Transaction Cost Economics in International Relations: The Case of International Antitrust Enforcement

(1)

Transaction Cost Economics in International

Relations: The Case of International Antitrust

Enforcement

By Martin Holterman, LL.M. University of Twente

(2)

1. Introduction

In 1934, economist Ronald Coase started asking himself a very simple question: If markets are so efficient, how come there is such a thing as a firm?1 His answer was so simple, that to an economist it is almost a tautology: firms exist because in many cases organising something within a firm rather than between firms is simply more efficient.2 This insight allowed

economists to understand a category of behaviour that they had never before been able to suc-cessfully penetrate. If economic theory shows that the firm can get the best value from its suppliers by making them bid for the contract, why would it vertically integrate by merging with one of them, thus locking itself in to that one supplier? The answer is that the costs of negotiating the contract and monitoring its execution are higher than the value gained in this way, even if you take into account the continued need to monitor performance within a single firm. In the language of Coase: by vertically integrating, the firm lowers the transaction costs associated with the transaction by so much that this gain outweighs the loss in production ef-ficiency.

Even though this 1937 article was at first “much cited and little used”3, it started off what is now known as neo-institutional economics, the branch of economics that tries to understand the reason for the occurrence of a myriad of institutions that seem to restrain the freedom of economic actors. The most important researcher in this field was Oliver Williamson, who, in a series of books and articles, successfully managed to explain how the forces of competition work to align transactions with their optimal governance structures. He did so by describing a number of key characteristics of each, and explained how they could be matched to achieve optimum efficiency. By the 1980s, his research was verified by detailed studies of the rela-tionships between electricity plants and their coal suppliers4 and the relationships between car manufacturers and their suppliers.5

By then, economists had set their sights to even more ambitious goals. In the 1990s, they started thinking about using the tools of transaction costs economics to explain other institu-tions than simply those created by companies on a competitive market. In 1999, Williamson

1_{Throughout this thesis, the word company will be used to describe the legal concept of a legal person, i.e. an}

organisation that is incorporated. A firm, on the other hand, is an organisation under unique leadership, which can consist of any number of corporations.

2_{Coase (1937), Williamson (1988), p. 65. For a discussion of Coase’s thinking in this period, cf. Coase (1988a,}

b, c).

3_{Coase (1988c)}

(4)

discussed the possibility of privatising the US State Department6, while others analysed the potentially even more intriguing case of the US prison system.7 However, in so doing, they started to reach the limits of where their science could take them.8

When Ronald Coase invented the Coase theorem, he took the rationality assumption to its very limit.9 In a world without transaction costs, rationality reigns supreme. Williamson chose a different path. He introduced the concept of bounded rationality in his work, setting limits to the kind of knowledge actors had and the kinds of calculations they could make.10 The con-cept of bounded rationality was originally introduced by Herbert Simon, who made it more concrete by describing satisficing behaviour as an alternative to optimising.11 However, economists are used to solving problems by calculating a specific outcome based on the as-sumption that the actors involved in the problem optimise something12, making bounded ra-tionality a very unpleasant assumption for modelling. This can be seen by comparing the work of Williamson, whose models are expressed only in words, to that of Grossman, Hart and Moore, who tried to formalise his work.13 While they still adhere to the assumption of bounded rationality, actors are assumed to be able to foresee the consequences of their deci-sions in a way that seems far removed from Simon’s original suggestions. This is a point that will not be developed further in this thesis.

Another problem arose when the institutional economists started describing the contract-ing problem of the government. The government can certainly be modelled as optimiscontract-ing something, but this optimising problem will clearly include some variables that are derived from sociology or political theory. When Williamson looked at the State Department, he called this variable probity, and modelled it as an additional characteristic of transactions. Unfortunately, it is not quite clear what he had in mind. Ruiter has suggested that probity is nothing more than good faith.14 However, good faith is probably better viewed as a source of rules that serve to partially resolve the problem of contractual incompleteness, rather than as the defining characteristic that expressed the difference between “sovereign transactions” and

6_{Williamson (1999). He already briefly discussed the application of TCE to public governance in Williamson}

(1997).

7_{Hart, Shleifer & Vishny (1997)}

8_{For a non-technical review cf. Ferris & Graddy (1998)}

9_{Coase (1960). Cf. Steenge (2004) for a discussion of the different formulations of the theorem, none of which}

comes from Coase himself. Different versions can be found in Cooter (1987), de Meza (1987) and Usher (1998).

10_{Williamson (1985), p. 43 – 67, Williamson (1996), p. 8 – 9.}

11_{Simon (1960, 1978). It should be noted that bounded rationality can refer to many different kinds of}

assump-tions about behaviour, only some of which can be described as satisficing.

12_{Cf. Dixit (1990)}

(5)

any other transaction.15 Therefore, the framework that is outlined in this thesis will show how, in some cases, additional restraints are necessary to restrain the actor’s inherent tendency to-wards opportunism to a level akin to good faith. This is one point where the present thesis adds new material to the received literature.

To solve this theoretical quandary, the present thesis will discuss a particular subset of government transactions: the transactions between states. In international relations, states still act based on non-economic restraints, and they still optimise an objective function that is dif-ficult to define in the manner that economists are accustomed to do, but at the same time they act more like regular economic actors. While, in their internal business, many contracts be-tween the state and a private person come in the form of regulation and legislation, the inter-national treaties that are concluded between states can much more readily be viewed as nor-mal contracts. Also, within the state the problem arises of the role of the courts: to what extent are these so-called contracts between the state and a private actor enforceable in court. In in-ternational relations this problem is wholly absent, because there is no court that could rule on such a question. In other words, treaties are not enforceable at all. This is particularly fortu-nate because many of the models that have been proposed by neo-institutional economists assume that court intervention is impossible or at least costly. Therefore, it is likely that these models are more appropriate to describe the behaviour of states in their international relations than their behaviour generally. Finally, a focus on international relations allows us to escape the dubious role of the citizen in the question of privatisation: at the same time constituent and consumer.16

In what follows, I will attempt to apply the existing analytical framework to explain state be-haviour in the field of international relations. More in particular, the question is whether it is possible to develop a rigorous analytical framework to explain the level of international inte-gration that is chosen. Of course, this is a subject that has been studied by historians and po-litical scientists for a long time17, but now the question is whether it is possible to approach the subject from the point of view of economics. As already indicated, the present thesis will inevitably involve the use of non-economic variables, but it will be interesting to analyse the tradeoffs involved here. It is likely that the descriptive power of the model will increase as

15_{Williamson (1999), paragraph 5.}

16_{Offe (1987), p. 501. It should be noted that the two most prominent discussions in this vain, Williamson}

(1999) and Hart, Shleifer & Vishny (1997) both concern cases where the role of the citizen as consumer is very modest.

