不同國家的反托拉斯法對於卡特爾行為的影響 - 政大學術集成

全文

(1)Department of Public Finance National Chengchi University Master Thesis Advisor: Jue-Shyan Ph. D. 政治 Wang, 大立 ‧ 國. 學 ‧. The Impact of Different Antitrust Laws on the Actions of Cartels n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v. Graduate Student : Yi-Shao Chen. June 2013.

(2) 感謝詞時光飛逝，兩年的研究所光陰就這樣過去了，首先我要非常非常感謝的就是我的指導教授－王智賢老師，老師在我寫論文的過程當中協助我非常多的地方，也因為有老師的鼓勵，我下定決心要用英文來完成我的論文，因為英文不是我的母語，要完成真的需要勇氣，也感恩思頤的鼓勵，讓我有信心動筆論文！回想研究所生活的點點滴滴，雖然有時候覺得時間很不夠、很累，甚至數學推導不對的時候，智賢老師總是悉心地協助我解決問題，真的非常感恩老師的指導，讓原本以為自己不可能研究理論的我完成了賽局研究，也讓我順利投稿上 TEL 期刊！碩士學位是我小時候立下的志向，也因為就讀研究所，我又認識了許多可愛的人們，也感謝同門的好同學：涵喻還有宛萱，以及不斷鼓勵我的葦杭學姊、巧. 治政很大的幫助與支持；感謝我的室友姝廷，總是在忙碌時刻願意跟我聊聊天，彼此大立互相加油打氣！. 芬學姊，以及協助我解決論文排版問題的艾儒、婉萍，你們都是我的貴人，給我. ‧ 國. 學. 感恩我的家人，外公、外婆、爸爸、媽媽、泊屹、幸愔，因為有你們背後默默的支持與鼓勵，才有今天的我，讓我無後顧之憂可以好好在台北念書，完成了感恩我的. 師父－大成就明師. ‧. 我的碩士學位，我是一個非常幸福的小孩。. 妙禪師父，安住我的心，開啟我的智慧，讓. Nat. sit. y. 我在忙碌的研究所生活與社團生活中取得平衡，讓我面對難題不會感到挫折害. io. er. 怕，還有如來實證社的大家們，能夠共心走在回家的路上真的很開心，特地感恩主原護持我的口試，我愛你們！. n. al. Ch. i n U. v. 雖然之後大家就要各奔東西，但我知道研究所的點點滴滴我一定不會忘記，. engchi. 我會永遠記得以及珍惜我求學生涯的這段日子，感謝每一位幫助我的人，你們都是我生命中的天使。. 宜卲. 2013.7.29 於宿舍.

(3) Abstract Title: The Impact of Different Antitrust Laws on the Actions of Cartels School: Graduate Institute of Public Finance National Chengchi University Advisor: Jue-Shyan Wang, Ph. D. Author: Yi-Shao Chen Keywords: Antitrust; Cartel; Leniency Policy; Bayesian Nash Equilibrium. 政治大 countries over the world adopt leniency policies to control the actions of cartels. We 立. The issues about antitrust laws are getting much attention nowadays. And many. ‧ 國. 學. used a game-theoretical model to discuss the equilibrium of cartels under different antitrust laws. And we modify the model of Blum et al. (2008) to analyze the. ‧. equilibriums of firms under the different mechanisms of leniency policies. We find. sit. y. Nat. out that the value of fine will affect the existence of a cartel, and the recognitions of. io. al. er. legitimacies for cartels are important as well. When the antitrust authorities ask firms to propose appliances in advance to let the cartels be legal, firms would incline not to. n. v i n C h from the otherUmechanism quite different engchi. become a cartel. It’s. which firms can. confess to the antitrust authorities after they’ve already become a cartel..

(4) CONTENTS CHAPTER 1. Introduction ................................................................................................... 1 CHAPTER 2. Literature Review .......................................................................................... 4 CHAPTER 3. The Basic Model ............................................................................................ 8 CHAPTER 4. Stable Equilibrium of Firms....................................................................... 11 CHAPTER 5. Conclusion .................................................................................................... 16. 政治大 Reference ................................................................................................................................ 19 立 Appendix ................................................................................................................................ 17. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v.

