The private and social functions of contracts and of contract law are examined here. In section 4.1 the basic theory of contracts is considered, in section 4.2 production contracts (which have been the focus of a substantial literature) are analyzed, and in section 4.3 several other types of contract are discussed.
Definitions and framework of analysis
A contract is a specification of the actions that named parties are supposed to take at various times, as a function of the conditions that then obtain. The actions usually comprise delivery of goods, performance of services, and payments of money, and the conditions include uncertain contingencies, past actions of parties, and messages sent by them. A contract is said to be complete if the list of conditions on which the actions are based is exhaustive, that is, if the contract provides explicitly for all possible conditions. Otherwise, a contract will be referred to as incomplete. Typically, incomplete contracts do not include conditions which, were they easy to include, would allow both parties to be made better off in an expected sense. It should be noted that an incomplete contract may well not have literal gaps in
that it covers all conditions at least by implication. Consider, for example, a contract stating merely that a specified price will be paid for a named quantity of wheat. Although this contract is incomplete because it does not mention many contingencies that might affect the buyer or the seller of wheat, it has no gaps, as it stipulates what the parties are to do (pay a price, deliver wheat) in all circumstances. A contract is Pareto efficient if it is impossible to modify in a manner that raises the expected utility of both of the parties; such a contract will sometimes be referred to simply as efficient or as mutually beneficial.
Contracts are assumed to be enforced by a tribunal, which will usually be interpreted to be a state-authorized court, but it could also be another entity, such as an arbitrator or the decision making body of a trade association or a religious group. (Reputation and other non-legal factors may also serve to enforce contracts but will not be examined here.70)
70See, for example, Bernstein (1992, 1998), Charny (1990), Greif (1998), and Klein and Leffler (1981).
Enforcement refers to actions taken by the tribunal when parties to the contract decide to come before it. Tribunals may impose money sanctions — so-called damages — for breach of contract or insist on specific performance of a contract — require parties to do what a contract specifies (for example, convey land). Tribunals may also fill gaps, settle ambiguities, and override terms in contracts.
4.1.2. Contract formation. The formation of contracts is of interest in several respects. Search effort. Parties expend effort in finding contracting partners, and it is apparent that their search effort will not generally be socially optimal. On one hand, they might not search enough: because the surplus gained when one party locates a contract partner will generally be divided between them in bargaining, the private return to search may be less than the social return.
On the other hand, parties might search more than is socially desirable because of a negative (“common pool”) externality associated with discovery of a contract partner: when one party finds and contracts with a second, other parties are thereby prevented from contracting with that party.71 Both of these externalities arise in Diamond and Maskin (1979), who examine a specific
model of search and contracting. Although policies to promote or to discourage search might be desirable, one wonders whether social authorities could obtain the information needed to determine the nature of problems with search effort.
Mutual assent and legal recognition of contracts.
A basic question that a tribunal must answer is at what stage of interactions between parties does a contract become legally recognized, that is, become enforceable. The general legal rule is that contracts are recognized if and only if both parties give a clear indication of assent, such as signing their names on a document. This rule obviously allows parties to make enforceable contracts when they so desire. Moreover, because the rule requires mutual assent, it protects parties against becoming legally obligated against their wishes. Thus, it prevents the formation of what would be undesirable contracts, and it means that search for contracting partners will not be chilled due to the risk of unwanted legal obligations.
However, certain legal doctrines sometimes result in parties becoming contractually bound without having given their assent; there exist cases in which a party became contractually bound when the other party with whom he was negotiating made substantial investments in anticipation of contract formation. This legal policy not only may result in undesirable contracts, it may also induce wasteful early investment as a strategy to achieve contract formation. It is true that early investment is sometimes efficient, but a party who wants to make early investment could attempt to advance the time of contract formation or make a preliminary contract about the matter. See Bebchuk and Ben-Shahar (1996), Craswell (1996), Katz (1996), and Wils (1993).
Compare our discussion in section 3.5.1 of excessive incentives to search for unowned property.
Offer and acceptance. Mutual assent sometimes is not simultaneous; one party will make an
If an offeror is held to his terms, offerees will often be led to invest effort in investigating contractual opportunities. Otherwise offerees might be extorted by offerors if the offerees expressed serious interest after investigation. The anticipation of such offeror advantage-taking would reduce offerees’ incentive to engage in investigation and thus diminish mutually beneficial contract formation. Hence, it may be in offerors’ and society’s interests for offered terms to be enforced for some period of time. Yet offerors’ circumstances may change, making it privately and socially advantageous for them to alter contract terms. On this and other issues concerning offer and acceptance, see Craswell (1996) and Katz (1990b, 1993).
Disclosure. The law may impose an obligation to disclose private information at the time of contract formation.72 Such a legal duty is beneficial in the respect that disclosed information may be desirably employed by the buyer; suppose, for instance, that he learns from the seller that the basement of his new house leaks and thus decides not to store valuables there. However, as initially emphasized by Kronman (1978a), a disclosure obligation discourages parties from investing in acquisition of information. For example, a company might decide against conducting aerial surveys to determine the mineral-bearing potential of land if it would be required to disclose
its findings to sellers of land, as sellers would then demand a price reflecting the value of the land.The social welfare consequences of the effect of a disclosure obligation on the motive to acquire information, analyzed in Shavell (1994), depend on whether the information is socially valuable or mere foreknowledge, on whether the party acquiring information is the buyer or the seller, and on inferences that would be made from silence.73
Duress and emergency. Even if both parties have given their assent, a contract will not be recognized if it was made when one of the parties was put under undue pressure, as when he is physically or otherwise threatened by another. This legal rule has virtues similar to those of laws against theft; it reduces individuals’ incentives to expend effort making threats and to defend against them.In addition, contracts may not be legally recognized if they are made in emergency situations, such as when the owner of a ship in distress promises to pay an exorbitant amount for rescue. Non enforcement in such situations beneficially provides victims with implicit insurance against having to pay high prices, but it also reduces incentives for rescue (yet rescue incentives might tend to be excessive, for the general reasons that there is excessive fishing effort).74
For discussions of various ways that asymmetry of information affects the contract terms that parties will agree to when there is no compulsory disclosure, see Ayres and Gertner (1989), Bebchuk and Shavell (1991), Spier (1992b), and Stole (1992).
