|
Article Excerpt
Economic models of contract typically assume that courts enforce obligations based on verifiable events (corresponding to the legal rule of specific performance). As a matter of law, this is not the case. This leaves open the question of optimal contract design given the available remedies used by the courts. This article shows that American standard form construction contracts can be viewed as an efficient mechanism for implementing building projects given existing legal rules. It is shown that a central feature of these contracts is the inclusion of governance covenants that shape the scope of authority and regulate the ex post bargaining power of parties. Our model also implies that the legal remedies of mistake, impossibility and the doctrine limiting damages for unforeseen events developed in the case of Hadley v. Baxendale are efficient solutions to the problem of implementing complex exchange.
And other things of this sort should be known to architects, so that, before they begin upon buildings, they may be careful not to leave disputed points for the householders to settle after the works are finished, and so that in drawing up contracts the interests of both employer and contractor may be wisely safe-guarded. For if a contract is skillfully drawn, each may obtain a release from the other without disadvantage.
Vitruvius, "Ten Books on Architecture," Chapter 1, Book 1, circa 1 st Century BC
1. Introduction
* There are several ways in which actual contracts vary from those portrayed in economic models. First, economic models of contract typically assume that contracts are enforced as written (a legal rule known as specific performance). In contrast, in the United States, the common law rule for breach of contract is that the court award damages for harm arising from the breach (the common law rule of specific expectation damages). (1) Second, economic models of contract are comprehensive in that a contract is modelled as a function that specifies outcomes/actions for every relevant state of the world. However, legally enforceable contracts are typically modular--they involve a great deal of "boilerplate" and only a few terms are modified to deal with special circumstances in the relationship at hand. (2) These differences between theory and practice imply that results obtained from the theoretical economics literature on incomplete contracts may not be directly applicable to many observed contracts.
It is generally recognized that contracts for the exchange of complex goods are necessarily incomplete. Building upon the insights of Williamson (1975), the early literature on contract incompleteness has shown that the need to renegotiate a contract in the face of an unforeseen contingency may lead to inefficient investment into relationship specific investments. (3) However, it is difficult to tell whether a particular observed contract is efficient or not. There is also a general presumption that failure to enforce contracts as written (i.e., lack of specific performance) is likely to lead to inefficient outcomes. (4)
In this article, we follow Goldberg and Erickson's (1987) suggestion that industry practice is a useful starting point for understanding efficient contract design. We analyze the standard form construction contracts developed by the American Institute of Architects (AIA). These forms are used to allocate a large fraction of the resources devoted to construction in the United States, currently about 9% of U.S. GDP. (5)
Our main result is that these contracts are an efficient mechanism for implementing complex building projects in the "shadow" of American law. (6) A key feature of these contracts is the inclusion of governance covenants that carefully allocate authority between the two parties and limit the ex post bargaining power of the parties. For example, the AIA has a set of forms that are used for implementing an auction for the selection of a contractor. When a contractor wins, he knows that he is the low bidder, and hence has an incentive to try to renegotiate an increase in the contract price. The AIA forms address this problem by requiring bidders to post a bond that makes such renegotiation difficult. Moreover, in addition to ensuring that the low-cost seller is awarded the contract, the auction also ensures that the buyer is the residual claimant to any future gains from trade. This in turn ensures that the buyer makes efficient investments into planning and design. AIA forms have many additional contractual instruments that allocate authority to either the buyer or the seller as a function of the events that occur during the construction project. Together, the various contractual instruments provided by the AIA forms use the law, as enforced in American courts, to ensure the construction of complex projects at the lowest cost.
Our model also provides a rule for contract damages that is a function of the extent to which an event is "foreseeable" or not. An implication of this rule is that it is optimal to excuse performance in situations when an event is not foreseen. This result implies that the current excuses from performance under American law are efficient damage rules. This includes the legal rule of "mistake," which can occur when one party did not understand an obligation, "impossibility," when an unforeseen event occurs that makes performance impossible, and the doctrine limiting damages for unforeseen events developed in the case of Hadley v. Baxendale.
The agenda of the article is as follows. In the next section, we briefly discuss the relationship of this work to the literature. Section 3 outlines the procurement model we study and characterizes the optimal allocation. Section 4 discusses several contractual instruments that are found in the AIA form contracts, and shows how together they implement the efficient allocation. In Section 5, we show that our model provides a rule for contract damages that can explain several of the excuses from performance of a contract as efficient default rules. The final section of the article contains a concluding discussion.
2. Relationship to the literature
* The early literature on incomplete contracts, including Rogerson (1984), Grout (1984), Hart and Moore (1988), and Tirole (1986), supposes that in states where a contract is incomplete or inefficient, parties play a renegotiation game that leads to ex post efficient allocations at the cost of providing inadequate incentives to make relationship-specific investments. (7) In addition to providing a coherent model of contract incompleteness, this literature makes precise Williamson's (1975) insight that contract incompleteness can lead to inefficient allocations. Bajari and Tadelis (2001) introduce a clever way to endogenize contract incompleteness based upon the idea that planning for the future is a relationship specific investment that affects the probability that a contract has a clear performance obligation for any realized state of the world. Taking the renegotiation game as given, and assuming the legal rule of specific performance, they show that the choice between a fixed price and cost plus contract depends upon the trade-off between the ex ante cost of planning and the ex post benefit in reduced costs of renegotiation. Tirole (2009) uses this idea to build an elegant and comprehensive model of incomplete contracts that provides predictions regarding contract duration and the extent to which relational contracts are optimally incomplete.