(6)

more “alien” elements are introduced, while the effect on the model’s prescriptive power is uncertain. In other words, as more political or sociological elements are introduced, the model will be more successful at explaining why things are the way they are, but less successful at predicting how they will be given a change in circumstances, or describing how things ought to be. Compared with earlier work, this thesis applies a different and narrower definition of transaction, which requires that the optimisation be taken over production efficiency and transaction efficiency together18, rather than simply stating that the choice of governance structure is driven by the attempt to minimise transaction costs, as in earlier work.19 Another key difference with the received literature is the role of probity, and its relationship with regu-lar institutions. At the same time, the problem is chosen so as to minimise the role of courts and ordinary citizens, which would serve to complicate the analysis in other cases. Moreover, in the case of international relations, it is likely that state actors will display a high degree of rationality, due to the careful preparation that goes into state behaviour in this field, and due to the fact that state behaviour is subject to a higher degree of competition in international relations than in most of its domestic policy choices.

In order to tie the model as closely as possible to the empirical observations of state be-haviour, this thesis will be based on a single dominant case study, which will provide material for illustration and inspiration for theoretical development. The issue chosen is the question of antitrust enforcement. In an increasingly globalised world, antitrust laws are still enforced by national authorities alone. The next chapter will discuss how this is done, and how this puts them at a disadvantage relative to the objects of their investigations. Only once this case study has been sufficiently elucidated, will I continue by describing the models of Williamson and others in chapters 3 and 4. Here, I will also have occasion to carefully define the key terms of the model that have so far been used only casually. Throughout, these definitions will be linked to the phenomena observed in the case study. In chapter 5, the model will be formally applied to the case study. Chapter 6 will analyse to what extent the problems of the case study have been answered satisfactorily, whether it is possible to extend this framework to other issues in international relations and whether there are other economic theories that would work better. Finally, chapter 7 concludes.

18_{North & Wallis (1994) come closest to such an assumption, since they view institutional change over a time}

period long enough that technical change inevitably becomes relevant.

(7)

2. International antitrust enforcement: a case study

In order to imagine the problem of international antitrust enforcement, imagine the European Commission’s current case against Microsoft. While the Commission wields the power of Leviathan20, there is no question that in this game the Commission plays the role of David to Microsoft’s Goliath. In money terms, DG Competition had a budget in 2006 of € 97,5 mil-lion21, which is equivalent to the profit made by Microsoft in 2005 every 3½ days.22 What’s more, only a small fraction of the company’s personnel and resources are located within the European Union, placing much of the evidence outside the Commission’s reach. Also, any fine imposed by the Commission or by the Community courts will have to be executed against Microsoft assets that are located outside the Union.23 Many of these problems can be solved with the usual tools of national or international law. The Commission could use 28 US 1782, which would allow it to ask a US District Court to force Microsoft to produce documents that are in its possession.24 Alternatively, it could rely on the Mutual Legal Assistance Treaty (MLAT) that is in place between the EU and the US.25 However, any short study of these tools will show that the Commission would never be able to investigate as effectively as it would within its own jurisdiction, giving Microsoft a legal advantage in addition to its finan-cial one. To overcome this problem, it is possible to create an international legal regime spe-cifically for antitrust. Such a regime could vary from the type of mutual assistance treaties that are already in place, to a supranational organisation that would enforce antitrust laws globally in the same manner that the European Commission enforces the law in Europe. The advantages and disadvantages of various levels of integration will be discussed in subsequent chapters, but first I will explain the current regime, starting with multilateral treaties, before continuing with bilateral ones and with unilateral efforts.

20_{Hobbes (1651), CITE}

21_{For the past five years, the appropriations were: 2006: € 97,5 million, 2005: € 88,8 million, 2004: € 81,4}

mil-lion, 2003: € 75,9 million and 2002: € 73,9 million. Source: DG BUGDET website. For comparison, the budget of the Dutch competition authority (NMa) in these years varied between € 32 million and € 39 million, while the budget of the competition division of the UK’s Office of Fair Trading lay between £ 25 million and £ 30 million.

22_{Microsoft’s 2005 net income was $ 12,599 million. Source: Microsoft 2005 annual report.}

23_{The Commission imposed an original fine of € 497.196.304,- on March 24, 2004. (Commission Decision}

C/2004/900, d.d. 24 March 2004) and an additional fine for non-compliance with the original decision of € 280,5 million on July 12, 2006 (Commission Decision C/2005/4420 d.d. 10 November 2005, and press release ME-MO/06/277, d.d. 12 July 2006.)

24_{Cf. Intel v. AMD, Supreme Court June 21, 2004, 542 US 241, 124 S.Ct. 2466.}

25_{Agreement on mutual Legal assistance between the European Union and the United States of America, OJ d.d.}

(8)

Multilateral Treaties in Antitrust Enforcement

With the United Nations more concerned with establishing a New International Economic Order26 than with policing the old one, the natural multilateral forum for antitrust enforcement cooperation is the World Trade Organisation. While the WTO’s main goal is the reduction of barriers to trade27, the Doha declaration that initiated the current negotiating round included a proposal for negotiations on antitrust.28 Even though as a result of this proposal a working group was set up to investigate this matter farther, subsequent Ministerial Conferences did not devote much attention to it.29 Even though the initial proposals were not very ambitious, it is safe to say that it is unlikely that any future agreement resulting from the current negotiating round will contain any reference to antitrust matters.

An important reason why the High Contracting Parties of the WTO are less than keen to create a WTO antitrust agreement is the success of the International Competition Network. The ICN is a collaboration between the competition authorities of 82 countries, both develop-ing and developed.30 Its aim is strictly intergovernmental: by organizing work shops and de-ciding on best practices, it hopes to improve antitrust enforcement throughout the world. Be-cause its membership consists solely of antitrust authorities, with no involvement for legisla-tures or governments, no attempt at harmonization or supranational international collaboration is possible even if the members wanted to. What is possible is the encouragement of lower level agreements regarding collaboration between agencies. While cooperation between agen-cies has undoubtedly been improved by the informal contacts created at the ICN's annual meeting, international collaboration has so far failed to get its own place on the agenda. Nev-ertheless, significant progress has been made in achieving improved enforcement of antitrust legislation in the fields of merger review and cartels, each of which has its own ICN working group. It appears that legislators and government officials in many countries recognize the value of this work, and therefore think that this is the best road to improvement for the time being.31

26_{UNGA resolutions 3201 and 3281, dated May 1, 1974 and December 12, 1974, respectively.} 27_{Agreement Establishing the World Trade Organization, articles II and III.}

28_{Declaration of the Doha Ministerial Conference, paragraphs 23 – 25. Cf. Anderson. & Jenny (2002)}

29_{Draft Declaration of the Cancún Ministerial Conference, Annex E, paragraph 2. The Declaration of the Hong}

Kong Ministerial Conference did not contain any reference to antitrust matters. Cf. Fox (2003) and OECD (2003).