73On inferences from silence in other contexts, see Fishman and Hagerty (1990), Grossman (1981), and Milgrom (1981).
74On rescue, see Landes and Posner (1978).
4.1.3. Why contracts and their enforcement are valuable to parties. At the most general level, parties make contracts when they have a need to make plans. They want contracts enforced to ensure that promised payments are made and to prevent opportunistic behavior that otherwise might occur over the course of the contractual relationship and stymy fulfillment of their plans. There are two basic contexts in which parties make enforceable contracts.
The first is that concerning virtually any kind of financial arrangement. The necessity of contract enforcement here is transparent. For example, because borrowers would not be forced to repay loans in the absence of contract enforcement, loans would be unworkable without enforcement. In financial arrangements, there is often a party who extends credit to another for some time period, and contract enforcement prevents his credit from being appropriated, which would render the arrangements impossible. In addition, financial contracts that allocate risk would generally be rendered useless without enforcement, even though there might not be an initial extension of credit, because once the risky outcome became known, one of the parties would not wish to honor the contract.
The second context in which parties make enforceable contracts involves the supply of custom or specialized goods and services — those which cannot simply be purchased on a spot market in a simultaneous exchange for money. The need for enforcement of agreements for supply of custom goods and services inheres mainly in averting what is often described as the holdup problem (discussed further in sections 4.2.1 and 4.2.2). To illustrate, consider a buyer who wants a custom desk which would be worth $1,000 to him and would cost $700 for a seller to produce. In the absence of contract enforcement, the buyer will not pay the seller in advance (for the seller could walk away with what he receives).
The buyer will pay the seller only after the seller makes the desk. But at that point, the seller’s production cost is sunk and he is vulnerable to holdup; the situation is that he has a desk which, being custom-made, has little or no alternative value.75 The outcome of bargaining between him and the buyer might thus be a price lower than the seller’s cost of $700; say the price is $500. If so, and the seller anticipates receiving only the $500 price, he will not produce the desk. This is true even though production and sale at a price between $700 and $1,000, such as $800, would be mutually beneficial for the seller and the buyer.
75Similar forms of holdup would arise in the absence of contract enforcement where parties want to convey property that already exists, such as land; for instance, a seller might worry about being held up by the buyer if he waits and forgoes a present opportunity to sell his land to a new party who makes a bid for it
Enforcement of the buyer’s promise to pay $800 for the desk on delivery, or of the seller’s promise to produce and deliver the desk (if the buyer paid the price of $800 in advance), is thus desirable for the parties.
More broadly, enforcement of contracts will stimulate all manner of investments which, like the seller’s expenditure on production, have specific value in a contractual relationship.
Enforcement will lead buyers to train workers to use new contracted-for equipment, sellers to engage in research to reduce production costs, and so forth. In the absence of contract enforcement, there would be too little investment in these things, for, at the final stage of negotiation for performance and for payment, each side would be subject to holdup by the other, so would tend to obtain only a part of the surplus created by its investment.
The foregoing idea of contract enforcement as a cure for holdup-related underinvestment was initially stressed in the economics literature by Klein, Crawford, and Alchian (1978), Grout (1984), and Williamson (1975). However, the general notion that contract enforcement is privately and socially desirable because it fosters production and trade is made (usually with little articulation) by most writers on contract law and, one supposes, has always been appreciated.
See, for example, Farnsworth (1982, pp. 16-17) and Pound (1959, pp. 133-134).
4.1.4. Incomplete nature of contracts and their less-than-rigorous enforcement.
Although enforceable contracts are desirable, they are observed to be substantially imperfect. They are significantly incomplete, leaving out all manner of variables and contingencies that are of potential relevance to contracting parties, and they also often fail to employ included variables in a
mutually beneficial manner. Moreover, contracts are not enforced rigorously, despite the seeming strength of the reasons for contract enforcement: penalties for violation of contractual obligations are often modest, and breach is not an uncommon event. There are three important reasons for the incompleteness of contracts. The first is the cost of writing more complete contracts.
Parties may not include variables in a contract, or not in a detailed, efficient way, due to the cost of evaluating, agreeing upon, and writing terms. (In particular, parties will tend not to specify terms for low probability events, because the expected loss from this exclusion will be minimal, whereas the cost of including the terms is borne with certainty.)
The second reason for incompleteness is that some variables (effort levels, technical production difficulties) cannot be verified by tribunals.76 Of course, many such variables can be made verifiable (effort could be made verifiable through videotaping), but that would involve expense.
The third reason for the incompleteness of contracts is that the expected consequences of incompleteness may not be very harmful to contracting parties. Incompleteness may not be harmful simply because a tribunal might interpret an imperfect contract in a desirable manner. In addition, as we shall see, the prospect of having to pay damages for breach of contract may serve
as an implicit substitute for more detailed terms. Furthermore, the opportunity to renegotiate a contract often furnishes a way for parties to alter terms in the light of circumstances for which contractual provisions had not been made. Finally, in some settings parties’ concern for their reputation may induce them to refrain from opportunistic behavior. That contracts are less than rigorously enforced is intimately related to their incompleteness.
