difficult to collude, while a tariff has no such This raises the intriguing possibility that this governments seem However, it it to prefer quantitative restrictions to tariffs is the reason 12 must be pointed out that the robustness of the monopoly outcome with respect to a tariff is sensitive to the use of maximal punishments. If instead, reversions to the Bertrand outcome are used, a tariff which raises the foreign firm's costs barely below the monopoly price of (a+c)/2 makes the monopoly price unsustainable. The reason for this is that to maintain this price the domestic firm must give a sizeable fraction of the market (1-6) to the foreign firm. Thus if the foreign firm's costs are near (a+c)/2 the domestic firm will actually earn more during the period of punishment (when price barely below the foreign firm's costs) than when it it charges a goes along with the monopoly price The results of this section are sufficiently striking that they deserve to be qualified somewhat further. The analysis hinges critically on the presence 15 Since this means that only the foreign firm can of only one domestic firm. punish the domestic firm, of a quota. punishment this ability is curtailed in the presence there were more domestic firms in the market, they could If revert to a price of c if any of all them deviated and thus keep the oligopoly the monopoly outcome even in the presence of a quota. On at the other hand, if the domestic firms are capable of concerted cheating, the approximation considered only one domestic firm in this section of Ill may be a good one. Quotas and quantity competition at home In this section we assume This has two consequences. competitor amount. when First, deviates. neither firm eliminates the sales of These are maintained in quantities, which is prices are the strategic variables. The its at the implicitly colluding Second, punishments are characterized by reversion to the single period Nash equilibrium when it that quantities are the strategic variables. a much milder form of punishment than 13 results of this section are directly comparable to Davidson's (1984) analysis of tariffs, in which he also assumes that quantities are the strategic variables. collusion to import We show that the result that quotas lead to less sustainabUity of now only holds very little. for very restrictive quotas that allow the foreign firm For larger quotas, we find that collusion anything, enhanced by the quota. Thus our Davidson's (1984) since he shows that is, if results here are similar to small tariffs (which correspond to large quotas) enhance collusion while large tariffs do the opposite. Again, the firms are assumed to be able to monopolize the domestic market in the absence of quotas. 2 earns (a-c) /8b. If By going along with selling (a-c)/4b each, each firm either firm were to deviate it would choose to sell 3(a- 2 c)/8b which would give profits of 9(a-c) /64b in the period of the deviation. it From then on, the firms would revert would will to Cournot-Nash equilibrium so that each 2 (a-c)/3b and earn profits of (a-c) /9b per period. sell be deterred from cheating [9/64 - Now 6 - 1/9] (a-c)^/b(l-6) So suppose that level of sales, it is to the If e is above 3/8, so that the quota monopolizing level of output, has no effect. it lower than this level but higher than the Nash equilibrium that i.e. , exceed 9/17. consider a quota of E(a-c)/b. exceeds the best response each firm if: l/8](a-c)^/b < 6[l/8 which requires only that So, z is between 3/8 and 1/3. Then the quota does not affect the foreign firm's ability to punish the domestic firm from deviating (since each firm produces (a-c)/3b in makes the foreign firm's own deviation cannot now exceed E(a-c)/b. Cournot-Nash equilibrium. it done in the So such a quota has no effect when the firms can absence of a quota. is For suppose that selling (a-c) /4b, constrained, somewhat lower. since it 6 is it when it now earns will still not This means that With z not only wUling to go along with is less when it deviates than may be possible if In essence, it its does if it sales are to sustain the output by having the foreign firm produce less than half of the domestic firm produce more than half. may just below 9/17 so that be induced to cheat even it actually this cannot be not sustainable in the absence of a quota. between 3/8 and 1/3 the foreign firm isn't Furthermore, possible to sustain this monopolistic outcome even the collusive output it less profitable since this deviation sustain the monopolistic outcome without a quota. make However, monopoly total, and the the "excess" unwillingness to cheat on the part of the foreign firm can be used to free up the constraint on 17 the domestic firm. Thus for a quota to promote competition equilibrium (a-c)/3b, detail. We t must be below the Nash it We now analyze such must be less than 1/3. on the part of the foreign first look for the lowest level of sales firm which makes willing to go along with an equilibrium it Then we analyze whether (a+c)/2. Suppose that the foreign firm and that the firms attempt the domestic firm by the foreign this price given these sales p(a-c)/b in equilibrium to sell of output. derive the minimum level of y for which the foreign firm By cooperating deviate. 