Printer Friendly

Contract interpretation redux.



   A. The Timing and Nature of Interpretation Issues
   B. The Problems the State Must Solve
   C. An Introduction to Our Theory of Interpretation
      1. The Case for Party Control
      2. Parties Prefer Textualist Defaults

   A. Risk Neutrality
   B. Strategic Behavior
   C. Accuracy, Context, and Party Preferences
   D. The Evidence

   A. Recovering the Contract Terms: The Parol Evidence Issue
   B. Resolving Disputes over the Meaning of Contract Terms



Interpretation issues are hard; the doctrine is difficult and the issues are complex. Courts divide over whether to retain the common law's "textualist" rules of interpretation or to apply "contextualist" interpretive principles that are reflected in the Uniform Commercial Code (UCC) and the Second Restatement. (1) States holding the former view are led by New York, while California is the most significant contextualist jurisdiction. (2) Perhaps understandably, given the doctrinal complexities and jurisdictional divisions, much of the scholarly commentary is unhelpful. This is unfortunate because contract interpretation remains the largest single source of contract litigation between business firms.(3)

In 2003, we published an article that, among other things, set out a theory of contract interpretation to govern agreements between business firms. (4) The arguments we advanced were part of a larger project arguing that the law should pursue the first order goal of maximizing contractual surplus when it chooses rules to regulate merchant-to-merchant contracts. (5) The article has generated substantial academic commentary. Of greatest interest has been our treatment of interpretation issues. In that section, we supported a formalist theory of contract interpretation. Our article has become the iconic formalist statement. As such, it has engendered a number of antiformalist responses. (6)

Recently, Professor Steven Burton has published a book on contract interpretation advancing a theory he calls "objectivism" that purports to resolve the divisions between these competing claims. (7) In the course of developing his argument, Burton also discusses and rejects our approach. (8) Instead, he purports to chart a reasonable middle course between the Scylla of contextualist theories of interpretation and the Charybdis of formalist interpretive theories. But in order to create this space, Burton mischaracterizes both formalist and antiformalist arguments. The former, he claims, reduces to a literalism that urges the interpreter to focus only on a single word or phrase; the latter is a subjectivism that seeks to recover the parties' subjective intentions rather than the objective manifestations of those intentions. (9) As we show in the analysis that follows, Burton's intermediate position is untenable. He urges courts to consider course of dealing and usage of trade evidence in addition to the contract and the parties' objective circumstances, but he would have courts exclude extrinsic evidence (such as prior negotiations) because it is too "subjective." (10) Burton's solution thus precludes the resolution of a dispute on summary judgment while it denies to parties a full trial in which all evidence that might reveal their ex ante intentions is considered. In short, his "objectivism" sacrifices both cost and accuracy-two goals, we argue, that firms prefer the courts to consider when interpreting contracts.

Professor Burton's response to our article, as well as a number of other responses, demonstrate that the interpretation debate has become both livelier and more highly contested than ever. Unfortunately, the responses to us often mistake or misconstrue our claims. To bring some clarity to the debate, therefore, we begin by restating the four arguments our article advanced.

Our first argument begins with the premise that, although accurate judicial interpretations are desirable, accurate interpretations are costly for parties and courts to obtain. (11) If contract writing were free, parties could minimize interpretive error by exhaustively detailing their intentions. And if adjudication were costless, courts could minimize interpretive error by hearing all relevant and material evidence. Contract writing and litigation are costly, however. Since no interpretive theory can justify devoting infinite resources to achieving interpretive accuracy, any socially desirable interpretive rule would trade off accuracy against contract-writing and adjudication costs. Such a rule, we argue, tells courts in some cases to exclude relevant evidence.

The second argument follows from the fact that parties to commercial contracts have preferences over the rules courts use to interpret their agreements. These preferences exist because the parties are the primary beneficiaries of accuracy and the primary bearers of contract-writing and litigation costs. Our second argument, then, rests on three premises: (i) courts that interpret commercial contracts should share the parties' goal of maximizing expected contractual surplus; (ii) parties are better informed than courts about benefits and costs, so parties commonly have a comparative advantage over courts in making the requisite tradeoffs; and (iii) different contractual relationships may make different tradeoffs-that is, party preferences over interpretive rules are heterogeneous. Together, these premises imply that the state should defer to party preferences regarding interpretation, just as it defers today to party preferences over a contract's substantive terms.

We argue, therefore, that the state should choose interpretive rules that conform to majoritarian party preferences, and courts should obey party instructions to depart from those rules in particular cases. Much of the criticism we face stems from the failure to appreciate this claim. Thus, we do not argue that the state should enact mandatory rules that require courts to make formalist interpretations. Rather, we argue that the state should create interpretative rules that instantiate party preferences; it is the business parties that commonly prefer formalist interpretations.