For incomplete contracts not to disadvantage parties, tribunals must be able to reinterpret or override imperfect contractual terms rather than always enforce these terms as written. Also, for damage measures to be employed beneficially by parties, notably for parties to be able to escape from contractual obligations when performance and renegotiation are difficult, damages for breach must not be excessive. Additionally, for parties to avoid bearing high risks in the form of
payments that they would be induced to make when renegotiating imperfect contractual terms, the damages for breach must again not be severe. These points will be expanded in the discussion below of contract interpretation, remedies for breach, and renegotiation.
4.1.5. Interpretation of contracts. Contractual interpretation, which includes a tribunal’s filling gaps, resolving ambiguities, and overriding literal language, can benefit parties by easing their drafting burdens or reducing their need to understand contractual detail.77 For example, if it is efficient to excuse a seller from having to perform if his factory burns down, the parties need not incur the cost of specifying this exception in their contract, assuming that they can trust the tribunal to interpret their contract as if the exception were specified.78
The problem of unverifi ability of variables is diminished by the possibility that parties can plan in their contract to use a tribunal of experts in their area, such as individuals in the same business as the contracting partners. In many industries, this practice is common.
77On various aspects of contract interpretation, see, for example, Ayres and Gertner (1989), Hadfield (1994), and Schwartz (1992).
78Another example, where it may be efficient for a tribunal to override particular terms that appear in contracts, is when a seller offers only a detailed, fine-print contract in conjunction with the sale of an inexpensive good or service. Because it would be irrational for consumers to read such contracts, sellers would have incentives to include inefficient, onesided terms if such terms would be enforced. See Katz (1990c). The extent to which such contracts will be problematic will depend on the fraction of consumers who are informed about contract terms and shop among competing sellers. See Schwartz and Wilde (1979).
It may be worthwhile elaborating somewhat by viewing contract interpretation more formally, as a function that transforms the contract individuals write into the effective contract that the tribunal will enforce. Given a method of interpretation, parties will choose contracts in a constrained-efficient way. Notably, if an aspect of their contract would not be interpreted as they want, the parties would either bear the cost of writing a more explicit term that would be respected by the tribunal, or else they would not bear the cost of writing the more explicit term and accept the expected loss from having a less than efficient term. The best method of contract interpretation will take this reaction of contracting parties into account and can be regarded as implicitly minimizing the sum of the costs the parties bear in writing contracts and the losses resulting from inefficient enforcement.79
4.1.6. Damage measures for breach of contract. When parties breach a contract, they often have to pay damages in consequence. The damage measure, the formula governing what they should pay, can be determined by the tribunal or it can be stipulated in advance by the parties to the contract.80 One would expect parties to specify their own damage measure when it would better serve their purposes than the measure the tribunal would employ, and otherwise to allow the tribunal to select the damage measure. In either case, we now examine the functioning and utility of damage measures to contracting parties (assuming here that there is no renegotiation of contracts). Clearly, the prospect of payment of damages is an incentive to perform contractual obligations, and thus generally promotes enforcement of contracts and the goals of the parties, as
discussed in section 4.1.3. As emphasized in section 4.1.4, however, damages for breach in fact are not chosen to be so high that they virtually guarantee performance of contracts as written. Under the commonly employed expectation measure, damages equal the amount that compensates the victim of breach for his losses; these damages are often quite willingly paid by a party who commits a breach.
Why are damages not chosen to be so high as to guarantee performance? An important explanation is that parties do not always want performance of the less-than-complete contracts that they write. For example, suppose that a contract is very incomplete: it merely states, “The seller will produce a custom desk for the buyer and receive full payment of $800 in advance.”
The determination of the optimal method of interpretation may involve subtleties. For example, according to the optimal method, a term might not be interpreted in the way that is best in the majority of transactions. Suppose that term A is best in the majority of transactions and that the parties to these transactions can include A explicitly, at little cost on a percontract basis, because they are repeat players. Suppose that term B is best only in the minority of transactions, but that for the parties to these transactions to include B explicitly will not be cheap on a per-contract basis because they are not repeat players. Then the optimal method of interpretation would make B the default term even though it is best only in a minority of transactions.
80A contractual provision that states a particular amount of damages is referred to as a liquidated damages clause.
The buyer and the seller do not really want the desk always to be produced. It is readily shown that, had they made a Pareto efficient complete contract, they would have specified that there should be performance if and only if the production cost is less than the $1,000 value of the desk to the buyer. (For instance, in a complete contract, they would have jointly decided against a contractual term specifying performance when the production cost is $2,000, for the seller would have been willing to reduce the contract price sufficiently to induce the buyer to strike the term.)
Now if the incomplete contract calling for the desk always to be produced is enforced by the expectation measure of damages of $1,000, the seller will behave exactly as he would have under the Pareto efficient complete contract, that is, he will perform if and only if the production cost is less than $1,000. Higher damages than the expectation measure might induce performance when it is inefficient, and lower damages might lead to breach when that is inefficient. Indeed, for this reason, the parties would often agree to choose the expectation measure over other measures of damages.
This understanding of damage measures as a device to induce the behavior that the parties would have specified in more complete contracts sheds light on the notion held by some legal commentators and philosophers that contract breach is immoral, that it constitutes the breaking of a promise. That belief is often incorrect, it is submitted, and might fairly be considered to be the opposite of the truth. The view that a contract breach is the breaking of a promise overlooks the point that the contract that is breached is generally an incomplete contract, and that the “breach” constitutes behavior that the parties truly want and would have provided for in a complete contract. In the example of the simple incomplete contract calling for a desk to be produced, the seller who finds that his production cost would be $2,000 will commit breach under the expectation measure. But in so doing, he will be acting precisely as would have been set out in a complete contract, and it is that contract which is best regarded as the promise between the parties that ought to be kept.
The point that a moderate damage measure, and in particular the expectation measure, is desirable because it induces performance if and only if the cost of performance is relatively low was apparently first clearly stated (informally) in Posner (1972), who emphasized the social efficiency of the measure. Shavell (1980b) formally demonstrated this and also stressed the mutual desirability of the measure for contracting parties and its role as a substitute for more complete contracts.