(5) Tables Table 1. Normal form of the cartel game ................................................................... 9 Table 2. The Bayesian Nash equilibrium of model ................................................. 11 Table 3. Type of games and payoffs .......................................................................... 17. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v.

(6) Figures Figure 1. a  c ....................................................................................................12 Figure 2. a  c   z .............................................................................................13 Figure 3. a  c   z .............................................................................................14 Figure 4. Different regimes of the cartel game .....................................................18. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. i n U. v.

(7) CHAPTER 1. Introduction Much literature discussed about the antitrust laws which are not all the same among the countries. The USA is the pioneer to set the antitrust law. After that, so many other countries passed legislation of antitrust laws, one after another. Antitrust law is also called competition law. It’s often used to keep the market from unfair competition or stop firms forming a private cartel. A cartel is a formal agreement among competing firms. Firms get together to set their business strategies in order to raise their benefit. In this way, firms can definitely expand their magnitude and get. 政治大 social welfare. In order to regulate or forbid collusion (also called cartel agreement), 立. abnormal profit by reducing competition. But these actions can be harmful for the. the antitrust authorities would set their own antitrust laws in view of their own. ‧ 國. 學. circumstances. Leniency policy is a new policy that has been applied in recent years.. ‧. In this paper, we used a game-theoretical model to analyze the equilibrium. sit. y. Nat. belonging to cartel. It isn’t always bad for social welfare when there are cartels in. io. er. society. Sometimes, cartels may be good for the whole society. Hence, the antitrust authorities have to find their own solutions to raise the whole social welfare.. al. n. v i n C hif firms have collusive In Europe commission (EC), agreement and become a engchi U. cartel, they will be forbidden by the Cartel office unless the actions of collusion are. helpful for production or distribution of goods. At the same time, cartels should improve the development of technique and economy and ensure that consumer’s welfare is guaranteed definitely. That is to say, firms won’t be punished if the collusion is beneficial for consumers’ welfare as well as the whole social welfare. And it wouldn’t be necessary to submit an appliance in advance in the most European countries. This kind of mechanism in Europe is much similar with the United States. Antitrust division has a power of prosecutorial discretion which means they can process the cases as they want. Hence there would be case selection. If the antitrust 1.

(8) authorities think the cartels are harmless to the society, they may ignore and drop the cases. The reason is that they think they don’t have enough resource to handle all the cases. In Taiwan, antitrust authority is called Fair Trade Commission. Collusion won’t be allowed basically, but under some particular circumstances, firms can also exempt from the Fair trade Act. The most obvious difference between American law (or EC’s law) and Taiwanese law is that firms must submit appliance in advance no matter what situations in Taiwan. If not, cartels are always illegal without registration. Even if the collusion is beneficial to the whole social welfare, it’s surely unacceptable in. 政治大 In order to control the actions of cartels, the antitrust authorities introduce a 立. Taiwan.. mechanism to influence the strategies that firms would take. Antitrust law has a kind. ‧ 國. 學. of reward program which we call leniency policy or leniency program. Leniency. ‧. policy was introduced for the first time by the United States in 1978 that allowed. sit. y. Nat. firms to be the whistle blowers and to have a reduction of fine payment under some. io. er. specific conditions. Sometimes the whistle blowers can exempt from all the punishment. At first, only a few firms applied for this leniency program because it. al. n. v i n wasn’t very transparent and it C brought some kind ofUuncertainties. Consequently, it hengchi was revised significantly in 1993. Leniency policy is really a key factor for the cartel. office to affect the actions of firms. So the legislations of this kind of laws become much more and more important. Consequently, we discuss whether the equilibrium derived from the game-theoretical model, fit the antitrust law or not. And we will measure the welfare from the standpoint of the whole society. At the end of this section, let’s introduce the framework of this article: section 1 is the introduction that we talk about the history of Antitrust Law and leniency policies, section 2 is the literature review, section 3 is the basic model that we used to 2.