2 sells Its It output The domestic firm earns (a-c) y/2b. requirement that deviates sells - We y)/b. This y. is We now not want to (a + c)/2 so that at a price of it monopoly output show below that the be smaller than 1/3 Implies that when the foreign firm z E(a-c)/b. Thus when it deviates 2 it Note that the profits to the foreign firm when p)/b. in it will sells that portion of the not sold by the foreign firm, or (a-c) (1/2 is willing to go along with is monopoly level to sustain the whose price firm. supposed is a quota in because the higher is y the earns (a-c) e(1/2 it - + e deviates are increasing lower are the domestic firm's sales when they both go along with the proposed equilibrium output. After game. its deviation the market reverts to the equilibrium in the one-shot Since the quota is less than the Nash equilibrium value foreign firm sells E(a-c)/b while the domestic firm sells this level of output, E)/2b. Hence the foreign firm (a-c)^{E[l/2 i.e. or (a-c) (l-E)/2b. if: + y - e] - will its Thus the foreign - e the best response to 2 firm earns (a-c) e(1- be deterred from deviating y/2!/b < 6(a-c)^[y of (a-c) /3b, if: (1-e) ]/2b(l-6) y > E - 6e^/[1 2e(1-6)]^^. - Note that for values of for this is when that near zero, the foreign firm e the implicit agreement allows (9) is deviating unless it that for sufficiently small is already selling all as allowed to is that The reason price does not differ appreciably from small, Then (a+c)/2, even during the punishment phase. much deviate unless to sell nearly its entire quota. it the quota firm's interest to sell as will it can and it always is in the foreign cannot be deterred from essentiaUy sell it its This means entire quota. e the foreign firm has no punishment power since it is allowed to bring in. Given surprising that we establish next that for sufficiently small this, e, it is it not the equilibrium cannot involve the monopolization of the domestic market. Suppose that the Consider the domestic firm's incentive to deviate. foreign firm best response to this, c) 2 (a-c) (l-y)/2b, 2 (l-y) /4b in the period in which would sell The domestic firm would selling (a-c)vi/b as above. is it if deviated. it It would then earn From then on the foreign firm deviated. 2 2 (1-e) /4b per period. So the domestic firm would be deterred from deviating only - v)^/4 - (1/4 - increase is y, easy to see that when u falls, as long as y is if: p/2) < 6[l/4 It is (a- (a-c)E/b so that, by an argument similar to that above, the domestic firm would earn (a-c) (1 sell its - y/2 - (1 - E)^/4]/(l-6). in spite of the fact that profits the inequality in (10) smaller than 6/(1-6) is more (which is (10) from deviating likely to hold the smaller true by assumption). The 19 reason for this Is that the principal effect of a decrease in y that the is domestic firm earns more by going along with the collusive agreement. make the domestic firm as firm sell only as much as willing to go along as possible required to make is (9) we Thus the foreign let hold with equality. Substituting this expression for y in (10), multiplying by (1-6) 1-2e (1-6) [ collecting terms, E^(l-26^) - we divide E^(l-6)[4(l-6^) + + by e this expression 2 by (after dividing (11) range of t. and 26] e'^(1-6)[4(1-6) obtain (1-26 )<0, which requires of 2 ] the inequality (10) becomes: + If to 2 e This means that 6 46(1-6) 2 + 6^] < 0. and take the limit as e if 6 exceeds .71 it is goes to zero we Furthermore, the LHS smaller than about .71. decreases monotonically with ) (11) e in the relevant always possible to sustain the monopoly price. If, instead, 6 between 9/17 and is .71, the monopolistic outcome sustainable with free trade but not with a sufficiently small quota. is In this case of linear demand, however, the quota must be very small for this result to hold. If we take the smallest discount factor consistent with collusion under free trade (9/17), one can demonstrate numerically that the monopolistic outcome is not sustainable for e equal to .01 but These are such low values of e is sustainable for e equal to .015. that even the noncooperative solution looks essentially identical to the monopolistic solution. IV Quotas and competition abroad: Constant demand We now turn to the implications of the existence of a quota at home for competition abroad. Sections and II to III we adapt the arguments presented In this section in argue that under certain circumstances they can be interpreted as providing conditions under which competition increases abroad. We assume that the domestic and foreign markets are demand curves are given by If (1). both firms have infinite capacity, there a monopolistic equilibrium in which (a + c)/2 as 6 identical so that their is charged in both markets as long equals at least 1/2 when prices are the strategic variables and least 9/17 when becomes So marginal cost equals infinite. in is equals at that the domestic firm actually has limited capacity given If k equals c for instance, 2, that this capacity is by untU the firm produces k(a-c)/b and then the domestic firm can supply the The quantity demanded at the competitive price in both markets. section 6 quantities are strategic variables. Now assume k(a-c)/b. is idea of this sufficient to maintain the monopolistic outcome both countries with free trade. However, with a quota that forbids all imports into the domestic country, the domestic firm must devote a larger fraction of its capacity to producing for the protected domestic market. then effectively has a very small capacity for selling abroad. It This has the same effect abroad as would a quota imposed by the foreign country. We is first treat the case in which prices are the strategic variables. equal to 1/2 (the benchmark case) and k is less than 2, If 6 the two firms cannot maintain the monopolistic outcome in both markets even in the