Two other writers, Birmingham (1970) and Barton (1972), adumbrate these points, although the meaning of their articles is at times obscure. See also Diamond and Maskin (1979), who consider damage measures in analyzing search behavior
Several more comments should be made about damage measures and incentives. First, damage measures influence the motive of contracting parties to make reliance investments (so called because the investments are made relying on contract performance). Reliance investments are illustrated by the earlier-noted instance of a buyer training workers to use a contracted-for machine or by advertising the contracted-for appearance of an entertainer. Under the expectation measure, there is a tendency for reliance investment to exceed the Pareto efficient level: the buyer will treat an investment like advertising as one with a sure payoff — either he will receive performance or receive expectation damages, a form of insurance — whereas the actual return to investment is uncertain, due to the possibility of breach (advertising will be a waste if the
entertainer does not appear). This tendency toward overreliance due to the receipt of contract damages was initially noted in Shavell (1980b), and stands in contrast to the problem of inadequate reliance investment associated with lack of contract enforcement. The issue of reliance investments has been elaborately analyzed, as will be described in section 4.2.2.
A second comment is that the value of damage measures as an incentive toward efficient performance would not exist if renegotiation of contracts in problematic contingencies would always result in efficient performance. But, as will be discussed below, it seems plausible that renegotiation would not always result in efficiency.
An important function of damage measures which is quite distinct from their incentive role concerns risk-spreading and compensation. Notably, because the expectation measure compensates the victim of a breach, the measure might be mutually desirable as a form of insurance if the victim is risk averse. However, the prospect of having to pay damages also constitutes a risk for a party who might commit breach (such as a seller whose costs suddenly rise), and he might be risk averse as well. The latter consideration may lead parties to want to lower damages (see Polinsky 1983) or to avoid use of damages as an incentive device, by writing more detailed contracts (for instance, the parties could go to the expense of specifying in the contract that a seller can be excused from performance when his costs are high).82 A full consideration of damage measures and efficient risk allocation would also take into account whether the risk that a party bears is detrimental or beneficial,83 whether the risk is monetary or non-monetary,84 and whether the parties can obtain insurance.
82When parties do not so specify in advance, certain legal doctrines may serve this function. See Joskow (1977), Posner and Rosenfield (1977), and Sykes (1990).
83For example, if a party wants to breach because he has a superior opportunity, optimal damages might be higher, although adjusting damages in the case of beneficial risks is not likely to matter as much on risk-bearing grounds.
84For example, if the victim’s loss is non-monetary, such as the loss due to failure of musicians to appear at a wedding, financial compensation in the form of damages may not constitute an optimal form of insurance. See section 2.4.2.
4.1.7. Specific performance as a remedy for breach. As observed at the outset, an alternative to use of a damage measure for breach of contract is specific performance: requiring a party to satisfy his contractual obligation.85 Specific performance can be accomplished with a sufficiently high threat or by exercise of the state’s police powers, such as by a sheriff removing a person from the land that he promised to convey. (Note that if a monetary penalty can be employed to induce performance, then specific performance is equivalent to a damage measure with a high level of damages.)
It is apparent from what has been said about incomplete contracts and damage measures that parties should not want specific performance of many contracts that they write, for they do not wish their incomplete contracts always to be performed. It is therefore not surprising that, in fact, specific performance is not used as the remedy for breach for most contracts for production of goods and for provision of services. Additionally, it may be observed that specific performance might be peculiarly difficult to enforce in these contexts because of problems in monitoring and controlling parties’ effort levels and the quality of production.
However, specific performance does have advantages for parties in certain contexts, such as in contracts for the transfer of things that already exist, like land, and specific performance is the usual legal remedy for sellers’ breaches of contracts for the sale of land. This point is discussed briefly below, in section 4.3.1. On specific performance and its general comparison to damage remedies, see Bishop (1985), Kronman (1978b), Schwartz (1979), Shavell (1984b), and Ulen (1984). (Specific performance also is examined in some of the articles on production contracts cited in section 4.2.2.)
4.1.8. Renegotiation of contracts. Parties often have the opportunity to renegotiate their contracts when problems arise. Indeed, the assumption that they will do this has appeal because, having made an initial contract, the parties know of each other’s existence and of many particulars of the contractual situation. For this reason, much of the economics literature on contracts assumes that renegotiation always occurs when inefficiency would otherwise result; see, for example, Hart (1987), Hart and Holmstrom (1987), and Rogerson (1984).
Nevertheless, in many circumstances contracts will not be renegotiated because parties are not in contact with each other when difficulties are experienced and one party would benefit from acting quickly.
85Some economists have employed the term “specific performance” in an unconventional sense, to refer to
enforcement of all provisions in a contract, including any damage measure named in it. Thus, they would say that a contract is specifically performed when the parties name expectation damages in their contract and parties who breach are thus required pay these damages.
A problem may occur during the course of production and the producer may have to decide on the spot whether to abort the process or proceed at greater cost. Or a new bid may be heard and have to be immediately answered. Furthermore, even if the parties are in contact with one another, asymmetric information may lead to breakdowns in renegotiation.
In any event, let us assume that successful renegotiation tends to occur and consider how it affects the welfare of contracting parties. Plainly, renegotiation often allows parties to avert Pareto inefficient breach decisions. For example, if damages exceeding the expectation measure or specific performance were the remedy for breach, a seller might be led to perform when his production cost exceeds the value of performance to the buyer. To avoid this inefficient outcome, the seller might pay the buyer to release him from his obligation to perform. That renegotiation may result in performance if and only if it is efficient means, as we noted, that damage measures for breach are not necessary to accomplish this, and also helps to explain why contracts lack detail.