(9) analyze the equilibriums, section 4 is the part about the stable equilibrium of firms’ strategies, and section 5 is the conclusion that we got.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 3. i n U. v.

(10) CHAPTER 2. Literature Review We spend some time to know some papers over the past. Motta and Polo (2003) studied and concluded that the government should let the confessors or the whistle blowers have the full immunity to all the other firms under the model is the equilibrium. And they also put emphasis on that a leniency program may encourage pro-collusive effect as the Antitrust Authority is short of resources. Hence, they leniency program is the second best and should be implemented as the authority has sufficient resources. Ellis and Wilson (2003) assumed that there are three ways firms. 政治大 leniency policy, squeal to the Antitrust Authority and be detected by other forms under 立. can choose: comply with the collusion rules and become a cartel, comply with the. a probability or not be detected fortunately. The studies showed that firms may squeal. ‧ 國. 學. to the cartel office because they want to beat the competitors by lowering their profit. ‧. and raise their cost which caused by the fines. And this is a strategic advantage for the. sit. y. Nat. firms to squeal. They also thought the concentration of market is an important element. io. er. to affect cartel. When the market is under less concentration, leniency policy may destabilize the cartel. On the contrary, leniency programs make firms become stronger. al. n. v i n C h to confess to theUAntitrust Authority. Firms have cartels when they don’t induce firms engchi. the mechanism of self-stabilization. Spagnolo (2004) thought that under dynamic analysis, leniency policy would be useful if the cartel office create incentives for firms to cheat on partners. So the incentive is the super high reward for the first one to squeal. And the reward is from the fines paid by all other parties. When leniency policy is implemented in reality, may also destabilize and deter cartel by (a) protecting the squealers from fines; (b) protecting them from the others agents’ punishment; (c) increasing the riskiness of being a cartel. Compare to the view of Motta and Polo (2003), the difference between them is that leniency is affective only because of immunity by the view of Motta and Polo (2003). Aubert et al. (2006) discussed about 4.

(11) the leniency policy with some brand new view. They compared the impact of reduced fines and positive reward, and argue that rewarding individuals, including employees of firms, may be more effective to deter cartel than traditional leniency policy. Chavda and Jegers (2007) used an oligopoly model to analyze the policy. At first glance, we usually think fines deter the cartel in reason of sapping the benefit and raise the risk involved in illicit behavior. However, their study showed that fines tend to increase the stability of an agreement. If the impunity is granted for firms, cheating is the most profit way. To fight with the collusive behavior, Antitrust Authorities all over the. 政治大 leniency policy. Blum et al. (2008) challenged the contemporary view that standard 立 world use combinations of fines and fine exemptions. That is the effectiveness of. leniency privilege is incentive- compatible to increase competition. They thought. ‧ 國. 學. leniency policy is a preemptive strike for firms against competitors when cartels. ‧. become unstable. The defectors may have more economical privilege in the future.. sit. y. Nat. Accordingly, if the leniency policy would lead to more competition in the market, the. io. er. policy should be welcomed by the national cartel offices. Brisset and Thomas (2004) discussed about the impact of leniency policy on incentives within cartels. And they. al. n. v i n C hthe cartel membersUto confess and implicate their wanted to find a way to encourage engchi. co-conspirators with hard evidence about their collusive agreement. They developed a simple model of cartel behavior under a first-price sealed-bid procurement auction and found whether an effective leniency program can prevent the internal coordination of cartel members. Somehow, they found out that the rules of the actual European program present some inefficiency. For example, any member who reports the cartel before any investigation can has a full exemption from fine. Besides, they also showed that how the leniency program could become an effective tool for the prevention of anti-competitive behavior by rewarding firms that provide hard evidence of the cartel’s existence. Then the leniency programs can create private 5.