Our third argument urges participants in the interpretative debate to distinguish two principal interpretive questions-whether a contract term is ambiguous; and whether the term is written in the standard language or in a private language (for example, technical terms in a trade). (12) This distinction is important to draw because the categories of evidence a court should admit when making an interpretation arguably should turn on whether the issue involves ambiguity or language.

Our fourth argument, however, holds that the majoritarian party preference is textualist in both cases: business parties commonly prefer judicial interpretations to be made on a limited evidentiary base, the most important element of which is the contract itself. (13) With regard to ambiguity, we claim that parties prefer a narrow evidentiary base when a court has enough information to reach the "correct answer" to an interpretation question on average. Put more technically, the correct answer is the mean of the distribution of possible interpretations. The claim for textualism then follows because a risk neutral firm is indifferent to the magnitude of the variance around the mean; hence, this firm is unwilling to incur additional costs in order to further increase the accuracy of any particular finding. With regard to language, we consider two defaults: a plain meaning default supposes the parties to be communicating in the standard language; and a contextual default admits extrinsic evidence that the parties were communicating in a "private" (technical) language. We argue that the preferred default should aim to reduce the risk of strategic behavior. A plain meaning default that presumes the parties have written in the standard language reduces this risk by requiring parties to specify the terms that take technical meanings. (14) Our argument that parties have textualist preferences over questions of meaning supports our further claim that parties also have preferences over how a court determines the terms of a contract; that is, parties prefer courts to use a "hard" parol evidence rule, one that restricts courts to a narrow evidentiary base when identifying the contract's terms.

This Review clarifies and extends the arguments sketched here. We briefly summarize empirical data that support our theory, and we respond to our critics. With respect to the data, although much academic commentary suggests otherwise, both the available evidence and prevailing judicial practice support the claim that sophisticated parties prefer textualist interpretation. (15) Indeed, the large majority of common law courts, led by New York, continue to follow the traditional Willistonian approach to interpretation, which embodies a hard parol evidence rule, retains the plain meaning rule, gives presumptively conclusive effect to merger clauses, (16) and, in general, permits the resolution of many interpretation disputes by summary judgment. (17)

In responding to our critics, we make three preliminary remarks. Initially, though our article was the first cut at a difficult subject, and so was less clear than it could have been, the mistakes in representing our view are hard to explain as resulting only from a lack of clarity on our part. "Formalism," to many, has come to represent almost every bad thing, (18) so we as formalists are believed to commit almost every sin. Thus, we often are taken to be literalists who want to restrict an interpreting court to consider only the contract terms at issue and a dictionary. (19) But literalism is impossible. Courts necessarily see the pleadings. When these are contested, a court will and should consider briefs and evidence as to what the parties did or failed to do, the damages the plaintiff claims, and so on. Rather, we argue that interpretive defaults acceptable to firms require courts, when making interpretations, to exclude particular categories of evidence, such as the parties' practice under prior contracts between them and their precontractual negotiations. (20) Moreover, we concede that a court is more likely to make an accurate interpretation if it sees more evidence, but we argue that sometimes accuracy is not worth the costs of achieving it. Our critics reject this view largely on the ground that contextual interpretations are more accurate than formalist ones. This response does not engage our arguments, which importantly rest on cost concerns. (21)

Second, critics claim that we advance a monist normative theory--economic welfare--to govern a world that is too complex for any single theory. (22) criticism is misguided and incomplete. Our theory applies to contracts between sophisticated firms. It is a questionable theory if the world of sophisticated firms that we analyze is too complex for the theory. But it is not a questionable theory if only the wider world is too complex. Further, this criticism confuses procedural with substantive norms. Professor Burton thus criticizes our theory as excessively monist because, he alleges, we neglect the rule of law virtues of security of transactions, nonarbitrary adjudications, and administrability. (23) Rule of law virtues are "procedural," in the sense that the virtues create a framework in which decisionmakers should implement the substantive norms the decisionmakers prefer. The virtues, however, do not identify the substantive norms that are to be implemented. For example, a procedural norm requiring courts to avoid arbitrary decisions does not identify the substantive norms that courts should apply in principled ways. If a theory is pluralist when it combines the procedural virtues of the rule of law with a single substantive norm, then we have a pluralist theory; for we actually argue that our theory better serves the procedural virtues Burton identifies than other theories do. (24)

Our theory is monist in a substantive sense. We argue that the contract law that regulates transactions between firms should seek only to maximize efficiency. This argument rests on two grounds. First, efficiency is the only implementable norm when firms prefer efficient rules and have the resources and sophistication to contract out of rules they dislike. (25) Second, scholars have yet to show how the efficiency norm can be combined with other substantive norms to yield predictable-that is, nonarbitrary-adjudications. In the absence of such a showing, substantively pluralist theories necessarily serve the rule of law virtues less well than substantively monist theories do. These two grounds cannot be dismissed on the mere assertion that the world is too complex for a single theory.