But even if renegotiation tends to occur, it may represent only a partial substitute for explicit contractual terms or for appropriate damage measures for breach. One reason (see section 4.2.3) is that renegotiation cannot affect actions that are taken before the time of renegotiation, which influence the likelihood of nonperformance; renegotiation can only affect future decisions about
breach. Another reason involves the allocation of risk-bearing. Consider, for instance, the substantial risks borne by a producer who may have to purchase a release from an obligation to perform when his production costs would be extremely high. Such risks could be mitigated by use of a clause excusing him from performance or by a damage measure such as expectation.
Additionally, the prospect of renegotiation affects the incentives of parties to invest in the contractual relationship. A party’s level of reliance investment will be inefficient if renegotiation results in the extraction of part of the surplus that the party’s reliance investment creates. Yet renegotiation is influenced by, among other elements, the damage measure that applies for breach; and if the damage measure is appropriately chosen, the damage measure together with
renegotiation may, in principle, spur desirable reliance investment; see section 4.2.2.
One presumes that the ability to renegotiate is usually desirable for contracting parties, because it allows them to improve their situation when difficulties arise and to write simpler contracts than otherwise.
Thus, we would expect that parties will want their renegotiated contracts enforced, and the law generally does enforce renegotiated contracts. However, the ability to renegotiate can also work to the detriment of parties because they might thereby be prevented from committing themselves to particular outcomes in their initial contract. See Jolls (1997) and the literature cited therein, especially Fudenberg and Tirole (1990). Nevertheless, the law usually prevents parties from binding themselves not to renegotiate, even though that could in theory be done.86
4.1.9. Legal overriding of contracts. A basic rationale for legislative or judicial overriding of contracts is the existence of externalities. Contracts that are likely to harm third parties are often not enforced, for example, agreements to commit crimes, price-fixing compacts, liability insurance policies against fines, and certain simple sales contracts (such as for machine guns).87 Another general rationale for nonenforcement of contracts is to prevent a loss in welfare to one or both of the parties to contracts. This concern may motivate nonenforcement when a party is incompetent, lacks relevant information, or is in an emergency situation (see section 4.1.2).
The rationale also applies in the context of contract interpretation by tribunals; as discussed in section 4.1.5, contract interpretation may amount to overriding terms of contracts, and this may promote the welfare of contracting parties by allowing them to write simpler contracts.88
86It is true that parties will not usually be able to bind themselves against engaging in renegotiation, for they could ordinarily renegotiate in secret. However, as Jolls (1997) observes, one of the parties will usually prefer that the original contract be enforced, so that if the courts stand ready to enforce the original contract, renegotiation cannot result in a new contract. For example, in the standard principal-agent contract, after the agent exerts effort, the principal and the agent will have an incentive to arrange for the agent to be paid a constant amount. But if this were contemplated, then after the output is realized, the agent would have an incentive to assert the original contract if his pay would be higher according to it, and the principal would assert the original contract if he could pay less under it.
87See also Posner (1995), who suggests that such contractual limits as usury laws, which constrain consumers’ ability to borrow, might be justified by a type of externality: when high-risk borrowers fail, they may become eligible for social welfare programs, imposing costs on taxpayers.
88Also, at least in theory, nonenforcement of contracts might also be beneficial to parties where they would be led to include terms constituting wasteful signals of unobservable characteristics. See Aghion and Hermalin (1990).
Additionally, contracts sometimes are not enforced because they involve the sale of things said to be inalienable, such as human organs, babies, and voting rights. In many of these cases, the inalienability justification for lack of enforcement can be recognized as involving externalities or the welfare of the contracting parties.89
4.2. Production Contracts
In this section, the literature on production contracts is discussed. The first case considered is that where symmetrically informed, risk-neutral parties enter into contracts, and the only variables of concern are the value of performance and production cost. Then the case where parties make reliance investments to raise the value of the contract during the contract period is examined. Finally, several other issues, including risk-bearing and asymmetric information, are reviewed. Throughout, when remedies for breach are discussed, one can imagine them either to be chosen by the parties or by the courts.
4.2.1. Value of performance and production cost. Assume that a risk-neutral buyer and a risk-neutral seller have met; the seller faces uncertain production cost c, which he will learn before he decides whether to produce; v is the certain value of performance to the buyer; and the parties are symmetrically informed. The Pareto efficient outcome is for the seller to produce if and only if c < v. (That is, in a complete contract, with terms for all contingencies, performance would be required if and only if c < v; a change in the contract price would compensate a party for agreeing to alter a term from any initially considered contract under which performance does not occur if and only if c < v.)
In the absence of contract enforcement, then (amplifying on section 4.1.3) there would be too little production because the buyer would only pay the seller for actual delivery of the good and cannot guarantee the price. In particular, supposing that the seller would obtain a fraction á of the surplus from a transaction (á reflects bargaining strength), he would obtain a price of áv.
(After the seller produces the good, the surplus from the transaction would be v, presuming for simplicity that the custom good has no alternative value for the seller.) Thus, the seller would decide to produce only when c < áv, rather than whenever c < v.
Suppose now that there is contract enforcement and that the parties are not able to renegotiate before the seller decides whether to produce (an assumption that is relaxed below). If c is verifiable by the tribunal, the parties could write a complete contract specifying performance if and only if c < v. The parties would want a damage measure d for breach of this contract to be sufficiently high to induce performance when c < v, and thus any d exceeding c would work.
89See generally Rose-Ackerman (1985) and Trebilcock (1993).