(12) incentives to play inefficient firms against efficient ones. Ishibashi and Shimizu (2010) used a model of quantity competition to analyze the leniency policy in antitrust laws. They concluded that an amnesty to the second or later candidates of a leniency program is useless if colluding firms can choose the most profitable collusion. And their result implied that the design of a leniency policy depends on which kinds of market structure are prevalent in the countries. Bigoni et al. (2012) thought there are more countries using leniency policy to control the actions of collusion. And they studied how fine, leniency programs and rewards for whistleblowers affect cartel. 政治大 to observe. In their experiment, 立. formation and prices from an actual experience. But their effects on cartel formation and prices are hard. traditional antitrust law. enforcement without leniency policy has a significant deterrence effect that fewer. ‧ 國. 學. cartels form. But without leniency, surviving cartels’ prices grow, which means the. ‧. overall prices do not fall. Leniency can help fix this problem. Leniency programs. sit. y. Nat. further increase cartel deterrence, but also stabilize surviving cartels relative to a. io. er. laissez-faire regime. So when fines are used as rewards for self-reporting agents, prices fall significantly and antitrust enforcement improves welfare. Hinloopen (2003). al. n. v i n analyzed with a dynamic modelCto know the cartel members’ incentives to report the hengchi U. existence of the cartel to antitrust authorities under leniency programs. And the model incorporates detection probabilities that can differ in each period. It takes time for antitrust authorities to detect a cartel and then to provide enough hard proof for it to be dismantled. The author concluded increasing the reduction in fine payment in return for reporting an illegal cartel and increasing the per-period detection probability in any future period enhances the efficiency. In this paper, we would discuss about the impact of different antitrust laws on firms. For example, the antitrust authorities ask firms hand in their appliance beforehand to be legal. Or after-the-fact policy which means cartels can be legal as 6.

(13) they are good for the whole society even they don’t tell the antitrust authorities first. Blum et al. (2008) only categorized the games corresponded the areas in the figures, but we directly solved Bayesian Nash Equilibriums of the cartel game. And furthermore, we discussed the degree of difficulty to form cartels under different national antitrust laws.. 立. 政治大. ‧. ‧ 國. 學. n. er. io. sit. y. Nat. al. Ch. engchi. 7. i n U. v.

(14) CHAPTER 3. The Basic Model We based on the model of Blum et al. (2008) and we did some reformation. Let’s set a model of a cartel with two suppliers in the market, player 1 and player 2. The two suppliers are two firms competing with each other. We assume the firms as homogeneous under the general condition, which means the two firms have the same condition of abilities. There are two kinds of situation for firms to react. Firms can comply, which means the two firms adhere to their cartel agreement. Still, firms can defect, which means one or both of them would squeal to the antitrust authorities.. 政治大 and be attractive for any party to defect. So, leniency privilege must be known for 立 As for the antitrust authorities, they must want the leniency policy to be useful. every player. Moreover, antitrust authorities would prefer to take side of the whole. ‧ 國. 學. society to maximize social welfare.. ‧. Let’s begin to introduce the meaning of parameters: We assume the market. sit. y. Nat. reward is  1, the cartel surplus is c , the fine for a proven cartel is f , and the. io. er. leniency reduction is w . The first mover has an additional advantage, a , it can be regard as an improved cost structure or getting an increased market share. If leniency. al. n. v i n privilege is not only for the firstCmover but also granted h e n g c h i U to the subsequent confessing. parties, the authority may offer the confessors a reward, z , instead of w . As we can see, z applies to the second confessor in the model, and z  [0, w] . And note that z is non-negative because it would reduce the fine, f . And p is the probability of a confession being processed by the antitrust authority which might be less than 1 because of work overloaded. p  (0,1) . And the normal form of the game is presented in Table 1.. 1. The market reward of oligopoly is  , and c is the cartel surplus (net effort to sustain the cartel). Total payoff of a firm is   c When the market is Monopoly, the total advantage of a firms is. 2(  c) .. 8.