30_{Cf. www.internationalcompetitionnetwork.org and the Annual Reports on Competition of the Commission of}

the European Communities, 2002 - 2005.

31_{For example, the European Commission's report for 2004, op cit, p. 183: "It is now widely recognized as a}

(9)

Bilateral Treaties

Apart from multilateral efforts in international antitrust, an important category of international effort in the field are the bilateral treaties and agreements. Such treaties are extremely com-mon, with new ones being agreed each year. The most important agreements of this kind are the those between the EU and the United States of 199132 and 1998.33 The advantage of such an arrangement is that it enables the state to, in each case, surrender just as much sovereignty as it wishes. For example, the United States are generally reluctant to accept supranational institutions such as the International Criminal Court; they prefer intergovernmental

ar-rangements.34 The European Union, on the other hand, is itself a supranational institution, and there is no reason why it would have a problem, on principle, with surrendering its antitrust authority to a new global institution. By using bilateral treaties, the EU can create a very in-tergovernmental framework for its relations with the US, while creating a much stronger re-gime with some of its other trading partners. A multilateral framework would always be lim-ited to the lowest common denominator, because the regime can only be as strong as all par-ticipating states find acceptable and beneficial. A framework based on bilateral treaties is much more flexible.

In order to better understand the usual scope of such a treaty, it is useful to take a closer look at the aforementioned 1991 EU-US Agreement. It contains obligations such as the obli-gation to notify each other when enforcement activities affect the “important interests” of the other party, including detailed rules about the exact moment the notification has to occur, and examples of cases where an important interest would normally be affected35, the obligation to share information and coordinate their efforts36, a more detailed arrangement for the case where anticompetitive behaviour in one Party’s territory aversely affects the other37, as well as a detailed framework to avoid conflicts over enforcement in case both Parties pursue the same case.38 Significantly, the approach chosen does not involve any conflict resolution mechanism, but merely the mutual obligation to give due regard to the other Party’s legitimate interests.

32_{Agreement between the Government of the United States of America and the Commission of the European}

Communities regarding the application of their competition laws, OJ 1995 L 095, p. 0047 – 0052, d.d. 27.4.1995.

33_{Agreement between the Government of the United States of America and the European Communities on the}

Application of Positive Comity Principles in the Enforcement of their Competition Laws, OJ 1998 L 173, p. 0028 – 0032, d.d. 18.09.1998.

34_{Cf. Delrahim (2004a, b) and Guzman (2003)} 35_{Article II}

36_{Articles III and IV} 37_{Article V}

(10)

In addition, the subsequent 1998 Agreement requires the EU and the US to come to each other’s aid, if such assistance is requested, for example by investigating and remedying an antitrust violation39 and it creates the possibility that one party may suspend or defer its en-forcements activities because the other party is already in the process of solving the problem. However, the Agreement merely creates the possibility that this will be done, followed by a few properties of cases where this will “normally” be done.40 It should be noted that, strong as this language may be, it does not create an obligation.

In practice, the transatlantic cooperation in antitrust matters leaves somewhat to be de-sired. In a number of important cases, both countries’ competition authorities have disagreed. Examples vary from mergers that were blocked by one, while being allowed by the other41, to cases where one state prosecutes a dominant firm shortly after it has reached a settlement agreement with the other.42 In general, both countries carry out their own merger review, which makes this the field where the differences between them are most obvious.

Unilateral legislative action

Finally, in addition to international law, states could facilitate antitrust enforcement by giving their laws a degree of extraterritorial effect. What should be distinguished here is the question where the unlawful act took place, and whether the person who incurred damage as a result is a resident or a citizen, or whether they are a foreigner. Bearing this in mind, the most restric-tive option is that the laws only protect citizens against unlawful acts committed within the jurisdiction.43 This is a solution that is not normally observed. After all, it would be consid-ered quite immoral if the criminal code outlawed the murder of a citizen, but not the murder of a foreigner. Instead, the criminal laws usually outlaw certain behaviour towards everyone, as long as that behaviour takes place within the jurisdiction. Dutch laws outlaw the murder of anyone, but have nothing to say about a murder that takes place in Germany. Similarly, article 81 EC outlaws cartels that affect competition within the Common Market, without setting any conditions as to the range of people this law seeks to protect. That is because it protects eve-ryone who operates on the Common Market, i.e. within the EU.

39_{Art. III of the 1998 Agreement.} 40_{Art. IV (2) of the 1998 Agreement.}

41_{The merger between General Electric and Honeywell was blocked by the European Commission in July 2001,}

after the US Department of Justice had indicated they had no objections. Cf. Wood (2003). This was the first time the European Commission blocked a merger between two American companies.

42_{Eg Microsoft case, chapter 4, below.}

43_{The term jurisdiction is used to describe the right of a court to rule on a particular case as well as to refer to the}

(11)

So far the situation seems pretty clear. But what about European citizens trading in Rus-sia? Could they claim protection from the EC Treaty?44 At first sight, the answer would have to be no. On the other hand, a clever lawyer could make the argument that a cartel in Russia affects the trade within the Common Market indirectly, if the customers of the cartel sell the product within the EU. What’s more, what about a cartel that is truly global in its effect, cov-ering both the Common Market and the rest of the world. Could a customer of such a cartel sue for damages under EU law even if she bought good from the cartel in Japan? While this problem has not yet come up before the EU Court of Justice45, the US Supreme Court ruled on just this problem in 2004.46

In the years before, a global cartel had been uncovered that had controlled the global vi-tamins market for over a decade. When it was uncovered, the participants received huge fines under both US and EU law.47 Moreover, a large number of private victims of the cartel brought a case before the US courts, using the rule laid down in 15 USC 16, which states that a final judgement or settlement in a case brought by the Department of Justice counts as

prima facie evidence in a subsequent private action, meaning that the defendant firm would

have to disprove the charges, rather than having the plaintiff prove them. In response, the de-fendant companies settled with the American plaintiffs, while the non-US plaintiffs ended up in the Supreme Court. The rub was whether what the defendant companies had done should be considered a single “act”. If so, than this act was clearly in violation of US law, and all persons who suffered as a result would be able to recover their damage under 15 USC 15.48 However, the Supreme Court, as well as the lower courts, did not see it that way. Instead, they treated the defendants’ behaviour with regard to the plaintiffs as acts committed abroad, and

44_{Ryngaert (2004)}

45_{The Court has specified its criterion for jurisdiction in another case: [I]t must be possible to foresee with a}

sufficient degree of probability on the basis of a set of objective factors of law or fact that the agreement in ques-tion may have an influence, direct or indirect, actual or potential, on the pattern of trade between Member States. Case 56/65, Société Technique Minière, [1966] ECR 235 at 249 and Case 89/85, In re Wood Pulp, [1988] ECR I-5193. This language seems to suggest that an indirect influence, as described in the text, would be sufficient.