If c is not verifiable, the parties are able to write an incomplete contract specifying “The seller shall deliver the good to the buyer, who will pay price p at the outset,” accompanied by damages d for seller breach. Under such a contract, the seller will perform when c < d and will commit breach otherwise.90 If the expectation measure is employed, that is, d = v, the seller will perform if and only if c < v, so that performance will be efficient.91 If damages d exceed v, there will be excessive performance, as there will be if there is specific performance. If d is less than v, there will be too little performance. The points of these paragraphs were, as noted, emphasized in
Posner (1972) and Shavell (1980b). If, instead, it is assumed that the buyer and the seller can renegotiate their contract after c becomes known but before the seller decides whether to produce, then, given symmetric information, it is natural to suppose that there will always be Pareto efficient performance, regardless of d. Let us also note that if the buyer’s value v is uncertain as well as the seller’s production cost c, the major difference in the outcome is that, since v cannot be prescribed as damages in the contract, v must be verifiable for the expectation measure d = v to be applied by the tribunal (c still need not be verifiable).92
4.2.2. Reliance investment during the contract period. Now assume that parties can make investments during the period of the contract that affect its value v or the production cost c. Such investments are, as noted, sometimes called reliance investments, since they are made in anticipation of contractual performance. We will begin with the case in which just one party invests before discussing the case where both sides invest.
Because we assume that the price p is paid at the outset, the seller faces cost c if he performs and will compare it to damages of d that he would have to pay if he breaches. If the price were to be paid only at the time of performance, then
the seller would perform if and only if c – p < d. Hence, the performance that is induced under d if the price is paid at the outset will be achieved under d¢ = d – p if the price is paid only at performance.
91A related issue concerns post-breach mitigation behavior of the buyer: efficiency requires that if there is a breach, the buyer should mitigate the consequences of breach by searching for alternative suppliers and the like. Let z be mitigation expenditure of the buyer to raise his post-breach value, say w(z). Efficiency requires the buyer to choose z to maximize w(z) – z; let z* be the optimal value of z. If y is the gross value of seller performance to the buyer, then we can define v, the net value of performance, as v = y – (w(z*) – z*). Thus, expectation damages for breach should equal this v, not the gross value y.
And if damages equal v, then the buyer will choose z* if he is the victim of a breach, and the net value of performance will actually be v. On this issue of mitigation of the consequences of breach, see, for example, Wittman (1981).
92However, if c is verifiable and v is not, Pareto efficient performance can be achieved by constructing the contract so that the buyer will commit breach by refusing to pay for performance when performance would be inefficient. Specifically, let the price p be paid at performance, and let damages for buyer breach be d = p – c, the seller’s profits. Then the buyer will breach and refuse performance whenever v – p < -(p – c), or when v < c. (If, as is realistic, it is assumed that p – c cannot be negative, then the parties can choose p high enough that it always exceeds c (assuming c is bounded), with the buyer being compensated for the high p through an up-front rebate.) The parties’ ability to determine who will make the breach decision, as described here, is emphasized in Edlin (1996).
93We comment in note 95 below on another case of reliance investment: where the party who chooses the reliance investment is the same party who faces uncertainty, such as where the seller chooses r to lower his production cost and faces uncertainty about his production cost.
In the absence of contract enforcement, there will be too little production, as before; it will occur only when c < áv(r). But now, in addition, the buyer will choose an incorrect value of r because he will only obtain a fraction 1 – á of the value created by investment.94 Assume next that there is contract enforcement and that the parties do not renegotiate before the seller’s production decision (we relax this assumption below). This is the setting analyzed in Shavell (1980b), who first studied reliance investment. If c and r are verifiable by the tribunal, the parties can write a contract specifying efficient performance (when c < v) and also specifying r*; again, they would want the contract enforced by a damage measure high enough to ensure performance, and any such measure of damages would serve their purposes.
Now assume that c and r are not verifiable, that the parties write a simple contract specifying “The buyer will pay price p at the outset and the seller will deliver the good to him,” and consider what occurs under different damage measures. If the expectation measure is employed, that is, d = v(r), the seller will perform when c < v(r), so that performance will be efficient. However, as the buyer will always receive v(r) (either he obtains performance, worth v(r), or damages of that amount), he will choose r to maximize v(r) – r. Consequently, the buyer will select an inefficiently high r; the problem is that the buyer does not take into account that investment does not have any value when performance does not occur.95 Under a sophisticated expectation measure based on efficient investment, namely d = v(r*), however, investment as well as performance can be shown to be efficient.
Specifically, he will choose r to maximize (1-á)v(r)G(áv(r)) – r, so the first-order condition determining r is (1-á)v(r)G(áv(r)) + (1-á)v(r)áv(r)g(áv(r)) = 1, or (1-á)v(r)[G(áv(r)) + áv(r)g(áv(r))] = 1. Although one might expect r to be
less than r*, it is apparent from the latter first-order condition that there is a possibility that the r chosen would exceed r*. The reason is that increasing r raises the probability that the buyer will obtain performance from the seller.
95We observe that the problem of an inefficiently high r does not arise under the expectation measure where the seller makes the reliance investment to lower his production cost and also faces uncertainty about it. Specifically, suppose
that production cost is c(r,è), where è is an uncertain state of nature, cè > 0, and c r < 0. In this case, the seller will choose the efficient r; the explanation in essence is that then the seller obtains the benefit of his reliance only when there is performance. The efficient r is that maximizing
è(v,r) (v – c(r,è))g(è)dè – r, 0 where è(v,r) is the è such that c(r,è) = v. Thus, r* is determined by -cr(r,è))g(è)dè = 1. Now under the expectation measure,
the seller will perform when c < v and pay v otherwise. Thus, the seller chooses r to maximize è(v,r) p – c(r,è)g(è)dè – v(1 – G(è(v,r))) – r, 0 and, differentiation of this yields the same condition as that which determines r*. This point is noted in Shavell (1980b).
96If d = v(r*), the seller will perform when c < v(r*), so the buyer will maximize v(r)G(v(r*)) – v(r*)(1 – G(v(r*))) – r. Accordingly, r will be determined by v(r)G(v(r*)) = 1, and this condition is clearly satisfied at r*. The explanation is that the buyer’s choice of r affects his return only when he obtains performance.