Another line of the literature explores the conditions under which one has efficient contracting. Chung (1991) and Aghion, Dewatripont, and Rey (1994), building upon the general results of Moore and Repullo (1988), show that if one has the rule of specific performance available, then one can achieve the first-best with a contract that regulates the ex post bargaining power of parties through the appropriate manipulation of the default terms. Maskin and Tirole (1999) show that renegotiation design can ensure efficient even though goods are complex in that they cannot be described ex ante. Recently, Evans (2008) has shown that one can ensure efficient investment and trade in a model with simple contracts and an infinite horizon bargaining model.
There is also a literature on legal defaults that asks the question--how should the law set damages when contract terms are missing or incomplete? The early work by Shavell (1984) and Rogerson (1984) considers the three legal rules that the courts might use in setting damages: expectation damages (the pecuniary value the harmed party expected from performance), reliance damages (the costs borne by the harmed party from entering into an agreement), and specific performance (an order to the defendant to perform as promised). They find that in general none of these rules achieve the first-best, although specific performance is generally preferred to the other rules. Edlin and Reichelstein (1996) show that specific performance can ensure efficient trade, but expectation damages cannot achieve an efficiency allocation in their model.
The paper most related to our work is MacLeod and Malcomson (1993), who provide an early result on efficient contracting in the shadow of the law. The goal of that paper is to explore how the holdup model can explain many features of observed contracts, such as wage and price rigidity. The model provides an explanation for Joskow's (1988) puzzling observations regarding long-term coal contracts. Joskow finds that these contracts often have complex indexing provisions that ensure that contract price tracks the market price for coal. The puzzle, then, is why have a contract at all rather than simply use the market price? MacLeod and Malcomson (1993) show that the combination of the U.S. law limiting the use of specific performance and the need to provide incentives to both parties to make relationship specific investments yields a contract with exactly this form that implements the first-best. (8) Thus, even though the law constrains the set of feasible contracts, parties are still able to find contractual instruments that implement the efficient allocation.
This article extends these results to the case of complex exchange exemplified by large construction projects. In this case, simple indexed contracts are not sufficient to implement an efficient allocation. Rather, efficiency is achieved with a complex combination of contractual instruments. Our results also illustrate that in contract design one must be cautious before concluding that expectation damages necessarily lead to an inefficient allocation, as the literature has thus far concluded.
3. A model of procurement
* Consider a risk-neutral buyer who wishes to contract with one of several potential risk-neutral sellers for the supply of a project that entails significant relationship specific investments by the selected seller. (9) The key ingredients of the procurement process are as follows:
(i) The preference ordering of the buyer over project characteristics is private information. Hence, the buyer must be induced to voluntarily reveal her most preferred project given the cost.
(ii) Investment into planning by the buyer is assumed to be observable by the potential sellers, but not contractible. Bajari and Tadelis (2001) observe that project design provides a "concrete" example of a relationship specific investment that is observable by both parties ex post, but cannot be explicitly contracted upon (Grossman and Hart 1986). It is well known in the construction industry that contractors use information on the quality of project design when setting their bids. (10)
(iii) Following Laffont and Tirole (1986), it is assumed that the ex post cost of production is observed, but not the ex ante investment by the seller into cost reduction.
(iv) The project is complex, in that it is built up from a set of components, such as the foundations of a building, the window frames, the roof, electrical system, and so forth. This complexity implies that the design is incomplete in two dimensions. First, it may be necessary to change the specifications of a component ex post. Second, the buyer may wish to add components or elements to the project that were not anticipated at the time the contract was signed.
Providing a precise definition of complexity is difficult, and certainly controversial. Here we follow the literature and use the notion of complexity in two senses. The first notion is due to Bajari and Tadelis (2001). Their insight is to recognize that investment in design affects the probability that the buyer will desire a change to the specifications of a project component. For example, one might realize that a paint color does not look quite right once applied, and after construction the buyer requests a color change. This change might have been avoided if the buyer had spent more time building prototypes of the project. The key feature of the Bajari and Tadelis (2001) model is that the investments into design are observed by the seller, and hence the seller can anticipate the likelihood of a design change ex post.
A project may also be complex because the buyer may require the addition of components to the project that were unforeseen at the time of the design. In the case of the Getty Museum in Los Angeles, the Northridge earthquake occurred during construction. From this event, the builders learned that they had to make substantial changes to the structure. Given that earthquakes are common to Los Angeles, the possibility of an earthquake was not unforeseen. The real issue is that it is costly to learn the detailed consequences of such an event. In this case, what was unforeseen was the incompleteness of their knowledge regarding the effect of an earthquake upon the existing structure. Accordingly, we explicitly model unforeseen events as a form of learning one's true preferences over project specifications.
[FIGURE 1 OMITTED]
Given that both the buyer and the seller have made relationship specific investments, it is cheaper to have the current seller carry out any unforeseen modifications. However, due to the asymmetric information that may exist between the buyer and seller, this may lead to inefficient ex post renegotiation. In addition, given the design, the seller can make noncontractible relationship specific investments that reduce production costs. The goal of the contract between the two parties is to ensure efficient investment into both the design and execution of the project.
The next subsection provides a timeline for the model, followed by a detailed description of each step.
* The timeline for the procurement process. The procurement process is divided into three major stages: ex ante, interim, and ex post. The ex ante stage encompasses the initial planning of the project and the selection of a suitable seller. The interim stage consists of a sequence of actions by the seller to carry out the construction of the project, and the ex post stage entails the final settling up of payments, including possible litigation. The next subsection provides a characterization of the optimal allocation subject to the informational constraints of the environment. The timeline for the procurement process is as follows (see Figure 1):...
|
