Micro foundations of public sector management 2 (updated)

Update: Added discussion of the Holmstrom and Milgrom paper.

This is my second post on the Treasury working paper "Contemporary Microeconomic Foundations for the Structure and Management of the Public Sector" by Lewis Evans, Graeme Guthrie and Neil Quigley.

In chapter 3 the authors focus on the "incomplete contracts" or "property rights" approach to theory of the firm.

In section 3.2 Evans, Guthrie and Quigley write,

"Transaction-cost explanations for contractual incompleteness are unsatisfactory, because there is more incompleteness than can be accounted for by transaction costs (specifically, because there are many elements where contracting is not possible rather than just more costly than the alternative). Examples include situations where information is symmetric, but key contractible elements are not verifiable by either party. Even when transaction costs are zero, incomplete contracts may arise because parties cannot observe relevant economic variables, cannot verify those variables to a legal standard of proof, or prefer not to disclose information about themselves that would be required for a complete contract."

Now this I don't get. Aren't all cases of incompleteness driven by some form of transaction cost? If we think of transactions costs as the costs of market transactions then in a zero transaction costs world contract would be complete since such contracts would be cost-less to write. Coase has written,

"The solution to the puzzles that I took with me to America [Why are there firms?] was, as it turned out, very simple. All that was needed was to recognize that there were costs of carrying out market transactions and to incorporate them into the analysis, something which economists had failed to do. A firm had therefore a role to play in the economic system if it were possible for transactions to be organized within the firm at less cost than would be incurred if the same transactions were carried out through the market. The limit to the size of the firm would be set when the scope of its operations had expanded to the point at which the costs of organizing additional transactions within the firm exceeded the costs of carrying out the same transactions through the market or in another firm."

The idea that when

"information is symmetric, but key contractible elements are not verifiable by either party"

we get incomplete contracts seems odd. If information is known to the contracting parties, what does it not being verifiable to the parties mean? If the information is not verifiable to a third party, e.g. a court, then a contract can be incomplete, but this is a different thing from information being non-verifiable to the contracting parties. If information is symmetric in that it is unknown to all contracting parties, will a contract be incomplete? The answer to this depends on whether or not the information is verifiable to a third party. If it is then the contracting parties can contract on it by just getting the third party to verify the information. If the information is not verifiable to the third party then a contract will be incomplete. But the reason for the incompleteness is not because the information is unknown to the contracting parties but rather because it is non-verifiable to the third party.

The idea that,

"Even when transaction costs are zero, incomplete contracts may arise because parties cannot observe relevant economic variables, cannot verify those variables to a legal standard of proof, or prefer not to disclose information about themselves that would be required for a complete contract"

also seems odd. I'm not sure what they mean when they say that information is non-observable to the contracting parties. If this means a moral hazard/adverse selection type framework then contract are comprehensive and not incomplete. As Hart explains it,

"Although the optimal contract in a standard principal-agent model will not be first-best (since it cannot be conditioned directly on variables like effort that are observed by only one party), it will be 'comprehensive' in the sense that it will specify all parties' obligations in all future states of the world, to the fullest extent possible. As a result, there will never be a need for the parties to revise or renegotiate the contract as the future unfolds. The reason is that, if the parties ever changed or added a contract clause, this change or addition could have been anticipated and built into the original contract."

and

"One would also not expect to see any legal disputes in a comprehensive contracting world. The reason is that, since a comprehensive contract specifies everybody's obligations in every eventuality, the courts should simply enforce the contract as it stands in the event of a dispute."

Clearly such a contract is not incomplete. If Evans, Guthrie and Quigley mean that neither of the contracting parties can observe the variable then we are in case discussed above in; which the important point is the verifiability of the variable to a third party. If the contracting parties,

"cannot verify those variables to a legal standard of proof"

then contracts could be incomplete. Non verifiability of information to a third party such as a court is the standard argument as to why contracts are complete. But this argument can be countered by the Maskin and Tirole critique. Maskin and Tirole argue that information which is observable to the contracting parties can be made verifiable (to a third party) by the use of ingenious revelation mechanisms. The contracting parties write into their contract a game which when played gives the appropriate incentives for them to truthfully reveal their private information in equilibrium. This undermines the non-verifiability approach to incomplete contracts.

If some of the contracting parties,

"prefer not to disclose information about themselves that would be required for a complete contract"

then its hard to see that we are in a zero transaction costs world. Isn't not providing information the same as saying the costs of contracting on that information are infinite? This looks like a very large transaction cost!

Under section 3.3 Evans, Guthrie and Quigley write,

"The starting point for this approach to the theory of the firm is the incompleteness of contracts. Since humans are boundedly rational, not all issues of relevance to a contract can be anticipated at the time of writing the contract."

But Oliver Hart argues,

"In the last few years, a literature has developed on the theory of incomplete contracts, and on applications of this theory to the understanding of organizations, such as firms. In this paper, I will argue that, while transaction costs of various sorts are a crucial ingredient of this literature, bounded rationality in the sense that agents have limited cognitive, computational or comprehension skills is not."

Given that Evans, Guthrie and Quigley argue that incompleteness is important for contracts there is one question that they need to answer. As it is possible for the contracting parties to fill any gaps in their contract as they go along, Why is contractual incompleteness important? The reason is that renegotiation itself imposes costs. These cost can be both ex post, incurred at the time of renegotiation, or ex ante, incurred in anticipation of renegotiation.

In section 3.4 the point is made that,

"The incomplete contracting perspective embodied in this example represents a sharp break with the earlier transaction cost-based literature on the firm. Whereas incomplete contracts imply that inefficiencies arise because it was hard to foresee and contract about the uncertain future, earlier literature tended to take a “complete contracts” perspective in which imperfections arise as a result of moral hazard and asymmetric information."