46_{Hoffman-La Roche v. Empagran, Supreme Court June 14, 2004, 542 US 155, 124 S.Ct. 2359.}

47_{In its decision dated November 21, 2001, case no. COMP/E-1/37.512, OJ 10.1.2003, L 6/1, the Commission}

imposed fines totalling € 855 million to Hoffman-La Roche and seven other companies for conspiring to reduce competition in the vitamins market. Before the Court of First Instance, BASF obtained a reduction in its total fine of € 59.315.000, leaving it with a fine of € 236.845.000 (Case T-15/02, 15 March 2006, not yet published) and Daiichi had its fine reduced from € 23,4 million to € 18 million (Case T-26/02, 15 March 2006, not yet pub-lished).

The American investigation led to a fine of about $ 900 million, and personal fines and prison sentences for a number of executives of the companies involved. Source: http://www.usdoj.gov/atr/public/criminal/212091.htm

48_{Cf. The amici curiae brief submitted by professors First and Fox and in the case of Kruman v. Christie’s,}

(12)

applied the Foreign Trade Antitrust Improvement Act (FTAIA).49 This law exempts export trade and strictly foreign trade from the scope of the American antitrust laws, unless both ex-porter is located in the United States and the damage is suffered in the United States, or unless the violation of the antitrust laws affects trade within the United States or imports to the United States. Since the Empagran case involved foreign plaintiffs suing foreign companies for damage suffered abroad, the conclusion was obvious: the claim was not admissible.

The FTAIA clearly discriminates against foreign residents. At the same time, the Empa-gran case showed that foreign governments consider it a good thing. Several of them submit-ted Amici Curiae briefs to argue for a restrictive reading of the statute.50 The reason why the did this is that they felt that a broad reading of the statute would interfere with their sovereign right to enact laws that govern their jurisdictions. This is a point that we well return to in chapter 4, below.

Before we return to economics, there is one more way that a state can allow its laws to have effect outside its borders. As mentioned above, 28 USC 1782 allows foreign prosecutors and parties to foreign court proceedings to subpoena documents and witness testimony in US courts. Around the same time as the Empagran case, there was an important case before the US Supreme Court that sought to make clear whether this statute applied to proceedings be-fore the European Commission.51

AMD had filed a complaint with the European Commission about the behaviour of its competitor, Intel. In that complaint, it noted the existence of court documents relating to pro-ceedings against Intel that had taken place in the US Federal District Court for the district of Alabama. However, since these documents had been sealed, it would require an action under 28 USC 1782 to obtain them. The European Commission refused to start such an action, so AMD decided to do so itself. Before the Supreme Court, the Commission argued that the sys-tem of investigation in place in the EU was designed in part to avoid the possibility that a company might launch a complaint in order to discover its competitor’s business secrets. That is why AMD’s rights under EU law were severely limited. Allowing it to circumvent these rules by using US law would thus violate international comity.

49_{15 USC 6a}

50_{An amici curiae brief, literally a brief by a friend of the court, is a brief submitted by a person with some}

in-terest in the case who is not a party, offering their point of view for the court’s consideration. The federal gov-ernment of the United States are regularly invited by the Supreme Court to give their views, both when the Court is considering whether it should hear the case at all (or in jargon: whether it should grant the writ of certiorari), and in the merits stage. Cf. 28 USC-Appendix, Rules of the Supreme Court of the United States, Rule 37. For a discussion of foreign governments’ response to unilateral US action of this kind, cf. Griffin (1998).

(13)

The Supreme Court noted that, even if the Commission was not a tribunal, its investiga-tive actions clearly led up to ultimate review by the EC Court of First Instance, which is. As a result, the statute applied to the situation at hand: the District Court could order the production of the documents. In response to the Commission’s concerns, the Supreme Court noted that the statute did not compel the court to issue such an order; if the plaintiff seemed to make the request simply to obtain access to sensitive documents, the request could be refused on that grounds.

As it has now been outlined, the case of international cooperation in antitrust matters has all the elements necessary to make it suitable for the present thesis. There are great and often competing interests at stake, but these are interests that can be described in terms of money, making it more likely that economic theory will be able to shed light on the problem. The case shows some signs of states’ willingness to cooperate, but no substantial progress has been made so far, with the exception of the interesting case of the European Community.

(14)

3. Transaction Cost Economics and Private Firms

Before we can apply Transaction Cost Economics (TCE) to state action, it is first necessary to look at how it can be used to explain the behaviour of non-state actors. In so doing, we will have the chance to discuss the details of the model and the types of questions it is able to an-swer in the empirical environment it was designed for. In what follows, I will first discuss transactions and their characteristics, followed by governance structures and theirs. Subse-quently, I will outline how they are predicted to be aligned. Finally, we will have opportunity to look at the formalisation of the model as it was proposed by Grossman, Hart and Moore. Throughout, it is important to remember that, while the theory introduced here has some no-table differences compared with the received literature, the main purpose of the present chap-ter is to summarise and explain a body of work that has its roots in the 1970s. Only in the next chapter will I analyse what changes are necessary to allow the model to successfully describe state behaviour.

Transactions

In TCE, “the transaction is the basic unit of analysis”.52 This means that the existence of insti-tutions is explained with reference to their effect on the efficiency with which transactions can be conducted. While there are many definitions of institutions53_{, probably the best one comes}

from Schumpeter (1983): “By institutions we mean (…) all the patterns of behaviour into

which individuals must fit under penalty of countering organised resistance, and not only

legal institutions, such as property or the contract, and the agencies for their production or enforcement.”54 In the study of international relations, the term often used to describe such concepts is regime, which is defined as “the principles, norms, and rules governing a

par-ticular issue area of international relations, and to the formal institutional structures

52_{Williamson (1996), p. 6.} 53_{Williamson (1996), p. 4 – 6.}

54_{Schumpeter (1983), p. 191. Throughout this thesis, the concepts of institution and governance structure will be}

(15)

and decision-making procedures through which those principles, norms, and rules are developed.”55 It follows that a regime consists of one or more institutions.