Next assume that the parties do renegotiate after the reliance investment is made and before the seller decides about production, so that, assuming symmetric information, there will always be efficient performance. This version of the model of production contracts was originally studied by Rogerson (1984). Here, damage remedies may influence investment through their effect on the outcome of renegotiation. To illustrate, consider what would occur under specific performance. Under this remedy, as suggested earlier, there will be renegotiation in which the seller pays the buyer to be allowed not to perform whenever c > v(r), since then performance would be inefficient. In particular, the assumption is that the seller would pay the buyer v(r) + (1 – á)(c – v(r)) to be allowed not to perform; for v(r) is needed to compensate the buyer for not receiving performance, 1 – á is the buyer’s share of the surplus from renegotiation, and c – v(r) is that surplus. Anticipating this, the buyer can be shown to choose an r exceeding the efficient level.97
The nature of the results about reliance investment in the case with renegotiation are very close tom those where there is no renegotiation. Indeed, they are identical under the expectation measure, essentially because there is no renegotiation under the expectation measure; thus, with d = v(r), investment will be excessive because the buyer will always be compensated for his investment. Furthermore, under the sophisticated expectation measure based on efficient investment, d = v(r*), investment will be efficient. See Spier and Whinston (1995).. For example, under the expectation measure, investment will tend to be excessive for both parties, but performance will be efficient. Much recent literature, beginning with Hart and Moore (1988), has focused on this general situation, assuming that parties can renegotiate after reliance investments are made and è is revealed, and that they will always then agree on efficient production decisions because information is symmetric. The literature in question furthermore usually supposes that none of the variables (costs, values of performance, reliance investments) are verifiable by the tribunal. Thus, a contract can depend only on what is recorded in it, certain subsequent communications between the parties, whether there has been performance, and, if not, who committed breach.
In general, the buyer will choose an r in between the excessive level he would choose under the expectation measure (determined by v(r) = 1) and r*. This can be explained as follows. If the buyer’s fraction of surplus is 0, he will receive v(r) whether or not there is renegotiation, so his situation will be the same as under the expectation measure. If the buyer’s fraction of surplus is 1, he will clearly choose r*. This suggests what can be shown, that if the buyer’s fraction of surplus is positive and less than 1, he will choose an r exceeding r* and less than the r chosen under the expectation measure. Of note are a number of results establishing the existence of contracts that will produce efficient outcomes, that is, in both parties choosing efficient levels of reliance investment (performance will always be efficient). Aghion, Dewatripont, and Rey (1994) and Chung (1991) demonstrate the efficiency result using a contract in which one party is effectively given the right to make a single take-it-or-leave-it offer to the other in renegotiation. It is evident that this party will invest efficiently, as he can extract in bargaining the full marginal return from his investment.
For instance, if the buyer has the right to make an offer and is paying the seller to perform, he will pay only the minimum needed to induce the seller to do so, and will obtain any increase in value v(r,è) due to his having chosen a higher r. Less apparent is how the other party is given an incentive to invest efficiently; that is accomplished by properly choosing the quantity of the good or the probability of delivery. (For instance, if the named quantity of the good is chosen to be higher than is likely to be efficient, the buyer will usually pay the seller to agree to lower the quantity. The amount the buyer will pay must compensate the seller for the profits he would have
made at that higher contracted-for quantity. But the profits the seller would have made at that quantity will depend on his investment in lowering production costs — thereby giving the seller an incentive to invest in lowering his production costs, and an incentive that is greater the higher the contracted-for quantity.) Also, Noldeke and Schmidt (1995) establish that a simple option
contract will induce efficient investments for reasons that are closely related to those just reviewed. Additionally, Edlin and Reichelstein (1996) and Hermalin and Katz (1993) adduce contracts leading to efficiency under somewhat different conditions, and Rogerson (1992) shows that efficiency can be achieved under wide circumstances, but assuming that parties can commi not to renegotiate their contracts. Cooperative reliance investments. It has been assumed above that a reliance investment benefits directly only the party who makes it. Another possibility is that a reliance investment benefits the other party to the contract; importantly, suppose that a seller’s investment raises
product quality and thus value for the buyer. Such cooperative reliance investment is studied in Che and Chung (1999). As they emphasize, when cooperative investment cannot be verified by courts, then under the expectation measure, there will be too little investment (in contrast to the usual case under the expectation measure, where investment is excessive). Indeed, there will be no investment if the seller who makes a cooperative investment will not benefit directly or in damages he receives in the event of breach. Moreover, there is no contract that will result in efficient cooperative investment (again in contrast to the usual case); this point is stressed in Che and Hausch (1999), who also demonstrate that contracting offers no advantage over no contracting in a wide set of circumstances.98
For recent attempts to provide a unified theoretical framework for the various problems concerning reliance discussed in section 4.2.2, see Maskin and Moore (forthcoming) and Segal and Whinston (1998). Asymmetric information. Another factor about production contracts that we have not examined is asymmetric information between the parties. When parties are asymmetrically 4.2.3. Further considerations. Risk-bearing. We have not discussed in this section on production contracts the allocation of risk among possibly risk-averse contracting parties, about which several comments should be made. First, if all variables are verifiable by a tribunal, the presence of risk-averse parties does not affect when it is Pareto efficient to perform; it continues
to be efficient to perform if and only if c < v. However, efficiency requires that the resulting risk be allocated appropriately; for instance, if the seller is risk averse and the buyer risk neutral, the seller would be insured against fluctuations in c by the buyer’s paying him c plus a constant. In addition, the level of efficient reliance investment will generally be affected by considerations of risk-bearing. Second, when variables of relevance are not verifiable, then damage measures and other mechanisms that may be employed to induce efficient behavior when parties are risk-neutral have to be reconsidered. For instance, the expectation measure imposes risk on the party who might breach and pay these damages; if that party is risk averse, the expectation measure would become less attractive relative to lower measures of damages. Furthermore, as we earlier noted, renegotiation does not generally lead to efficient risk bearing, even though it may lead to efficient performance.informed, renegotiation of contracts might not be successful, so that it becomes more important that the initial contract induces efficiency. Hermalin and Katz (1993) show that efficiency can be achieved under certain types of asymmetry of information using a relatively complicated mechanism in the contract.
