Navigation – Plan du site

Solving the « Selective Intervention Puzzle »: Some Reflections

Pierre Garrouste
p. 119-122

Notes de l'auteur


Texte intégral

1In his 1937 paper Coase addresses two problems. The first deals with the nature of the firm and is solved by assuming the existence of transaction costs. Without such costs the market is always the best coordination device. The second is: « why is not all production carried on in one big firm? » (Coase, 1937, p. 340)? This version of the selective intervention puzzle (Williamson, 1995, p. 231) is solved by Coase by comparing, as rightly written by Crémer, the internal and external marginal transaction costs. When they are equal, the firm stops growing. The way Williamson explains the puzzle is slightly different. He stresses that « the integrity of rule governance is unavoidably compromised by discretion (Williamson, 1985, chapter 6). Accordingly any effort to combine rule governance (as in markets) with discretionary governance (hierarchy) experiences trade-offs. » (Williamson, 1995, p. 231).

2I agree with Crémer (2010) on the fact that it would have been more satisfactory to explain the internal (bureaucratic) and external (market) transaction costs on the same basis. Such an explanation is however absent from the theories of the firm. Wording the problem differently, Holmstrom stresses that « I take the ownership of assets by firms, and the attendant feature that economics contracts are made with firms, not with their employees or owners, as one of the most significant and robust empirical regularities to be explained by any theory of the firm. » (1999, p.75) and adds « having said this let me confess at the start that I am unable myself to offer a well-developed explanation of asset ownership by firms. » (ibid.).

3The problem of the boundaries of the firm is indeed still open. The introduction of asymmetries of information does not solve this problem even if we usually consider that adverse selection is a case of pre-contractual opportunism and moral hazard a case of post-contractual one. As for the Grossman, Hart and Moore’s theory, if it explains why somebody owns an asset, it does not explain why a firm does own assets and how many.

  • 2 The limits of the firm should indeed be defined by the factors that explain the existence of (exter (...)

4My first point is that, even if Crémer’s critic is completely relevant, it seems to me that he argues as if Williamson was sharing the traditional incentive theory assumptions. It is however not the case. And this has very important consequences not only in terms of the explanation of integration or nonintegration but also in the very definition of the notion of contract. According to Grossman, Hart and Moore (henceforth GHM), the notion of contract is a basic concept and is a tie, i.e. a set of irrevocable commitments. Because individuals are rational and some commitments not contractible (they are linked with observable but non verifiable variables), contracting agents have no interest to respect them and nobody can constraint them to do so. The allocation of property rights determines the bargaining positions of the parties ex post (that is the way they share the surplus), and indirectly determines the decision to respect those commitments. The point is then to define the optimal allocation of property rights. In the GHM world, the relevant, i.e. the chosen type of contract (integration or not) depends on the allocation of property rights because ‘the only way in which the agents can be given incentives is through the allocation of property rights’ (Crémer, p. 7). Let’s compare this way of thinking with Williamson’s one (and as we will see with the tenants of the transaction costs economics, henceforth, TCE). Suppose few not perfectly rational and opportunistic individuals, transacting frequently in an uncertain world and both using highly specific assets. The possibility for some of them to hold-up the others if they sign a long term contract makes integration the only viable possibility. It follows that the characteristics of a transaction, that is its frequency, the level of uncertainty and the specificity of the assets that support it, are the determinant of the kind of contract (integration, hybrid forms or market) that is selected (what the TCE calls the governance structure). In fact the type of contract „chosen’ depends on the characteristics of the transaction. If the assets are not specific at all the possibility for opportunistic individuals to hold-up the rent is ineffective and it is therefore not necessary to integrate the associated transaction. To sum up the TCE proposes a solution to the make or buy problem, which is a binary answer but provides no explanation of the limits of the firm.2

  • 3 It follows logically as in Hart and Moore (2005) that an optimal command chain is characterized by (...)
  • 4 Here Coase considers a labor contract.