As for transactions, they are defined by Williamson by saying: “a transaction occurs when a good or service is transferred across a technologically separable interface.”56 Unfortunately, in its attempt to emphasise that transactions also occur within firms, this definitions suffers from several shortcomings. First of all, it is not at all clear what is gained by extending the concept of the transaction in this way. Quite the opposite, conceptual clarity would be greatly improved by distinguishing the efficiency of transactions conducted between firms from the efficiency of production which takes place within a firm. As the level of integration changes, so do both levels of efficiency, producing a trade-off where the entrepreneur can profit by finding a better solution than his competitors.57

On the other hand, it is important not to press this too far. The transfer of a good or service between two business units would, in casual language, be considered a transaction, particu-larly if this transfer takes place using a market-based transfer price.58 It should be conceded, though, that the business units of a firm are usually set up as legally separate corporations, so that we could base our definition on the legal reality of the situation, with the caveat that it might be necessary to view a firm that is organised as a single corporation, even though it consists of several separate business units as if it was organised as a number of corporations. Such a definition would have the additional advantage of avoiding the difficult empirical issue of when an interface is “technologically separable”, or even the question of what an interface is in this context. After all, it is probably undisputed that production and marketing are tech-nologically separable, which is why they are often carried out by different business units or at least different departments.59 On the other hand, it is much more a matter of opinion whether the activities of two production workers are technologically separable. Not only is this a sub-ject about which two people could reasonably disagree, it is also a subsub-ject that is quite irrele-vant for the question at hand, at least up to the point when someone suggests putting these activities in two different companies.

Because of these considerations, I will in the remainder of this thesis deviate from Wil-liamson’s definition, and restrict transactions to transfers between legally separate entities.

55_{Krasner (1983)}

56_{Williamson (1996), p. 58.}

57_{Usually, the model is phrased in terms of transaction cost economising. Exceptions are Williamson (1979), p.}

245 and North & Wallis (1994).

58_{Discussions of the determinants of transfer prices include Boyns, Edwards & Emmanuel (1999) and Elitzur &}

Mintz (1996).

(16)

What is left is to consider the difference between the transfer of a good or service in the nar-row sense, the combination of the transfer of a good or service combined with the payment for it and the contract that requires this transfer and its payment.60 Since part of the purpose of the analysis is to look at the relationship between the properties of the contract and the transfer that ultimately takes place, it is clearly important to distinguish between the contract and the transaction. Moreover, since the present thesis is going to be concerned with a number of pos-sible multiparty contracts, and the transaction or transactions resulting therefrom, it may be beneficial to define a transaction as a unilateral act, without the matching consideration. That is why, in this thesis, a transaction is defined as the transfer of a good or service between two legally separate entities.61 In international relations, generally, goods are obviously traded between states, so there is no reason to exclude the possibility of the transfer of a good in this context out of hand. This is especially so since we will have to consider the possibility of payment, or, more accurately, why it is not normally observed between states. Nonetheless, the most important example of a transaction in the context of international antitrust enforce-ment is a state carrying out an act of antitrust enforceenforce-ment that is of value to another.62_We

conclude that a transaction is the transfer of a good, a service, or money, between legally

separate entities.

Having defined a transaction, what is left is to consider is the relevant characteristics of trans-actions, i.e. the characteristics of a transaction that are relevant when considering alternative governance structures. If we, quite arbitrarily, consider the market as the default governance structure, the question is which characteristics of a transaction would make it inefficient to carry out the transaction in a market.63

60_{Alchian & Woodward (1988), p. 66 distinguish themselves from most authors in that they reckognise this}

distinction, although they do not define it quite correctly. Cf. also Hodgson (2002).

61_{What remains is the phrase “good or service”. In legal usage, it is more common to speak of a doing, a not}

doing or a giving (art. 3:296 of the Dutch Civil Code), or, in the words of the Restatement (second) of the Law of Contracts, § 71 (3): “The performance may consist of (a) an act other than a promise, or (b) a forbearance, or (c) the creation, modification, or destruction of a legal relation.” However, a detailed analysis of the subtle differ-ences between different possible expressions is beyond the scope of this thesis. Cf. Ruiter (1987, 2004, 2006)

62_{Enforcement actions that have no effect on any state other than the one carrying them out are outside the scope}

of this thesis. Whether or not such enforcement actions will be carried out depends on the level of rationality of the government and on whether such an action would actually be social welfare enhancing. The question of gov-ernment rationality will be discussed in the following chapter.

63_{Using the market as the default option is reasonable in the sense that it represents the absence of any}

(17)

First of all, we should consider a characteristic of contracts. The conclusion that some-times governance structures are efficiency enhancing compared to a spot market is based on the premise that many contracts are incomplete. A contract is considered complete if it com-pletely describes the transactions that will take place in any future state of the world in such a way that no value enhancing renegotiation is possible ex post. The observation that this is often not the case for real life contracts is connected to the assumptions of bounded rationali-ty64 and nonzero transaction costs. Since transaction costs include the costs of writing a con-tract, the assumption of zero transaction costs necessarily implies that a complete contract can be written costlessly. Also, a perfectly rational actor has, by assumption, all the information that is needed to write such a contract. In real life, on the other hand, contracts usually cover only a small part of the possible states of the world, leaving the more unlikely ones to be solved as they occur.65 Also, because actors are boundedly rational, they are unable to foresee all possible states of the world. When the UN Charter was concluded in 1945, no one could have foreseen the need for humanitarian intervention in the absence of a Security Council resolution, or if they did it would have been considered a proposition that was too vulnerable to abuse.66_{As a result, the only exceptions to the ban on the use of force are self defence and}

force authorised by the Security Council67, which meant that NATO’s 1999 intervention in Kosovo was in violation of international law.

The reason why this is a problem is that the ex post bargaining positions of parties may be substantially different from the ex ante positions. As a result, the party whose bargaining posi-tion is expected to weaken will realise that this weakened bargaining posiposi-tion will lead to a less advantageous distribution of the gains from trade, after renegotiation, since part of the gains from trade will be appropriated by the stronger party.68 This, in turn, will lead it to in-vest less than would be optimal ex ante, leading to a social welfare loss. In other words, if ex

In the context of international relations, it could be argued that the principle of state sovereignty is a reason why no integration should be the default solution. This is something that will be considered in the next chapter.

64_{Simon (1961), p. xxiv, defines bounded rationality as behaviour that is “intendedly rational, but only limitedly}

so”, which of course is not very helpful in and of itself. As a result, there have been many different versions of bounded rationality, even within TCE. A detailed discussion of this issue is outside the scope of this thesis, un-fortunately.

65_{For example, there may be conceivable states of the world that are disastrous for one contracting party. Rather}

than deciding in advance what should be done if such a state of the world is realised, the contracting parties realise that the bankruptcy code provides a fallback option, against which parties can negotiate ex post. In other words, the disadvantaged party can threaten to declare bankruptcy, rather than perform her contractual obliga-tions. Using this possibility as a threat point, parties can negotiate to find an ex post efficient solution.

66_{Note that parties’ inability to describe exactly when humanitarian intervention should be allowed even in the}

absence of authorisation from the Security Council is another way of saying that making the contract complete on this point would be too costly.