absence of a quota. This results because the domestic firm does not have sufficient punishment power outcome can to at best keep the foreign firm from cheating. be sustained for values of 6 Thus the monopoly that are greater than that required to sustain monopoly under free trade without capacity constraints. We consider how large the "critical" value of 6 must be for different values of k. Note first that k must be at least equal to 1/2 for the problem to 21 be interesting since otherwise the domestic country own market when its suppose that k is (the world market value of 6 that market 6 Now consider is 6 about is about . .65. If the effect of a quota. important to notice is is sustainable in the world as a instead k at This is is left over to that the domestic firm is firm sells (a-c)/2b at the foreign country. sell in is it is never willing punishing the foreign selling it abroad This means that the domestic firm sold at home. e we can use (6) 6 that sustains the monopoly outcome abroad. value of 6 strictly below to a value of is .65 in the is Letting k-1/2 be minimum For k barely above 1/2 no sufficient to maintain that outcome (in contrast absence of a quota). For k equal to 1, a value of .65 required when only .62 was required under free trade. We now turn to the case in which quantities are the strategic variables. Hence Here the monopoly outcome only broke down for tiny quotas. domestic firm has a capacity of (1/2 +e) where e is negligible, outcome can be maintained with free trade as long as other hand, once the domestic market firm 6 less than (or (8) for k bigger than 1) to discover the value of 1 if effectively is capacity constrained abroad with a capacity of (k-1/2) (a-c)/b. denoted by to sell so because the marginal revenue from selling a unit home always exceeds the marginal revenue from (a-c)/2b equal to one, the is The domestic more that (k-1/2) (a-c)/b units abroad, even when firm for deviating. equal to k/2 e 62 home and has (k-1/2) (a-c)/b units What with equivalent to twice the domestic market) to obtain the is for which the monopoly outcome value of critical (6) in So first to foreign firms. Then we can use barely above 1/2. This value of whole. becomes closed unable to meet demand is is is willing to sell at most e is abroad. 6 if the the monopoly exceeds 9/17. On the closed to the foreign firm, the domestic This means that for e small below .71, the monopoly outcome won't be sustainable abroad. enough if V Quotas and competition abroad: Fluctuating demand This section considers a somewhat different model in the country tends to increase competition collusion is The other country. idea in one market, they are small In this setting, one that is in the other market and the overall incentive to cheat (taking the two markets together) are fairly constant over time. If one looks at either market Then separately, however, the incentive to cheat fluctuates over time. in in maintained under free trade only because whenever the net benefits from cheating are big viceversa. which a quota in if, as Rotemberg and Saloner (1986), future punishments are largely independent the current period's incentive to cheat, then those punishments insufficient to deter cheating high. It if the current incentive to cheat may be is particularly thus becomes more difficult to prevent them from cheating. imposition of a quota has the effect of separating the two markets, makes implicit collusion Our specific model more of it Since the thus difficult. assumes that demand is random in both countries, as Rotemberg and Saloner (1986), and negatively correlated across markets as Bernheim and Whinston (1986). To simplify the analysis we assume in in that there are only two states of demand (high and low) whose realizations are independently distributed over time in any one country and perfectly negatively correlated across markets. So when demand In this section we assume is high in one market it is low in the other. that there are N/2 firms in each market while do not restrict attention to linear demand curves. symmetric, the monopoly price in the high state, p Since the countries are , is the same in both countries and similarly for the monopoly price in the low state, p the analysis by ensuring that these two prices p . We start and p are sustainable with 15 we 23 That free trade. is we consider a situation in which all N firms sell in both markets and derive the condition under which no single firm wants to deviate from charging p in the country in which demand is high and p in the other. Prices are again the strategic variable so that any deviation triggers a permanent reversion which price equals marginal to the static equilibrium in cost. Let be the total monopoly profits IT those in the market with low demand. former in the market with high demand and As long as p exceeds marginal the cost, bigger than the latter since the monopoly always has the option of is even when demand charging p is A high. firm that deviates from the monopoly outcome undercuts the price charged by the others slightly. the market in which deviating it demand gives up both these profits. (tt" it Thus each + TT^)(1 1 < 6/(1 - is high and now and firm 1/N) < 6(tt" is + in all In the other. it future periods On its thus earns tt in the other hand, by It share (1/N) of deterred from deviating as long as: tiVn(1 - 6) (12) or: N - - (13) 6). We assume that either the number factor big enough that of firms Is small enough or the discount (12) holds so that with free trade the two markets can be completely