5Accordingly, one of the main differences between the ways Grossman and Hart (1986) or Hart and Moore (1990) and Williamson (1985, 1995) solve the selective intervention puzzle is due to the different links they stress between incentives, property rights and residual rights of control. According to GHM residual rights of control are given by property rights that enhance incentives because decision rights are not separable from property rights3, even if decision rights can be delegated. As for Williamson, what determines the choice between make or buy is linked with the possibility for one party, depending on the characteristics of the transaction to act opportunistically and consequently hold-up the other party. The relation between property rights and incentives is not direct and the reason why integration takes place is not due to a problem of incentives. This is due basically to the fact that in the TCE property rights and decision rights are not linked as in the GHM approach. On this topic, Riordan’s critic of the GHM approach which Crémer refers to is perfectly clear. According to Riordan, it is not a problem of allocation of property rights that explains vertical integration but a problem of decision rights. This is in line with Coase when he writes that « the contract4 is one whereby the factor, for a certain remuneration (that can be fixed or fluctuating), agrees to obey the directions of an entrepreneur within certain limits. The essence of the contract is that it should only state the limits of the powers of the entrepreneur. » (1937, p. 391). The distinction Gibbons (2005) makes in Williamson’s theory between rent-seeking and adaption approach highlights the way the selective intervention puzzle is solved. According to the rent-seeking perspective « integration can stop socially haggling over appropriable quasi-rents » (Gibbons, 2005, p. 204) whereas for the adaptation conception, the problem is to ask « whether integration or non-integration better facilitates adaptive, sequential decision-making » (ibid, p. 208). According to Gibbons, the problem is that Williamson, in order to solve the selective intervention puzzle introduces the idea of costs of bureaucracy to balance the (external) costs of transaction and consequently does not explain the size of the firm with the factors (asset specificity, frequency of transaction, uncertainty, bounded rationality, and opportunism) that are at the origin of transaction costs.

6To sum up this first point, Crémer’s critic of Williamson’s selective intervention puzzle solution is perfectly relevant and points a problem of consistency: why internal transaction costs are not explained in the same ways that external ones are? This critic does not however take into account that Williamson’s conception is based on a very specific conception of a contract.

  • 5  « In the contractual literature, in contrast, we have seen decision rights moved across fixed firm (...)
  • 6  « We allow for a variety of observed governance structures for coordinating activities between fir (...)
  • 7 A relational contract is a non written, unformal agreement.
  • 8 This result is sensitive to the discount rate (as in Axelrod, 1984).

7The second point that I want to stress is the fact that the relationships between decision rights, property rights and incentives are perfectly disentangled by Baker, Gibbons and Murphy (2004, 2008). They begin by opposing two traditions: the first which assimilates decision rights and property rights (the GHM approach), the second, the traditional contractualist approach which considers that decision rights are not directly given by property rights5. This distinction is thereafter used to try to unify those two approaches. They « define a “governance structure” as an allocation of decision rights and payoffs to the parties, regardless of whether this allocation is achieved through contracts or asset ownership. »(Baker, Gibbons, Murphy, 2008, p.148). This definition allows them to present a richer number of organizational devices6 and to analyze the relationships between firms that they define as « conscious islands ». Accordingly even if they follow GHM when they define what integration and non integration are (2002), they improve this approach because they stress the idea that just because some commitments are not contractible the downstream party has to « provide incentives to the upstream party only if it is self-enforcing » for the relational contract7 to persist. (Baker, Gibbons, Murphy, p.41). If not, the upstream party has a high temptation to renege. In other words the short term payoffs have to be lower than the long term ones. A corollary of their model is that the firm cannot mimic the market just because the parties’ temptations to renege are then too high. In a dynamic perspective a tit for tat game permits to explain that as long as the parties’ temptations to renege (the one shot payoffs) are lower than the interest to pursue the relational contract (the cumulated payoffs) the parties have both interest to continue to agree on this contract8.