67_{Articles 2 (4), 51 and 42, respectively.}

(18)

post renegotiation can be avoided in all states of the world, creating a governance structure is

never optimal. Unfortunately, to do so requires the writing of a complete contract. While the fact that this is impossible is a characteristic of contracts, it is connected to the degree of un-certainty involved in the transaction. As the number of distinguishable future states of the

world increases, as they become more difficult to foresee and as their probabilities of being realised become more uniformly distributed, uncertainty increases and the contract

will necessarily be more incomplete, and thus give rise to greater ex ante inefficiencies.69 In the context of international antitrust enforcement, we see the limitations of our ability to make complete contracts already in the drafting of 28 USC 1782. Rather than describe when the district court can or cannot grant such a request, which would have made the court into “la bouche qui prononce les paroles de la loi”70, Congress decided to simply create the option of granting it, leaving it to the courts to decide when this should be done. In its ruling, the Supreme Court gave some guidance, but much still depends on the circumstances of the case and the trial judge’s assessment.71

Having considered this first characteristic of transactions, we have already encountered the second one. That characteristic is the capacity of the transaction to change the parties’ relative bargaining strenghts. But how can that happen? To answer that question, it is useful to briefly consider the issue of franchise bidding.72

For a long time, economists’ answer to the problem of natural monopolies was that of a choice between three evils: private monopoly, regulation or government ownership.73 But in 1968, Harold Demsetz imaginatively suggested a fourth possibility: Franchise bidding. What he meant was that the social optimum could be reached by creating an auction, auctioning off the right to be the natural monopolist. Whoever agreed to supply the service or sell the good at the lowest price would win the right to do so.74 This is an excellent idea, and there are cer-tainly markets where it can be usefully applied, but unfortunately there are also many cases where this solution is worse than any of the original evils. A classic example of franchise bid-ding going awry is Williamson’s case study of the supply of cable television in Oakland.75

69_{Another way to phrase this is to say that the inefficiencies occur because parties are unable to commit not to}

renegotiate ex post. Hart & Moore (1999), p. 128 – 132. The existence of commitment problems is a well known phenomenon in public choice literature, and we shall return to it later. Cf. Williamson (1984).

70_{Montesquieu, De l’esprit des lois, Livre XI, Chapitre VI.} 71_{Intel v. AMD, op cit, pages 20 – 23 of the ruling.} 72_{Cf. Williamson (1988), p. 76 – 78.}

73_{Friedman (1962), p. 128.} 74_{Demsetz (1968)}

(19)

The lesson we learn from this case study is that franchise bidding may involve competition at the outset, but once the contract is signed, the parties are locked in. This is so because once parties have started making transaction-specific investments, switching to another supplier is costly. As a result, the winning bidder in Oakland managed to substantially renegotiate the contract ex post.

The fact that transaction specific investments have been made, i.e. investments whose

expenditure cannot be wholly recovered should the transaction be cancelled, means that

the incumbent supplier has a competitive advantage compared to all other possible suppliers. As a result, the initially competitive situation is transformed into one of bilateral monopoly. In other words, the first key characteristic of transactions is the extent to which the assets that are required to perform the transaction are specific to it. In Williamson’s words: the degree of asset specificity. This variable can be measured by dividing the opportunity cost of the asset by its purchase price.

Williamson distinguishes between four forms of asset specificity:

• Site Specificity, whereby the location chosen for the investment is specific to the contractual counterparty.

• Physical Asset Specificity, whereby the characteristics of the physical assets pur-chased are tailored to the needs of the transaction.

• Dedicated Assets, whereby termination of the contract would leave the owner of the assets with a substantial overcapacity.

• Human Asset Specificity, whereby human capital is developed that is specific to the transaction.76

In the case of international antitrust enforcement, we can say that states are investing in mak-ing their markets more competitive. The extent to which competitive markets can be viewed as specific assets is something that will be discussed in more detail in chapter 5, below. In international relations, generally, it is clearly not very difficult to think of examples of spe-cific assets, for example the assets that need to be developed in order to participate in NATO, such as military hardware and human capital. These are assets that are to some extent tailored to the need of the organisation and can often represent an overcapacity, meaning that a state might spend more on its army than it would if it were not a member of NATO, so that it can participate more fully in the operations carried out by this organisation.77

76_{Williamson (1983), p. 526. Cf. Joskow (1987), p. 170.}

77_{Clearly this depends on the geopolitical situation of the country in question. The collective security concept}

(20)

The third and final characteristic is the frequency of the transaction. Since the cost of nego-tiating and formulating a contract is to some extent fixed, the total expense of transacting through the market increases with the frequency of the transaction. This means that, when the frequency of the transaction reaches a certain point, it can be more efficient to make arrange-ments that cover more than one transaction, i.e. to establish a governance structure. Such a governance structure could be something as simple as a long term contract whereby the sup-plier supplies a certain predetermined product if and when the customer needs it, for a price that is determined by a fixed formula. In the language of TCE, such an arrangement is consid-ered a hybrid, i.e. a solution somewhere between the extremes of firm and market.78

At this point, it is important to consider how the frequency variable is connected to the other two. It should be clear that the costs of contracting increase as uncertainty increases, and the consequences of a lack of contracting become more severe as asset specificity increases, leading to the result that the break-even frequency where market and governance are equally efficient depends on the other two characteristics of the transaction. In the absence of asset specificity, for example, there is no particular need to write any detailed contracts, because the transaction can just as easily be arranged with competitive spot contracts. That is why the contract between the government and the railroad company is gigantic, while the regulation of the taxi market is astonishingly simple. Since no specific investments are necessary to operate a taxi, the contract between the taxi driver and his customer dispenses entirely with any ex

ante contracting, arranging everything in the spot contract. As a result, contracting becomes

almost costless, and it is hard to imagine someone taking so many taxis that it becomes more efficient to buy the taxi company.79 Similarly, in the absence of uncertainty, contracting be-comes so cheap that it is almost never efficient to create a governance structure. In what fol-lows, the assumption will be that if any one of the three characteristics of a transaction is zero, it is always optimal to organise the transaction through the market.

Governance structures

Similarly to his analysis of the key characteristics of transactions, Williamson has identified three important characteristics of governance structures80: incentive intensity,

78_{Cf. Hodgson (2002) for a critique of the hybrid concept as it is used in much of the TCE literature.}

79_{Although there is still enough friction in the market to make it efficient for some people to have private}

chauf-feurs drive their cars.

80_{As before, and even though casual usage might suggest otherwise, what Williamson means by governance}

(21)

tive controls, dispute settling mechanisms and adaptability.81 Unfortunately, these

vari-ables lack the intuitive logic of the characteristics that were identified for transactions. This is in line with the general criticism of TCE that it has always been more successful at identifying the problems caused by trying to organise a transaction through the market, than at identifying why introducing governance would alleviate these problems.82 After all, by substituting hier-archy for market, the problem of how to persuade your contractual counterparty to behave in a value maximising manner is replaced by the problem of how to persuade your employees to do so. This problem is only partially resolved by the innovation of Grossman, Hart and Moore of defining ownership as the right to determine what happens in the absence of a contractual provision.83

Obviously, as one moves along the continuum from market to firm, the incentives become more low-powered, meaning that the responsiveness of the income of the decision maker to the value created becomes lower. Also, within a firm, there are more administrative controls, allowing for internal dispute settling. Finally, a firm is governed by company law and by em-ployment law, whereas market transactions are governed by private contract law.