monopolized. We now consider what happens when no imports are allowed country. This drastic quota has two effects. fewer (N/2) active firms in the First, it into the domestic means that there are domestic market while N firms continue to sell in the As a foreign market. the conditions under which monopoly will be result, As sustainable will be different depending on which market has high demand. Rotemberg and Saloner (1986) show, in this setting sustain collusion when demand monopolization is possible, monopolization is possible in the high Consider /N abroad. The Thus low. is always easier to whether complete to see we study the conditions under which such Any demand state. country when demand first the foreign firms try to charge p It is it high. Suppose that the is firm that undercuts this price earns losses from the ensuing competition are smaller for foreign countries since domestic countries may also lose any profits they make So we concentrate on the incentive to deviate of foreign firms. they lose the expected value of profits in the future. If a foreign firm will deviate •tt"(1 - 1/N) > 6(tt" Note that the of RHS + (it is +tt)/2. if: ttV2N(1-6). (14) of (14) equals half the RHS of (12). The costs in terms foregone future profits from deviating are half what they would be firms had access to both markets. of (14) and is (14) to hold simultaneously. in the it is In particular, if is bigger than it is When it is ti , the if in the profits (14) is impossible to sustain monopoly abroad high state of demand so the quota increases competition. It is LHS (12) holds as an equality just sustainable with free trade) then (14) holds, foreign if possible for (12) high and low demand states are different, the inequality always satisfied. in the Yet, insofar as bigger than half of the LHS of (12) so that (so that monopolization home. at they deviate Since demand independently distributed over time, these expacted profits equal So, instead of it easy to show that when (12) holds it Is always possible to sustain 25 monopoly abroad the low state of demand. in (12) ensures that deviations Yet, when demand would prevail if is low, will This results from the fact that when demand not take place even constant. is expected future losses from deviating exceed those that demand were expected This to stay low forever. from the fact that even when (14) holds, it is still possible to sustain This can be seen as follows; substantial profits in the high state. turn results in Let the sustainable price (the highest price that can be charged without inducing any firm to deviate) in the high state be p argument analogous one used to the refrain from undercutting p - Ti^(l 1/N) = 6(Tr^ + s and the ensuing profits to derive (14), s it Then by an . a foreign firm will just if TiV2N(l-6) or: TT^ = so that IT ttV{2(N-1)(1-6)/6 s equals n 1 when - 1]} (15) (12) holds as an equality and exceeds when it it holds as a strict inequality. So far, we have shown that under plausible conditions the foreign market becomes more competitive while still retaining substantial profits. This result does not hinge on the fact that prices are used as the strategic variables. As in Rotemberg and Saloner (1986), sustain when demand is high. it As we showed there, this quantities are the strategic variables We now turn only requires that collusion be difficult to when demand is is also true when linear. briefly to an analysis of the domestic market. If domestic firms were to deviate from the collusive arrangement they could be punished in both markets. oncG. On In this case, the other hand deviating firms would deviate in both markets at we have already derived the net benefits from deviating abroad under the assumption that punishments would only be meted out abroad. To complete the analysis we thus look at the net benefits from deviating in the domestic market alone under the assumption that firms will only be punished in Then, the the domestic market for doing so. the sum of these two net benefits. Suppose that the domestic market high, a deviating firm would earn then lose It it is instead of would thus be deterred from deviating tt"(1 2/N) < - The RHS meted out only the share fully monopolized. share of total expected profits in its total net benefits to deviating are 6(tt" of (16) in is + its its When demand share (2/N) of if: RHS On While the punishment of (12). one market, the share of each firm LHS would (16) the same as the substantially smaller than the It . own market from then on. ttVn(1-6). we previously considered. n is in this the other hand the of (12) market LHS is is double of (16) is since deviations are taking place in one market alone whUe, in addition, the net benefit is smaller because there are fewer firms active in the market and so the firm's share was larger to start with. So (12) is satisfied (even with equality) if strict inequality. In this case showing that monopoly is This applies to the case in then (16) is satisfied as a which demand is high at home. demand is low in the foreign market and our previous analysis there is a net loss from deviating there applies. sustainable in the absence of a quota home market, then When demand it is is when demand Thus is if high In the also sustainable with the quota. low at home the corresponding condition at home has the same 27 RHS as (16) but an even smaller LHS. So the net benefits to deviating are negative here as well. In summary, when holds, monopolization (12) is always possible in the protected domestic market. Indeed, since protected domestic firms