Finally, critics claim that we offer a law and economics account that may be appropriate for law and economics scholars but which need not be considered by courts or scholars of other persuasions. (26) This is an evasion, not an argument. We claim that interpretive rules should reflect the preferences of business firms that contract with each other. We justify this claim largely on welfare grounds and use economic analysis to identify the preferences that business firms typically hold. Such a law and economics account is either persuasive or it is not. Hence, it is open to critics to (a) explain why interpretation rules for firms should not be based on party preferences; or (b) show that we have misapplied economic theory; or (c) show that an alternative methodology would better recover what business parties prefer. (27)

The Review proceeds as follows. In Part I we further clarify our argument for textualist interpretation. Part II then applies that argument to the claims of our critics. In Part III, we show how the law can, and often does, implement a textualist interpretive theory. We briefly conclude.


A. The Timing and Nature of Interpretation Issues Interpretation issues can be illuminated by first considering the timing of a simple game:

[t.sup.0]: The state chooses the rules that govern contractual interpretation.

[t.sup.1]: The parties write a contract to trade a performance for a price. The contract describes the performance that is to be traded (for example, manufacture a machine or construct a building); sets out the parties' procedures (that is, when the performance is to be tendered; when the buyer is to pay); and sets a price and mode of payment. The contract may also set out the interpretive rules the parties prefer a court to use if (a) the state's rules are defaults; and (b) the parties prefer their own rules to the state's rules.

[t.sup.2]: The parties learn the state of the world.

[t.sup.3]: (a) The seller decides whether or not to perform; (b) if the seller chooses performance, the buyer decides whether or not to pay.

[t.sup.4]: If the seller decides not to perform when the buyer prefers to trade the contract price for the performance, the buyer sues; if the buyer refuses to pay when the seller tenders performance, the seller sues.

[t.sup.5]: A court resolves disputes.

These disputes may raise interpretive issues. For example, the seller may reject the proposed trade on the ground that the deal was for cash but the buyer instead insisted on credit. Alternatively, the buyer may refuse to pay on the ground that the seller failed to tender the quality of performance that the contract required. To resolve these disputes, a court must either apply the interpretive rules that the state promulgated at [t.sup.0] or, if the state's rules are defaults, apply any interpretive rules the parties substituted at [t.sup.1].

The interpretive rules that the court applies at [t.sup.5] largely regulate the admissibility of evidence. Assume, for example, that a plaintiff buyer proffers the following evidence in support of his claim that the seller delivered goods of the wrong quality: (a) the product the seller delivered; (b) the written contract that contains the specifications for the product; (c) two precontractual memoranda the parties exchanged that referred to the contract specifications; and (d) an oral narrative of party conversations prior to signing the contract that concerned the contractual quality standard. An interpretive rule selects the evidence that is admissible to resolve the interpretive question. A textual rule admits evidence bits (a) and (b) but excludes bits (c) and (d). A contextual rule admits (a) through (c) and sometimes (d).

B. The Problems the State Must Solve

The timing sequence set out above focuses attention on the three questions that the state must answer at [t.sup.0] so that the court can resolve an interpretation contest at [t.sup.5]. The relevant "state" consists of three separate legal sources: the UCC, the Restatement (Second) of Contracts, and the decisions of common law courts. The questions are: First, what goals should the state's interpretive rules attempt to implement? Second, what rules best implement these goals? Third, should the rules be mandatory or defaults ?

Regarding the first question, a state can have both first and second order preferences over goals. (28) We focus on first order preferences in this Review. The first order interpretive preference of each of the three sources of law is "goal neutral": it is uniformly held that a court resolving an interpretive dispute should recover the parties' intentions, whatever those intentions were. Intention, in turn, is determined prospectively and objectively; that is, the question for the court is what manifestation of intent a counterparty could reasonably understand the other was assenting to when it contracted. (29) Goal neutrality is justified by the freedom of contract norm. If parties can create their own contract, the interpreter's role is restricted to finding out what that contract was. The interpretive rules, therefore, should facilitate this quest. (30)

The second question the state must resolve at [t.sup.0] is that of determining which legal rules best implement the goal of recovering the parties' intentions. Here the consensus of the three sources of controlling law breaks down. On one view, a goal neutral preference to recover the parties' intentions is best implemented on a broad evidentiary base. The interpretive rules therefore should be contextualist. On the other prevalent view, courts commonly can recover the parties' intentions on a narrower evidentiary base. The interpretive rules thus should be textualist. The (almost) scholarly consensus shares the UCC and Restatement view that the appropriate first order interpretive preference should be goal neutral (that is, the court should recover the parties' intentions), and that this preference is best implemented by permitting the court to access a broad evidentiary base in determining both the terms of the contract and the meaning to be attached to those terms. (31)