(15) Table 1. Normal form of the cartel game Player 2 Comply Comply. Player 1. Defect. Defect. (1’). A1 , A2. (3’). C1 , C2. (2’). B1 , B2. (4’). D1 , D2. 政治大. 立. If the equilibrium is (1'), it means there is a stable cartel. If the equilibrium is (2'). ‧ 國. 學. or (3'), it means one part of them defects and be the crown witness. If the equilibrium is (4'), which means that there would be a price war and cartel fine.. ‧. In Table 1, the definitions of the symbols are as below:. y. sit. n. al. er. io. A2    c. Nat. A1    c. Ch. B1  p(a  w  f  c)    c. engchi. i n U. v. B2  p( z  f  c)    c C1  p( z  f  c)    c C2  p(a  w  f  c)    c D1  p 2 ( f  c  a  z )  p (a  w  z  2 f  2c)    c D2  p 2 ( f  c  a  z )  p (a  w  z  2 f  2c)    c. We will show how to calculate the value of B1 , B2 , and D1 . When we want to. 9.

(16) calculate the expected values, we also have to take the probability of a confession not being processed, 1  p , into consideration. Let’s take B1 as an example first. As player 1 goes to the authority to confess the truth of collusion, it has a probability of. p to get the value   a  f  w . And if the confession is ignored by the authority, it’s like the same situation that the firm never confesses to the cartel office. So the firms will get the payoff   c with a probability 1  p . Hence, if we want to obtain the expected value of B1 , we have to calculate as the following equations. B2 is obtained by the same method, firm 1 has a probability of p to get the value of.   f  z and probability of 1  p to get   c . As for D1 , it’s a little more. 政治大 complicated. We have to consider about the reaction of firm 2. Both of the two firms 立 confess to the antitrust authority. So when firm 1 thinks the probability whether it’s. ‧ 國. 學. squeal be processed or not, it still have to think about if the opposite’s confession be. ‧. taken or not. So the equation is calculated as follows.. y. Nat. B1 : p(  a  f  w)  (1  p)(  c)  p(a  w  f  c)    c. sit. B2 : p(  f  z )  (1  p)(  c)  p( z  f  c)    c. n. al. i n U.  p 2 ( f  c  a  z )  p(a  2 f  w  z  2c)    c B1  C2. ’. B2  C1. ’. D1  D2. Ch. engchi. er. io. D1 : p[ p(  f  w)  (1  p)(  a  f  w)]  (1  p)[ p(  f  z)  (1  p)(  c)]. v. We didn’t only care about the equilibrium payoff of two firms, but also put emphasis on the consumers’ surplus. That’s because we take the stance of the social welfare. And we will discuss it in the later section.. 10.

(17) CHAPTER 4. Stable Equilibrium of Firms In this section, we’re going to introduce the equilibrium, and Bayesian Nash equilibrium is used to analyze the model. Meanwhile, to avoid the complexity to blur the discussion, we based on the results of pure strategy. And the conditions of equilibrium are listed in Table 2. And we also specifically introduce different types of games to explain the equilibrium payoff and have an interpretation. And we can see this part in the appendix. Table 2. The Bayesian Nash equilibrium of model Equilibrium. 立. 政Condition治of Equilibrium 大. f  wac. (2’). ( B1 , B2 ). f [. w pz ac , w  a  c] 1 p 1 p. (3’). (C1 , C2 ). f [. w pz ac , w  a  c] 1 p 1 p. (4’). ( D1 , D2 ). f . ‧. ‧ 國. ( A1 , A2 ). 學. (1’). sit. n. er. io. al. y. Nat. w pz ac 1 p 1 p. i n U. v. We take the equilibrium (1’) as an example to show how the values of payoff. Ch. engchi. came out. As ( A1 , A2 ) is an Bayesian Nash Equilibrium, two conditions must be fulfilled at the same time, that is A1  B1 and A2  C2 . And we show the computational. processes. as. following. sentences.. A1  B1. means.   c  p(a  w  f  c)    c . Then we can get p(a  w  f  c)  0 . Finally we can simplify it as f  w  a  c . And the rest equilibriums are similarly derived with the same method. After we knew how to derive the condition of equilibriums, let’s draw figures to portray the areas of the Bayesian Nash equilibriums. Let w be the horizontal axis and f be the vertical axis. Figure 1 shows the equilibriums as a  c . We have 11.