strictly prefer to go along with monopolization than to deviate when (12) holds as an we can conclude that protection may make monopolization equality, home when The was not feasible with free trade. fact that collusion potential, To do it in a richer model, to make monopolization feasible abroad as well. the market unevenly as Section becomes easier for the domestic firms has the we must allow the this II. feasible at is set of firms that done charge the lowest price to split Bernheim and Whinston (1986) and as we do in Bernheim and Whinston (1986) show that the possibility in such of arbitrary divisions aUows "punishment power" to be shifted from a market in which there is more than enough monopoly outcomes. to one in which there isn't enough to sustain In our example, the domestic firms can offer the foreign Domestic firms would firms a share higher than 1/N of the foreign market. tolerate this for fear of losing their domestic profits if they deviate abroad. Foreign firms would earn more when they go along with the collusive arrangement and thus would be less willing to deviate. potentially achievable when demand does not work, however, when there is N is In this fluctuates relatively equal to 2 home when it little. deviates abroad. Then Its is This approach so that as in Sections only one firm in the domestic market. fear retaliation at way monopolization II this firm does not and need incentives to deviate abroad are the same as those of foreign firms and giving the foreign firm a disproportionately large market share only increases the domestic firm's incentive to deviate abroad. III to VI Quotas and Competition in Both Markets we consider the In this section competition at home (as in Sections Sections IV and V) fluctuating possibility that a quota will increase and 11 The model we use . demand model of Section as well as abroad (as in 111) to illustrate this possibility is the V except to permit competition to that, increase at home, we consider a strictly positive quota. analysis we one firm in restrict attention to the case in which each country, and 6 is Also, N equals to simplify the so that there 2, equal to 1/2, so that monopoly is is just sustainable in both markets with free trade. Since any firm that deviates in one market will be punished in both, any deviating firm wUl undercut the implicitly agreed upon price in both countries. So the incentive to deviate in any one country can be thought of as the sum of the net profits from deviating at home (i.e. the benefits from deviating at home minus the present value of the losses home) plus the net profits from at deviating abroad. Suppose that the foreign firm has a share its net profits from cheating when demand those when demand notice fall is that, as low are (1-3o/2)it is s is 1 high are (l-3a/2)Ti -otT/2. net profits from cheating those when demand a of the foreign market. when demand is s high are (3a/2-l/2)iT -(l-a)Ti /2. we increase a, -ott /2 while Similarly the domestic firm's low are (3a/2-l/2)iT 1 is s 1 1 s -(l-a)-!! /2 and The important thing to the foreign firm's net profits from deviating by the same amount as the domestic profits from deviating increase. as long as in each state the total net profits from deviating (the firm's net profits from deviating) are zero, redistribution of shares that will Then there make both firms We therefore now concentrate only on the will sum Thus, of each always be a willing to go along. total incentive to deviate. In 29 the foreign country this s So unless state. abroad is it is equal to home at We can then write the quota enough small Is )/2 -it 1 it in the low state and foreign firm is t is given by P=a -bQ (where in state Suppose that less than 1/2. is as equal to i decreased incentive to deviate is (a -c) e [e (1-6) + /2 6e - violated which 6z {2t -t This (a -c)/b at the )^ to maintain expression s 1 abroad. ) in monopoly (17) with i at if the quota monopoly price the (4) the we can obtain the ]/h y one firm's (17) home. replaced by 1 This 6 is the same as that (6) be is equal to 1/2 in Section in the is At least one of these net incentives i two markets taken together given by when demand plus the expression in (17) with II. domestic market. in the consider the net incentive to deviate we try where 1) given by: is for the case in which thus a net incentive to deviate Now (tt -tt we proved equals either u or i Once again by varying Note that the condition that (17) be positive by positive the other's increased incentive so we can focus on the total incentive to deviate. if is high in the -it Then we can obtain from allowed to seU y (a -c)/b. domestic firm's net profits from deviating. is si )/2 the total incentive to deviate foreign firm's net profits from deviating while from (5) There (it one of the two states. in Suppose demand is (tt is s (tt 1 -it ) plus the high abroad. It is replaced by u when demand is positive so that it given is low is impossible to sustain monopoly at home. Now we consider whether it is possible to have an outcome different from the monopoly outcome at home such that the net incentive to deviate at home is so negative that the monopoly outcome can be maintained abroad even in the high state. That this is unlikely for small quotas should be apparent since, then. the static game gives about as many profits as the monopoly outcome. This suggests that there are no prices between the monopoly price and those that prevail under static competition for which the costs of reverting to competition are high. We now consider and study the overall profits from deviating home. at a price p at home The amount demanded (a-p)/b and, since the distribution of this demand across firms this price is immaterial, We consider more formally. this issue we the foreign firm seU let sells the difference e at is (a -c)/b while the domestic firm between the quantity demanded and the quota. The domestic firm's net profits from deviating are then given by: E^(a'-c)(p-c)/b - (p-c)[(a'-p)-E(a^-c)]/b + (a'-c)^(l-E V/4b. The foreign firm gains nothing by cheating but the static game e (a -c)(p its cost from reversion to is: - a]/b so that the overall benefits from deviating are: {p-c)(t\a^-c) whose minimum is - (a^-p))/b (a^-c)^(l-E^)^/4b + reached for a p equal + E^a^-c) (o-c)/b [a +c-£ (a -c)]/2. There, this expression equals: {-E^/2 + e[1 - (2e-e^)^^^ + E^]!