New entrants. We have not examined the possibility that new buyers would appear and bid for the seller’s good (a similar possibility is that new sellers would appear and make offers to the buyer). In this regard, it should be noted that it is Pareto efficient for the initial contracting parties that a sale be made to a new buyer if and only if his bid exceeds the contract buyer’s valuation.
Moreover, the contracting parties will want to maximize the amount that they can extract from a new buyer if he purchases the good. This observation raises the possibility that the buyer and the seller may wish to set damages for seller breach at a high level in order to induce a new party to bid more (which he would have to do to make it in the seller’s interest to commit breach). Such an incentive of contracting parties to set damages at high levels can, though, result in too little breach and sale to new parties; thus, at least in principle, the incentive in question is a ground for tribunals not to enforce the high damage level specified by the contracting parties. This point was first made in Diamond and Maskin (1979) and has been refined in a number of articles; see Aghion and Bolton (1987) and Chung (1992).
However, Spier and Whinston (1995) observe that threeway renegotiation would seem to vitiate the advantage to the contracting parties of setting high damages. Yet they emphasize another reason (concerning induced reliance investment) that the parties will, after all, benefit from setting high damages. Precautions and probabilistic breach. It has been supposed throughout that breach occurs when a party decides not to perform, but often breach does not occur in this way: rather a party chooses a level of precaution which affects the likelihood of performance, and a random factor then determines whether breach or performance results. For example, a shipper’s care in packing dishes affects the likelihood that they will arrive unbroken, and a chance event (a jolt) determines whether they arrive broken or unbroken. In this setting, the conclusions reached about damage measures in the absence of renegotiation continue to apply: the expectation measure results in efficient precautions, the buyer’s reliance investment is excessive, and so forth. The very issue of renegotiation is made moot because the precautions are chosen before breach might occur (if the dishes arrive broken, it is too late for renegotiation). See Cooter (1985), Craswell (1988), and Kornhauser (1983).
4.3. Other Types of Contract
4.3.1. Contracts for transfer of possession. A different contractual context from production is where something that already exists is to be conveyed to a buyer. Examples include contracts for transfer of real estate, goods in inventory, and durable goods. Here a major uncertainty of interest concerns bids by new parties. With regard to these bids, the points just discussed concerning new entrants apply; the parties would like for there to be a sale to a new buyer when he will pay more than the contract buyer’s valuation, and so forth. It is of interest to explore why contracting parties often adopt specific performance as the remedy for breach of contracts for transfer of possession, even though damage measures are commonly employed for other types of contract. Initially, suppose that the contract buyer and the contract seller have equal access to bids from new parties. Then the buyer’s always receiving the good does not result in any loss of opportunity to sell to a new party willing to bid a high amount; that is, specific performance does not suffer from any clear disadvantage relative to damage measures that would allow the seller to breach and sell to a new party. Moreover, specific performance offers the parties an advantage over damage measures. Namely, because under specific performance it will always be the buyer who will be bargaining with a new party, the good will never be sold to a new party bidding less than the buyer’s valuation. In contrast, such a sale could occur if the seller might pay damages, commit breach, and bargain with a new party (suppose that bargaining is not three-way, involving the contract buyer as well). And, after such a sale, the buyer would have to obtain the good through repurchase from the new party, but in general this will be at a higher price than the seller obtained — meaning that some of the surplus would be shared with the new party. (Although the contracting parties would be worse off if the buyer repurchases at a higher price, society would not be worse off as the good would still be allocated to the user who places the highest value on it.) See Shavell (1984b) and Bishop (1985). The foregoing advantage of specific performance in preventing inefficient sales to new parties is clearly reduced if the buyer does not have equal access to bids from new parties (suppose that the seller is a dealer and the buyer is not). Also, the use of specific performance might increase transaction costs, if the new party turns out to purchase only after delivery of the good to the buyer.m Notice too that some of the disadvantages of specific performance in the production context are less significant in the present context of transfer of possession. In production contracts, specific performance imposes a possibly large risk of loss on sellers whose production costs might be very high; here, specific performance only reallocates a beneficial risk (of a sale at a high price) from seller to buyer. In addition, enforcement of specific performance in the context of contracts for transfer of possession is often easier than in the production context, where enforcement might involve policing the quality of production or services.
4.3.2. Donative contracts. An important category of contractual arrangement is donative, concerning gifts. Assuming that the motivation for gifts is altruism,99 a basic question is why a donor would want to defer his gift rather than make it immediately (in which case no contract would be required). The answers include the possibilities that the donor may face liquidity constraints and that he may wish to wait for resolution of uncertainties concerning, among other factors, his own needs and future income, and the donee’s needs, future income, and subsequently revealed character. However, if the donee knows about the altruism of the donor, a contract may not be necessary to induce donee reliance activity; if so, a contract would be disadvantageous for the donor. On these issues, see Goetz and Scott (1980) and Shavell (1991a); and see also Posner (1977b) and Posner (1997).
Other motives for gift giving exist (such as obtaining utility from expressions of gratitude from donees); some have similar implications to those of altruism.
4.3.3. Additional types of contract. In this section, mention has not been made of many additional types of contract, including principal-agent contracts, even though they have been studied, often intensively, in the economics literature. The omission of such contracts is explained in part by convention (by what is and is not considered to be a law and economics topic) and in part by the relative inattention that has been paid to contract enforcement.