(18) two lines to distinguish the areas of the Bayesian Nash equilibriums, line L1 and L2 . L1 : f  w  a  c L2 : f . w pz ac . 1 p 1 p. And the point e(  ( z, z  a  c)) is the intersection of two lines, under a  c . Here we show the Figure 1 as below:. Figure 1. a  c. f. (1') L1. (1'). (4 '). . e 立. 政治大. ( z , z  a  c). ‧ 國. 學. (2') (3'). io. w. n. al. er. Figure 1. a  c. Ch. sit. y. Nat. 0. ‧. (4'). L2. engchi. 12. i n U. v.

(19) Let’s talk about another situation under different condition. And Figure 2 is the situation when a  c and a  c   z . Because of a  c   z , the horizontal axis value of e is zero. There are also tow lines, L1 and L2 , to distinguish the equilibrium areas. And e is ( z , 0) . Under a  c   z , there is no equilibrium area for (2') and (3'). That is to say, there is no circumstance for two firms that one firm complies with the cartel agreement and the other one defects. Here we show the Figure 2 as below.. Figure 2. a  c   z. f. L2. (1'). (1'). ‧. (4'). (4'). n. er. io. e a l. ( z,0). sit. y. Nat 0. L1. 學. ‧ 國. 立. 政治大. Ch. engchi. iv n Uw. Figure 2. a  c   z. The third situation is the condition when a  c   z . And it will be shown in Figure 3. And the intersection of two lines, e , lies below the horizontal axis because of a  c  z .. 13.

(20) Figure 3. a  c   z. L2. f (1'). (1'). L1. (4'). (4') 0. w e  ( z , z  a  c). 立. 政 a治  c   z大. Figure 3. ‧ 國. 學. And what we want to focus on is the two firms’ stable equilibrium. When f is large, it’s easier to exist the equilibrium of being a cartel because of firms don’t want. ‧. to take any risk of being punished；On the contrary, when f is small enough, the. y. Nat. io. sit. equilibrium of both two firms’ defecting strategies would happen more likely. The. er. Bayesian Nash equilibriums may be summarized as the first two propositions in the. al. n. following.. Ch. engchi. i n U. v. 【Proposition 1】When f  max[w  a  c,( w  pz ) (1  p)  a  c] , (1') is an unique Bayesian Nash equilibrium for two firms to be a cartel. 【 Proposition 2 】 When. f  [ w  a  c,( w  pz ) (1  p)  a  c] , there are two. equilibrium (1') and (4') in this area. When the two (1') and (4') are the equilibriums at the same time, we call that coordinate game. Generally speaking, it depends on the value of A1 and. D1. to. know. which. one. is. easier. to. be. the. equilibrium.. As. f  [(1  p)(a  z )  w]/(2  p)  c , then D1  A1 , (4') would be the equilibrium more. 14.

(21) easily. On the contrary, it’s more likely to exist the equilibrium (1'). 【Proposition 3】When a  c   z , it won’t exist the equilibrium of (2') and (3'). If firms are aware of their own benefit is too small for them to confess to the authority, they have no incentive to defect. Hence, there are no equilibrium of (2') and (3'). And it is shown in Figure 2. Besides, the legitimacies of cartels in Taiwan and in America (as well as European countries) are different. In Taiwan, if firms don’t apply to the authorities, the actions of collusion would be recognized as illegal definitely. But in America or in. 政治大 cartels. Therefore, we have立 to compare the two different mechanisms under the same Europe, the legitimacy depends on the effect after the firms have already formed the. ‧ 國. 學. condition. We assume that collusion is the most efficient strategy first, which means the value of two firms’ welfare plus consumers’ welfare would be the highest of four. ‧. possible equilibrium values. Under the American or European laws, even the firms. sit. y. Nat. don’t apply to the cartel offices in advance, the probability of being processed by the. n. al. er. io. cartel offices, p , would still be regarded as zero. However in Taiwan, if firms don’t. v. apply for the permission of cartels to the authorities beforehand, p would be. Ch. engchi. i n U. definitely 1. When p  0 , firms’ equilibrium payoffs become   c under the antitrust laws of America or Europe. So there is no limitation for firms of being a cartel at all; When p  1 , it must fulfill the condition that f  w  a  c at least to form cartels. Hence, comparing the two kinds of different mechanisms, cartels are easier to exist in America (or in Europe) than in Taiwan. And here above is proposition 4. 【Proposition 4】It’s easier to exist cartels in America or Europe than in Taiwan.. 15.