(a-c)^/2b (18) 31 which is maximum negligible for small e . The cost from deviating at home, fact that 0, (18) small. is is small means that the maximum Yet, profits in the high state abroad must satisfy: (tt^ - TrV2 - so that for small to * it < impossible to make is s it equal to in which Tn it as would be required monopolize the foreign market. VII . Conclusions We have shown in a variety of models implicit collusion is important that quotas have the potential of increasing competition, not just in the domestic market, but also abroad. those obtained when there is These results stand a single domestic firm in sharp contrast to and goods from abroad are supplied competitively (Bhagwati (1965)) as well as those obtained when domestic and foreign firms act as The in a one- shot game. intuition underlying our seemingly paradoxical results are not restricted to quotas. competition less viable. Indeed, any action by the government which makes when coUusion breaks down more profitable Thus Government imposed minimum makes collusion itself price floors could have precisely the same effect. One natural question to ask is whether our results depend critically on the existence of implicit collusion or whether they can also be generated in other models of strategic interaction. would be modified in the In this conclusion One particular concern is how our results presence of entry. we sketch an argument that shows that quotas also tend to increase competition The the industry. by encouraging the entry we consider setting new (domestic) firms very closely related is to that into developed They analyze an industry with an established foreign and Kyle (1985). in Dixit of firm and a domestic potential entrant. They show that it possible that is if the domestic country prohibits imports that there will be more competition in the foreign market. This occurs because the potential domestic firm may find unprofitable to enter if By protecting it in the it must compete with the foreign firm domestic market (while market), the domestic government provides its fixed costs. competes both markets. in the foreign with sufficient revenues to cover then finds entry attractive, enters, and increases It Since the domestic market was monopolized in any event, the competition abroad. domestic country it it in it is made better off since consumer surplus is unaffected and rents have been shifted towards the domestic country. This analysis can easUy be extended to include the case of a nonzero Moreover, quota. it not only does competition increase abroad, but in that case, Consider the case where the firms increases in the domestic country as weU. compete with outputs as their strategic variables if entry occurs. Suppose further that the domestic potential entrant will find entry worthwhile earns its K is greater than than the domestic monopoly revenues. quota which is its Now domestic Cournot revenues and less let the domestic government impose a such that the domestic firm earns revenues of K when best-response to the quota. to enter it competitive revenues abroad plus some amount, K, of revenues in the domestic market, where its if and does so. Once it In that case the potential entrant is it produces willing has entered the firms produce their Cournot outputs in the foreign market, and in the domestic market the foreign firm its quota while the domestic firm increases in both countries. sells its best-response to that. sells Competition 33 The protect Dixit its and Kyle (1985) insight domestic firm above analysis points out, insulation in its fvirther order in is own market to is that the domestic country encourage it to enter. that the domestic firm in may may have However, what the not need complete which case domestic welfare can be improved even by allowing some foreign competition. In contrast to our analysis is this result holds whether the a quota or a tariff. All that is in Section II, instrument used by the domestic country required is that the domestic government increase the domestic profits of the domestic potential entrant to the point where entry tariff raises the costs of the foreign firm, is they make attractive. it Since a "less aggressive" in the domestic market which in turn makes that market more attractive to the potential entrant. Thus when considerations to of implicit coUusion or of potential entry are important, quotas can lead to greater competitiveness. APPENDIX appendix we consider the one- shot game In this and strategic variables which one in made establish the claims s = [a + namely that: in the text, The lowest price charged (ii) (a+c)/2 = and (Hi) - is: this lowest price is charged by both firms with probability zero the foreign firm earns e(o-c) (a-c)/b. In equilibrium, of