8This theory seems an elegant solution both to the selective intervention puzzle à la Williamson and to its Holmstrom’s expression (see above).

9Crémer points a real problem and shows that the contractualist theories of the firm have a challenge to take up. Fortunately some recent improvements of the GHM approach, as the relational contract theory quoted above accept it.

Haut de page


Axelrod, R. (1984), The Evolution of Cooperation, Basic Books.

Baker, G., Gibbons, R., Murphy, K.J. (2002) « Relational Contracts and the Theory of the Firm », Quarterly Journal of Economics, 39-84.

Baker, G., Gibbons, R., Murphy, K.J. (2004) Strategic alliances: Bridges between « islands of conscious power », Mimeo, Harvard MIT and USC.

Baker, G., Gibbons, R., Murphy, K.J. (2008) « Strategic alliances: Bridges between “islands of conscious” power » Journal of Japenese and Internaltional Economies, 22: 146-163.

Coase, R. (1937) « The nature of the Firm », Economica 4, 386–405.

Crémer, J. (2010) « Solving the “selective intervention” puzzle », Revue dEconomie Industrielle, no 129-130.

Gibbons, R. (2005) « Four Formal(izable) Theories of the Firm » Journal of Economic Behavior and Organization, 58, 202-47.

Grossman, S. J., Hart, O. D. (1986). The costs and benefits of ownership: a theory of vertical and lateral ownership. Journal of Political Economy 94, 691-719.

Hart, O., Moore, J. (1990) « Property rights and the nature of the firm », Journal of Political Economy, 98, 1119–1158.

Hart, O., Moore, J. (2005) « On the Design of Hierarchies: Coordination Versus Specialization », Journal of Political Economy, 113, 675-702.

Holmstrom, B. (1999) « The firm as a subeconomy », Journal of Law, Economics and Organization, 15, 74–102.

Williamson, O. (1985) The Economic Institutions of Capitalism, Free Press, New York.

Williamson, O; (1995), Organization theory, from Chester Barnard to present and beyond, Oxford University Press.

Haut de page


1  I thank Anne Yvrande-Billon for her useful remarks and comments.

2 The limits of the firm should indeed be defined by the factors that explain the existence of (external) transaction costs.

3 It follows logically as in Hart and Moore (2005) that an optimal command chain is characterized by the fact that the probability of an individual having an idea on the use of a set of assets is negatively related with his/her right to decide how to use this set.

4 Here Coase considers a labor contract.

5  « In the contractual literature, in contrast, we have seen decision rights moved across fixed firm boundaries (i.e., without changing asset ownership). These decision rights can be extracted (or « alienated ») from their native assets ». (Baker, Gibbons, Murhy, 2004, p. 4)

6  « We allow for a variety of observed governance structures for coordinating activities between firms, including acquisitions, unstructured collaborations, divestitures, licensing agreements, and royalty contracts. » (Baker, Gibbons, Murphy, 2008, p. 148)

7 A relational contract is a non written, unformal agreement.

8 This result is sensitive to the discount rate (as in Axelrod, 1984).

Haut de page

Pour citer cet article

Référence papier

Pierre Garrouste, « Solving the « Selective Intervention Puzzle »: Some Reflections », Revue d'économie industrielle, 132 | 2010, 119-122.

Référence électronique

Pierre Garrouste, « Solving the « Selective Intervention Puzzle »: Some Reflections », Revue d'économie industrielle [En ligne], 132 | 4e trimestre 2010, document 5, mis en ligne le 15 décembre 2012, consulté le 29 mars 2017. URL : ; DOI : 10.4000/rei.4424

Haut de page


Pierre Garrouste

Université de Nice-Sophia Antipolis et GREDEG

Articles du même auteur

Haut de page

Droits d'auteur

© Revue d’économie industrielle

Haut de page