(22) CHAPTER 5. Conclusion Blum’s paper only discussed about the equilibrium areas of different regimes of the cartel games, but we made some extensions. In this paper, we used a game-theoretical model to analyze the equilibriums of two firms’ strategies in a cartel. And the amount of fine payment is the key factor to affect two firms’ equilibrium. Different countries would have different p , which means the probability of a confession being processed by the antitrust authority. And that is issue what we are focus on.. 政治大 society. That is to say, the antitrust authorities will not interfere too much to forbid 立 The actions of firms being a cartel would be allowed if it is good for the whole. cartels all the time. However in Taiwan, antitrust authorities will not permit a cartel. ‧ 國. 學. absolutely if firms do not make applications in advance. The conclusion is that cartels. ‧. are easier to appear in America (or Europe) than in Taiwan. Hence, how to enact the. sit. y. Nat. antitrust laws becomes a very important issue to the cartel office.. io. er. This paper is surely to provide a point of view for antitrust authorities to set the policies of leniency programs. And it may arouse more people to study about this. al. n. v i n C h authorities would issue in the future and then the antitrust control the actions of cartels engchi U. more effectively. Maybe in the future, we can make a extension that leniency reduction is a proportion of fine payment, which means w  sf , s [0,1] . To see what would happen in this kind of situation.. 16.

(23) Appendix There are three kinds of games we would define, i.e. prisoners’ dilemma game, chicken game, and coordinate game (assurance game). For the row player, it has to fulfill the condition B1  A1  D1  C1 , and we can separate the condition as B1  A1 , A1  D1 , and D1  C1 at the same time. For chicken game, row player’s value must. achieve B1  A1  C1  D1 , and we can separate it as B1  A1 , A1  D1 , and D1  C1 . As for coordinate game, the payoff of row player must fulfill A1  D1  C1 and A1  B1 . And we can regard them as three conditions, A1  D1 , D1  C1 , and A1  B1 . We show it in the following Table 3:. Player 1. player2. 學. Type of game. ‧ 國. 治政 Table 3. Type of games and payoffs 大立 B1  A1  D1  C1. C2  A2  D2  B2. Chicken game Coordinate game. B1  A1  C1  D1. C2  A2  B2. A1  D1  C1 and A1  B1. ‧. Prisoners’ dilemma. D2. A2  D2  B2 and A2  C2. sit. y. Nat. n. al. er. io. Blum et al. (2008) use a figure with horizontal axis and vertical axis to portray. i n U. v. these three games and the situation. As they take the probability of a confession being. Ch. engchi. processed by the antitrust authority, p , into consideration, we think there are something that we can modify and make an improvement. The way that Blum put the probability into the formulae is too simple. According to the conditions, we will portray the three kinds of game. We set the conditions of the figures under a  c and a(1  p)  z . B1  A1 is forming the condition of the line L1 , D1  C1 represents L2 , and A1  D1 represents L3 . We will explain how would the condition, a(1  p)  z , come out next. L2  L1 . pz 0 1 p. 17.