these in turn: The highest price charged by both firms the domestic firm, it will never choose to charges its highest price, But the price that maximizes which (11) is its than Hence the domestic firm knows that be undercut with probability one. when it is undercut for sure is s, therefore the highest price charged by the domestic firm. The lowest price charged is if it e - (a-c) (2e-e2)1/2/2 and First, it . (o) This price cannot be the case its In which the domestic firm would be willing o (i.e. has a number of interesting in an equilibrium that the foreign firm always charges a price strictly above a. would always make more than (iii) (o-c) (a-c)/b. could thereby be sure to capture the entire market undercut the foreign firm) properties. (a+c)/2 o = Consider the lowest price charge will profits equilibrium, the foreign firm earns to it sell less charge a price higher than the highest price charged by the domestic firm. it s is given that the foreign firm can only First note that, when is: {a-c){2t-t^)^'^^/2 We discuss each (i) We E(a-c) ]/2 - c which prices are the of the firms is capacity constrained. The highest price charged by both firms (i) in profits at s If it did so the domestic firm by charging a price between o and 35 the lowest price charged by the foreign firm, which contradicts the fact that is highest price. its domestic firm, mean unless would benefit from undercutting at o so that it would be unwilling the lowest price charged is charged by the it with positive This would imply that the domestic firm couldn't expect to capture demand of a is the lowest price if cannot be charged with positive probability since that would that the foreign firm probability. all it Secondly, s charge to by the domestic On o. the other hand the foreign firm (who firm, seeks only to undercut the domestic firm) would never charge a price of who charges must be lowest price charged by the domestic firm it o. So o with zero probability. Since the foreign firm must make equal expected profits at prices it charges we can compute in equilibrium its profits when being able to sell its entire quota so these profits equal If its it the domestic firm charges is a. sure to When sell it does this the entire £ is is assured of (o-c) (a-c)/b. demand given by profits from charging a price p are given by: (p-c)(a-p)/b = (a-c)^/4b 2 - [p - 2 which must equal (a-c) (1-e) /4b = (a+c)/2 - (a-c)(2E-£^)^^^/2. the expected profits simply by its computing all (a+c)/2]^/b at a price of o. (Al) So o equals 17 (A2) (1), FOOTNOTES proven by Mookerjee and Ray (1986) and Bernheim and Whinston (1986) 1 This 2 The notion is of segmented markets is discussed, for the case of static imperfectly competitive models, in Helpman (1982). 3 For the specific demand and cost functions used here is it presented, for instance, in Rotemberg and Saloner (1986). 4 The first of demand, if these assumptions 5 it The second assumption punishment 6 is is large. is the division of the spoils that able to resist a larger temptation to deviate Thus collusion what leads Abreu (1982) The abUity considered 7 is easiest to collude. Each individual firm This not restrictive since, with constant the firms can sustain any collusive outcome, they can also sustain the monopoly outcome. makes is is easier the larger to focus if the the punishment. is on the maximal possible punishments. of tariffs to shift rents from foreign to domestic firms is in a static In other words, model by Brander and Spencer (1985). following a deviation, the owner of the deviating firm signs a contract to pass ownership of the firm to a third party for a nominal fee (which can even be less than the scrap value of the plant) specifies that the current owner will date at which ownership passes. involved in a finite-horizon . The contract receive the profits that accrue until the Then the current owner and the rival are game which has as its unique equilibrium the outcome from the one-shot game. 8 Such signalling might be credible, for example, managers who are "old-fashioned" in the if there are some types of sense that they have never conceived of maximal punishments, and who therefore revert to the one-shot Nash equilibrium 37 in any reversionary period. 9 This result does not depend on the use of one-shot Nash punishments and can The reason be derived also with maximal punishments. mentioned above, the domestic firm can be sure e) 2 at the is this loss. this loss, below o. To obtain It must then be compensated a lower 10 firm earn less than it in later this price v it periods for taking we assume that, after taking 2 - [ (a-c)/b] {e/2 + [ Then 6(2e-e^)/(1-6) ]^^^| so that, to order the foreign firm earns £(a + c)/2 even when Footnote 7 establishes that for e small, \i it is being punished. must be essentially equal for the foreign firm not to cheat even with maximal punishments. valid in this case as well, monopoly is does must charge a price the domestic firm earns the entire monopoly profits (a-c) /4b. V must equal at least (a+c)/2 first bound on as that, 2 one- shot Nash equilibrium for at least one period V which is earn at least (a-c) (1- to To make the foreign /4b(l-6) at any equilibrum. for this unsustainable for small e to e Since (5) is even with maximal punishments. 