I would read the "earlier literature" comment to refer to the transaction cost literature. But incomplete contracts are a central feature of the transaction cost approach. As Hart and Moore (2007) explain

"Transaction cost economics (see, e.g., Oliver Williamson (1975, 1985), Benjamin Klein et al. (1978)) argues that firms are important when contracts are incomplete, and parties make large relationshipspecific investments."

Section 3.5 of the paper looks at the link between transaction costs, incentive-based theories and incomplete contracts. When discussing incentive-based theories of the firm Evans, Guthrie and Quigley write,

Incentive-based theories of the firm have their foundation in the analysis of the incentive problem between a principal and an agent. This approach assumes that there are many tasks and many instruments associated with the agency problems in a firm, and asset ownership is merely one of the instruments. Papers in this paradigm consider two ways to structure the agency problem: (i) where the agent does not own the asset (is an employee) and therefore has incentives provided by being paid on measured performance, and (ii) where the agent does own the asset (is an independent contractor) and receives both a payment based on measured performance and the value of the asset after production occurs.

This approach to the theory of the firm has in effect focused on the claimed distinction between the low-powered incentives associated with employment, and the high-powered incentives associated with contracting. Employees require low-powered incentives because they are not distracted by the contractor’s incentives to increase the value of the assets used for production. More generally, joint optimisation over asset ownership and contract parameters determines whether to conduct activity within the firm or outside.

The incentive-system theory of the firm is therefore related to the incomplete contracts literature, both in its use of ownership as an instrument and in its ability to provide a unified account of the costs and benefits of integration.

Incentive theory is normally thought of as a comprehensive contracts based theory and much of the literature is of this form. Think of moral hazard models. Incentive theory of this type is probably best understood as a extension of the neoclassical theory of the firm that inquiries into the incentive conflicts that may hinder the firm from reaching its production possibility frontier. But not all incentive theory is of this kind. While its not exactly clear what set of papers is being referred to above. I assume that papers like Bengt Holmstrom and Paul Milgrom’s 1994 paper, "The Firm as an Incentive System" fall into this group.

Holmstrom and Milgrom here stress the importance of viewing the firm as "a system", specifically as a coherent set of complementary contractual arrangements which mitigate incentive conflicts. In their opinion, it is misleading to focus on any one single aspect of the coherent whole: the firm is characterized by the employee not owning the assets, by the employee being subject to a low-powered incentive scheme, and by the employee being subject to the authority of the employer. These “incentive instruments” are complementary: For example, in the presence of measurement costs, it is important that a person who does not own the assets which he uses is not subject to high-powered incentives, since he then is likely to care too little for the assets. Likewise, low-powered incentives make it important for the employer to be able to exercise authority over the use of the employee’s time, since the employee will lack the proper incentive to be productive. Due to this complementarity it is logical that independent contracting has the exact opposite constellation of instruments from the employment relationship.

The choice between the two different incentive systems depends importantly on the extent to which every dimension of a person’s contribution can be measured. When an important dimension is unmeasurable, it might be counterproductive to remunerate the person through a high-powered incentive scheme since the person is likely to allocate too little attention on the unmeasurable activity. Thus, according to Milgrom and Holmstrom lack of measurability is an important variable determining the size of the firm [...]. (Foss 2000)

An important point to note about this paper is that it is not only a principal-agent theory but also an incomplete-contracting theory. So the relationship between the two theories can be a very close one, the theories are not just related but can be usefully merged.

In the past I have argued that transaction cost and property rights theories are "orthogonal" to each other. In a discussion of the differences between the Grossman-Hart-Moore (GHM) theory of the firm and the transaction-cost approach, Williamson (2000, pp. 605–606) argues that the most important difference between them is that GHM introduce inefficiencies at the ex ante investment stage while the transaction-cost approach emphasises that ex post haggling and maladaptation drive inefficiencies. There are no ex post inefficiencies in GHM due to their assumption of common knowledge and ex post costless bargaining. Gibbons (2010, p. 283) explains it this way:

‘[t]he model in question is Grossman and Hart’s (1986), which explores an alternative to Williamson’s (2000, p. 605) emphasis that “maladaptation in the contract execution interval is the principal source of inefficiency.” Instead, in the Grossman-Hart model, there is zero maladaptation in the contract execution interval, and the sole inefficiency is in endogenous specific investments.

It is striking how different the logic of inefficient investment can be from the logic of inefficient haggling. In their pure forms envisioned here, the two can be seen as complements. For example, the lock-in necessary for Williamson’s focus on inefficient haggling could result from contractible specific investments chosen at efficient levels. But by assuming efficient bargaining and hence zero maladaptation in the contract execution interval, Grossman and Hart focused attention on non-contractible specific investments and hence discovered an important new determinant of the make-or-buy decision: in the Grossman-Hart model, an important benefit of non-integration is that both parties have incentives to invest; in Williamson’s argument, an important cost of non-integration is inefficient haggling. In short, the two theories are simply different’.

This emphasis on ex post haggling and maladaptation can be interpreted as reflecting a view thatinternal organisation is better at reconciling the conflicting interest of the parties to a transaction and facilitating adaptation to changing supply and demand conditions when such cost are high.

One point worth making is that the reference point approach (not much discussed in the Evans, Guthrie and Quigley paper) to the firm that has grown in very recent times out of the property rights approach can be seen as a move away from the ex ante GHM approach and back towards transaction cost thinking in so much as contracting is not perfectly contractible ex post.

Chapter 4 of the Evans, Guthrie and Quigley paper is on Real Options and Investment.