(24) L3  L1 . a  ( p  1) z 0 p2. L2  L3 . a(1  p)  z 2  3 p  p2. It’s ambiguous whether the value a(1  p)  z exceeds zero or not, so we can discuss this under the different conditions. And we take the conditions a(1  p)  z as an example. Figure 4 portray the equilibrium area of the three kinds of game.  1 is the area of prisoners’ dilemma game. And the equilibrium area of prisoners’ dilemma game must fulfill B1  A1 , D1  C1 , and A1  D1 .  2 is the equilibrium area of chicken game,. 政治大. and the conditions are B1  A1 , D1  C1 , and A1  D1 . The last one is the coordinate. 立. game, the equilibrium of this game must fulfill B1  A1 , D1  C1 , and A1  D1 . We. ‧ 國. 學. get  3 is the equilibrium area of coordinate game.. And we can do further discussion under the condition that a(1  p)  z or. ‧. a(1  p)  z . When the conditions are different, the distribution of the equilibrium. n. al. er. io. sit. y. Nat. won’t be the same.. i n U. Figure 4. Different regimes of the cartel game. Ch 3. f . 2. e. L2. eL1n g c h i. v. ( z , z  a  c). 1. L3. w. 0 Figure 4. Different regimes of the cartel game. 18.

(25) Reference Aubert, C., P. Rey, and W. E. Kovacic (2006). “The Impact of Leniency and Whistle-blowing Programs on Cartels,”International Journal of Industrial Organization, 24, 1241-1266. Bigoni, M., S. Fridolfsson, C. Coq, and G. Spagnolo (2012). “Fines, Leniency, and Rewards in Antitrust,” Journal of Economics, 43, 368–390. Blum, U., N. Steinat and M. Veltins (2008). “On the Rationale of Leniency Programs: A Game-theoretical Analysis,” European Journal of Law and Economics, 25,. 政治大 Brisset, K., and L. Thomas (2004). “Leniency Program: A New Tool in Competition 立 209-229.. Policy to Deter Cartel Activity in Procurement Auctions,” European Journal of. ‧ 國. 學. Law and Economics, 17, 5-19.. y. Nat. Cartel Stability,” Metroeconomica, 58, 231-243.. ‧. Chavda, A. and M. Jegers (2007). “The Effects of Leniency Programs and Fines on. er. io. sit. Ellis, C., and W. Wilson (2003). “Cartels, Price-fixing, and Corporate Leniency Policy: What Doesn’t Kill Us Makes Us Stronger. Manuscript, University of Oregon.. al. n. v i n C h Analysis of Leniency Hinloopen, H. (2003). “An Economic Programs in Antitrust engchi U Law,” De Economist, 151, 415-432.. Ishibashi, I., and D. Shimizu (2010). “Collusive Behavior under a Leniency Program,” Journal of Economics, 101, 169-183. Motta, M. and M. Polo (2003). “Leniency Programs and Cartel Prosecution,” International Journal of Industrial Organization, 21, 347–379. Spagnolo, G., “Divide et Impera (2004). “Optimal Deterrence Mechanisms Against Cartels (and Organized Crime),” FEEM working paper, Milano.. 19.

(26) 不同國家的反托拉斯法對於卡特爾行為的影響國立政治大學財政學研究所碩士班研究生:陳宜卲指導教授:王智賢. 論文提要:. 立. 政治大. 反托拉斯法在近期以來已受到越來越多的關注與重視，而為了全球經濟的穩. ‧ 國. 學. 定與公平，近年來有越來越多的國家採用反托拉斯法裡面的寬恕政策來控制企業壟斷的行為。本文利用賽局理論的模型來分析在不同國家的反托拉斯法之下會產. ‧. 生的均衡對應為何，並且我們以 Blum et al. (2008) 的模型加以修改來分析在不同. y. Nat. sit. 寬恕政策的機制之下，對於廠商們壟斷行為的可能影響。我們研究發現，對於壟. n. al. er. io. 斷行為的罰款金額大小會影響壟斷行為的存在與否，而各個國家對於壟斷行為的. i n U. v. 合法性認定也會是一個非常重要的因素，我們研究得出的結果是，當一個國家的. Ch. engchi. 主管機關要求廠商一定要事先提出壟斷行為(卡特爾聯盟)申請，才能夠使聯合行為合法的時候，廠商們會傾向決定不要採取聯合壟斷的行為。. 關鍵詞: 反托拉斯、卡特爾、寬恕政策、貝氏 Nash 均衡. 20.

(27)