11 This conclusion, which is also derived by Mookerjee and Ray (198G) for a slightly different game, is somewhat striking since the domestic firm makes there is static competition positive profits if by simply charging less than the foreign firm's marginal cost inclusive of tariff. that yields zero profits for the domestic firm is The equilibrium constructed as follows: the first period the domestic firm charges a price, u, (in that period) a price just In so low that its losses are exactly equal to the present value of the profits from the one- shot game from the second period on. slightly higher than u. The foreign firm charges The foreign firm charges this price a price just every period until the domestic firm charges u for one period; thereafter both firms revert to the one -shot outcome each period. For evidence on this fact and some alternative explanations see Deardorff 12 (1986). This hinges critically on the fact that we do not use Abreu's (1982) severe 13 These are however punishments. difficult to characterize in this case. This inequality can be used to demonstrate that the foreign firm always 14 seUs the entire quota when market, the domestic firm amount this it greater. Then by suffices to RHS When deviates. To show show that this inequality is e it that this latter amount less than 15 if e to larger than is Suppose (l/4 + vi/2). This means that of the inequality. impossible response it were would also have to be greater than 1/4 plus expression whose maximum, reached when is the firms are monopolizing the selling (a-c) (l/2-y)/b so that the best (a-c) (l/2 + y)/2b. is the quota half the is it e is e must exceed 1/2 minus an 1/3 and 6 is This equals 1/18. 1, less than 1/3. is Bernhein and Whinston (1986) consider the more general case of imperfect They show correlation. that the sustainability of collusion a monotonically is decreasing function of the degree of correlation across markets, so that we are considering the extreme that makes collusion most sustainable. 16 This result is in contrast to that of consider a model in which the incumbent They Brander and Spencer (1981). is committed to selling its pre-entry output once entry occurs. In that setting, the incumbent can deter entry by producing a large enough output. to prefer to produce Imposing a a lower output tariff on the incumbent may lead even though that will result in However, since the incumbent stUl has the option of detering entry, chooses instead to allow entry it price when it does so. So, must be the case that in their analysis, any it entry. if it receives a higher tariff big enough to induce the domestic firm to enter necessarily raises the domestic price above what it it 39 would have been if however, depends Since there is the domestic firm was deterred from entering. critically on the absence of perfecteness of their equilibrium. no logical link between the incumbent's pre-entry output and post-entry output, the incumbent should be expected entrant's post-entry output. post-entry. 17 This result, to respond optimally This leads to the Cournot outcome in its to the both markets In that case the above analysis applies. Note that this expression for s (the highest price is strictly declining in as are the espressions 2 2 charged by the domestic firnm) and (a-c) (1-e) /b, the So our analysis of competition expected profits of the domestic firm. consistent with that of Krishna (1985). differentiated products. £ is She analsyzes a duopoly producing She shows that, in the single shot static game, reducing a quota raises domestic prices as the domestic firm need not be so concerned with foreign competition. to the case in Our analysis shows which the duopoly produces a that her result extends homogeneous good. REFERENCES Dilip, "Repeated Games with Discounting: A General Theory and an Application to Oligopoly", Princeton University, December 1982, mimeo. Abreu, Bernheim, B. Douglas, and Michael D. Whinston, "Multimarket Contact and Collusive Behavior," Harvard University, April 1986, mimeo. Bhagwati, Jagdish, "On the Equivalence of Tariffs and Quotas" in R.E. Baldwin et eds. Trade, Growth and the Balance of Payments: Essays in Honor of Gottfried Haberler .Noth Holand, Amsterdam, 19G5. al. Brander, James and Barbara Spencer, "Tariff Protection and Imperfect Competition", in Monopolistic Competition and International Trade edited by Henryk Kierzkowski, Oxford: Oxford University Press, 1984. ^ "Tariffs and Extraction of Monopoly Rents of Economics 14 (1981), 371-389. under Potential Entry," Canadian Journal . Brock, WUliam, A. and Jose, A. Scheinkman: "Price Setting Supergames with Capacity Constraints" Review of Economic Studies 52, July 1985, 371-82. , Davidson, Carl, "Cartel Stability and Tariff Policy," Journal of International Economics 17 (1984), 219-237. . Avinash K. and Albert S. Kyle, "The Use of Protection and Subsidies for Entry Promotion and Deterrence," American Economic Review 75 (March 1985), 139Dixit, 152. Friedman, James W. "A Non- Cooperative Equilibrium for Supergames", Review of Economic Studies 28, 1971, 1-12. : . Helpman, Elhanan, "Increasing Returns, Imperfect Markets and Trade Theory," Foerder Institute for Economic Research Discussion Paper 18-82, 1982 Kreps, David M., and Jose A. Scheinkman, "Quantity Precommitment and Bertrand Competition Yield Cournot Outcomes," Bell Journal of Economics 14 (Autumn 1983), 326-337. Krishna, Kala, "Trade Restrictions as Facilitating Practices," Harvard University, October 1984, mimeo. Mookherjee, Dilip and Debraj Ray, "Dynamic Price Games with Learning-By-Doing," Stanford University, March 1986, mimeo. Rotemberg, Julio J. and Garth Saloner, "A Supergame-Theoretic Model Wars During Booms," American Economic Review forthcoming. , of Price- 7 'V %MMeff Lib-26-67 1BRAR1ES ! 3 TDflD D |!|!' M I I 111'! '11 |||| QMS T33