What is clear is that high-powered incentives, which are considered a good thing in many economic models, can be harmful to value creation in many situations that are successfully analysed by TCE. This is possible because actors are assumed to behave opportunistically. Williamson defines opportunism as “self-interest seeking with guile”.84 This opportunism is the second behavioural assumption in the model, in addition to the assumption of bounded rationality that was already mentioned before. Without opportunism, parties would simply renegotiate in good faith, making the bilateral exposure caused by asset specificity to be much less of a problem.85

Asymmetric Information

A related source of transaction costs is the presence of asymmetric information; the joint ef-fect of moral hazard and adverse selection. In his work, Williamson does not discuss these at

not confined to agencies and other organisations, but also include all rules, laws and customs, as well as the functioning of the reputation effect between the contracting parties.

81_{Williamson (1996), p. 101 – 105, Williamson (1999), paragraph 2.2 and Ruiter (2005), p. 289.} 82_{Cf. Ruiter (2005)}

83_{Grossman & Hart (1986). The Grossman, Hart & Moore (GHM) approach to transaction cost economics will}

be discussed shortly.

84_{Williamson (1985), p. 30, Williamson (1996), p. 6.}

85_{Williamson (1985), p. 47 – 52 compares and contrasts opportunism with other possible behavioural}

(22)

great length86, but for the present problem, that of the international relations between states, a brief discussion is indispensable.

Put simply, the problem is that in almost any transaction parties have different information about relevant variables. In any simple consumer purchase, the consumer has information that the seller does not, i.e. his subjective valuation of the product. If the seller is a monopolist, even the absence of this information already leads to a welfare loss; it makes first degree price discrimination impossible. In many other cases, the relevant information can be obtained, but only at a cost. In the presence of moral hazard, for example, both parties have an incentive to expend real resources on bonding and monitoring.87 These expenditures are transaction costs, which can be reduced by vertically integrating. After all, the extent to which such information asymmetries are a reason for concern depends on the incentive intensity of the governance structure chosen; a person who is in possession of superior information will take advantage of this information only if and insofar he has an incentive to do so, and this incentive can be re-duced by vertically integrating.88 In this way, we obtain a second set of transaction characte-ristics that needs to be aligned with the governance structure.

Alignment

Now that we know what the relevant characteristics of transactions and governance structures are, what is left is the question of how they are aligned. When Williamson discusses the dif-ference between markets and hierarchies, he constructs his discussion around the concepts of autonomous adaptation, which he derives from Hayek, and which is found in markets, and cooperative adaptation, which he derives from Chester Barnard, and which is found in hierar-chies.89 In this thesis, I will suggest three forms of adaptation, where Williamson’s coopera-tive adaptation is one, while autonomous adaptation turns out to consist of two distinct alter-natives. Which of these three is most relevant for any given case depends in part on the time horizon concerned, with evolutionary adaptation working only over longer time periods, while cooperative adaptation can respond more quickly90, while competitive adaptation works quickest of all.91

86_{Williamson (1985), pp. 51, 82, 293, Williamson (1996), p. 65. Cf. also Barzel (1982).} 87_{Alchian & Demsetz (1972), Barzel (1982).}

88_{The extent to which such incentive remain in an integrated organisation is analysed by agency theory,}

particu-larly Jensen & Meckling (1976).

89_{Williamson (1996), p. 26, 228 – 229. The sources he cites include Hayek (1945) and Barnard (1938).} 90_{Cf. Eisenhardt (1989) for a study of the determinants of the speed of corporate decision making.}

91_{Williamson (2000) distinguishes between four categories of institutions, from resource allocation decisions}

(23)

Firstly, great misalignments are filtered out because they would cause the transaction to collapse altogether. This is the evolutionary model, which has been discussed by Alchian (1950).92 In a large number of situations, this is a very useful explanation of why people who are not rational act as if they are. Alchian gives the example of a city with only one fuel sta-tion. If cars are modelled as driving through this city at random, stopping for fuel whenever they pass by the petrol station, what will be observed is that, after a while, only the “rational” cars, i.e. the cars that stop for petrol every once in a while, are still driving. Or, in other words, all driving cars act rationally, even though they are not.93 An example from recent history is communism. Although many experts felt that organizing all production in a hierar-chy was not such a bad idea94, the eventual demise of the Soviet-Union showed that it was.95 While this approach has been used with considerable success in a number of influential studi-es96, there are some drawbacks. To begin with, there needs to be a clear way in which subop-timal behaviour is punished. While mrs. Thatcher, for example, often argued against harmoni-zation in Europe with the argument that it was preferable to let the differing systems compete, it is in many cases difficult to see how this would happen. If the railways are privatized in one country, and nationalized in another, how would evolutionary forces ever show which is more efficient? Also, such evolutionary mechanisms often work over longer time horizons. Over shorter periods, inefficient systems can be held upright if the political will is there. After all, it took over 70 years before the Soviet-Union collapsed.

Closely connected with evolutionary alignment is alignment through competition, specifi-cally. This approach only works for private companies, in most cases, although I will argue in the next chapter that there is some merit in using it for international antitrust enforcement as well. In a competitive corporate environment, where shareholders force management to maximize profits, best practices will develop, although one would not expect to see the same solution everywhere. Employment laws, for example, are an important factor in taking the make or buy decision. Since these are quite different in different countries, it is expected that different governance structures will be chosen in response. However, the point is that in a competitive environment companies are forced to learn from each other, as well as from their high-paid corporate consultants.

92_{Later contributions include Penrose (1952, 1953), Winter (1964, 1971) and Hirshleifer (1977).} 93_{This is equivalent to the survivor bias in statistics. Cf. Studenmund (2001), p. 544.}

94_{E.g. Schumpeter (1942), p. 165 – 232.} 95_{Cf. Fukuyama (1992)}

(24)

Many of the problems of the evolutionary approach apply to the competitive approach as well. In fact, one could argue that the difference between them is merely a difference of de-gree. However, that would not be an accurate description of the concept. The difference be-tween evolution and competition is a different level of rationality. Evolution operates without any real decision making, rational or otherwise, at all. Competition requires that managers and consultants think about the different solutions, and have some understanding of which options should be preferred to others. The more rational these analyses are, the more quickly the com-petitive environment will force the market participants to tend to the most efficient solution.

Finally, the third approach is that of complete rationality on the part of the decision maker(s). This covers the model of the benevolent dictator, but also the more realistic sce-nario where the government commissions a committee to study what would be the best solu-tion. When the building of a railroad, for example, was first considered in The Netherlands in 1836, the government asked a commission, chaired by Anton Falck, to study what the role of the government in this market should be.97 Although given the present state of knowledge in the field of economics the answer would probably be different today, it is clear that the com-mission’s answer represented an adequate assessment of the problem given the knowledge of the time. Its conclusion was that the state should generally stay out of the running of the mar-ket, limiting itself to helping the market develop by using its powers of eminent domain, by returning the import levies on coal and steel that these companies would otherwise have to pay, etc. It is interesting to see that, after a century of state-run railroads in Europe, many countries are returning to a more market-based governance structure. Truth be told, though, it begs the question whether this is the result of a better understanding of the issues, or rather the result of changes in political fashion.98

Finally, it should be emphasized that Williamson is by no means suggesting a blind adher-ence to such an oversimplified prescription. When using his framework to analyse real life cases99, he instead applies his so-called “remediableness criterion”, which states: “an extant

mode of organization for which no superior feasible alternative can be described and

implemented with expected net gains is presumed to be efficient.”100 As before, the extent to which the proposed alternative has to be feasible to be relevant depends on the horizon over

97_{Veenendaal (2004), p. 27.}

98_{A connection should here be noted between the work of Williamson and that of Hirschman (1970). Rational}

adaptation is more likely if the customers of the organisation express their dissatisfaction through “voice” rather than “exit”.

(25)

which the analysis is undertaken.101 Any proposed alternative is only feasible if it compatible with those institutional factors that are at a higher level of analysis. In other words, when the time horizon is several years, any proposed alternative has to take the existing customs, norms and institutional environment as given.

A more thorough analysis of this point requires a discussion of the concept of lock-in. Lock-in can occur because of the factors described above, or because of switching costs. In fact, those two categories are by no means mutually exclusive. If the term switching costs is taken broadly enough, it can cover all costs associated with changing the higher levels of ana-lysis, reducing the existence of lock-in to all cases where switching costs are observed. The received literature goes well beyond this point, analyzing the circumstances in which switch-ing costs commonly occur. However, a discussion of this literature would be well outside the scope of this thesis. The reader is referred to Woerdman (2002, 2004), and the references therein.

Described schematically, the argument so far looks as follows:

(26)

efficiency

Asset Specificity Information Impactedness

Transaction Cha-racteristics Governance Alter-natives - Asset Specificity - Uncertainty - Frequency - Moral Hazard - Adverse Selecti-on - Incentive Strength - Administrative controls - Dispute settling mechanisms - Adaptability

Grossman, Hart & Moore

This chapter so far contains all the key elements of transaction cost economics. However, before we see what happens when we use it to explain state behaviour, it is useful to examine the formalization of the model the way it was proposed in a series of papers by Grossman, Hart and Moore.102 In their attempt to express the model in mathematical form, these authors have made a number of important contributions, even though they had to make some addi-tional assumptions that were absent from Williamson’s original work. It is these contribu-tions, more than the mathematics, that are of interest for the present thesis, which is why I will not go into the mathematical details of the GHM model.

Firstly, GHM have made a more careful analysis of the nature of incompleteness. In their work, they distinguish between variables that are unobservable and variables that are

102_{Grossman & Hart (1986), Hart & Moore (1988, 1990, 1994, 1999), Hart (1988, 1989, 1995). Other important}

(27)

able. While the former cannot be observed by anyone other than the party that controls it, un-verifiable variables can be observed by both parties to the contract, but not by an outsider. This means that it is impossible to have a court enforce an obligation with regard to a variable that is either unobservable or unenforceable, but that it may be possible to use an unverifiable variable in the contract if an adequate ex post renegotiation game can be designed.103 The classic example of such a variable is effort. In some circumstances, no one but the actor exert-ing the effort knows how much effort he has put in, but in other circumstances the parties to the contract will work together closely enough that they know of one another how much effort each has exerted. In our antitrust enforcement case, the exact level of enforcement is clearly unobservable, because it depends not only on the effort of the authorities, but also on the compliance rate of the state’s companies. As a result, information about the number of cases that have been prosecuted only provides limited information about the effort put in by the antitrust authorities.

While variables that are not observable or verifiable are an important reason for contrac-tual incompleteness, it should be noted that the behavioural assumptions of the general model provide many more reasons why contracts are incomplete. At the same time, introducing the concept of an unverifiable variable into their model allows GHM to analyse the consequences of the absence of courts, which is a question that is of great importance in international rela-tions. The details of this point will be discussed further in the following chapter.

The other main innovation of the GHM model is its unusual approach to ownership. As noted above, students of TCE have always had difficulty explaining how creating a governance structure resolves the problems they see in the market. By replacing the market with a hierar-chy, the contracting problem is not removed. The only thing that has changed is the legal re-gime that controls the relationship between the manager and the person who has to carry out the transaction. GHM resolve this issue by admitting that it exists and that it cannot be re-solved; they assume that ownership does not mean that the owner has complete control over the firm, but rather that he has the ultimate control rights over the assets he owns.104 If the actor who works with the asset reneges on his promise, he can be replaced. Also, if the con-tract provides no rule for what should be done, the fallback position depends on who controls

(28)

the physical asset.105 This comes back in their model because they analyse the contracting problem between two actors who each own an asset. The question then becomes which is preferable: to contract, or for one of the actors to buy the asset of the other. The answer is that it depends on the relationship between the actor and the asset, on the importance of the in-vestment of each actor and on the extent to which this inin-vestment is observable and verifiable. In other words, the actor whose investment is much more important for the ultimate gains from trade should own both assets, especially if his investment is uncontractable. However, if an asset and an actor are closely connected, in the sense that the actor has skills that are essen-tial for the productivity of the asset, he should own it. After all, his unique skills give him veto power over the use of the asset, and in the basic GHM model, joint control over an asset is never optimal.106

Having discussed the transaction cost model the way it can be used to analyse a firm’s make or buy decision107, we can now proceed to an analysis of the particular problems that arise if we try to apply the model to the public sphere.

105_{Cf. Hart (1988), who gives an example based on the famous GM/Fisher Body case study (as discussed in}

Klein (1988), for example): what happens if GM suddenly need more car bodies? If GM owns all the assets, the additional bodies get made unless the managers of Fisher Bodies negotiate otherwise. If both companies are separate, the bodies do not get made, unless the managers of GM negotiate otherwise.

106_{Hart & Moore (1990), propositions 3 and 4, on page 1132.}

107_{For empirical studies confirming the validity of the model, cf. Joskow (1985, 1987, 1988), Ménard (2000),}

Transaction Cost Economics in International Relations: The Case of International Antitrust Enforcement