To understand the litigation implications of these competing views, consider the key differences between hard and soft versions of the rules governing, respectively, parol evidence and the meaning of terms. The parol evidence rule functions to determine the terms of the contract. A hard or textualist parol evidence rule thus narrows the evidentiary base the court considers when finding the contract's terms; a soft or contextualist rule expands the base. (32) A separate question asks: what do the contested terms mean? This issue arises when a term in the written agreement has a plain, unambiguous meaning in the standard language, but one party claims that the parties attributed a different meaning to the term. (33) Courts that follow a hard parol evidence rule also tend to follow the plain meaning rule, and thus exclude extrinsic evidence to resolve such disputes. Courts that follow a soft parol evidence rule also tend to reject the view that words can have invariant meanings; hence, these courts permit a party to introduce extrinsic evidence of the particular--that is, nonstandard--meaning that the parties attached to the contract language.

The final question that the state must resolve is whether its interpretive rules should be mandatory or defaults that parties are free to alter. Just about everyone who creates, applies, or analyzes the interpretive rules believes that they should be mandatory. (34) And, in fact, the rules are mandatory in the sense that parties cannot contract directly for textualist or contextualist interpretive rules. Textualist and contextualist courts differ, however, in the degree to which they permit parties indirectly to control how disputes over their contract are adjudicated. For example, textualist courts apply a hard parol evidence rule, but this version of the rule permits parties to "contract out" either by not integrating their writing fully or by including context evidence in the integrated contract. In contrast, contextualist courts apply an invariant soft parol evidence rule, so that parties cannot narrow the evidentiary base. In contextualist jurisdictions, therefore, contextualism is the only option. (35)

C. An Introduction to Our Theory of Interpretation

1. The Case for Party Control

We introduce our theory here, and extend it as required in the course of this Review. (36) We now assume for purposes of discussion that interpretations should be goal neutral; the state should seek to recover the parties' objective ex ante intentions. Before discussing the interpretive rules that this goal implies, we note two related questions-ambiguity and language-that the rules must answer. Some terms are ambiguous because they can take more than one objective meaning. For example, if a seller agrees to sell "one Ming vase" to a buyer, but the seller owns four Ming vases, the term "one Ming vase" is ambiguous because it could refer to any of the four vases; the phrase does not identify the vase that the parties intend to trade. Other terms are ambiguous in a different sense: because they are vague, they cannot precisely confine the parties' meaning. Consider a contract to sell "hard, dark red Macintosh apples." The terms "hard" and "dark red" are vague because they do not precisely determine what constitutes a conforming apple: how hard and how dark red must the apples be? In both instances of ambiguity, the interpreting court may be unable to recover the parties' intentions on a narrow evidentiary base. (37)

As an example of the language question, consider an acquisition agreement that excuses the buyer if the target seller suffers a "material adverse change" that substantially reduces its value. A sudden shortage of parts disrupts the seller's production. As a consequence, the seller's market capitalization drops substantially. An ordinary observer likely would think that the parts shortage constituted a material adverse change. The seller, however, argues that, in the market for corporate control, a material adverse change is understood to denote a reduction in the seller's value that the seller could have prevented, and it could not have prevented the parts shortage. (38) Did the parties want the phrase "material adverse change" interpreted as ordinary speakers of English would understand it? Or did the parties want the phrase to be interpreted as participants in the market for corporate control would understand it?

We argue, for various reasons, that the interpretive rules that regulate both types of meaning questions should be defaults. (39) This argument has two implications. First, the rules should reflect majoritarian party preferences. Second, courts should obey party interpretive instructions that alter the defaults. As an example of the latter implication, parties should be free to choose a hard rule to govern parol evidence questions if the default is soft, and the converse if the rule is hard. In contrast, many courts and commentators assume (at least implicitly) that a goal neutral interpretive regime requires mandatory interpretive rules. (40) This assumption is rarely defended. Our article contested this assumption by arguing that (a) the rules should reflect the parties' preferences; and (b) parties prefer textualist interpretive defaults.

As we noted above, critics primarily attack our claim for textualist rules but essentially ignore the logically prior claim that the rules should be defaults that reflect party preferences. This is unfortunate because the argument for interpretive defaults is the more important claim. Our critics defend contextualist rules primarily on the ground that contextualism yields more accurate interpretations than textualism. This defense is not responsive. If the interpretive rules that a goal neutral interpretive theory implies should be defaults, and if good defaults reflect party preferences, the search cannot be for the most accurate rules, simpliciter. Rather, the search must locate the most accurate rules that parties are willing to pay for. One can refute our view, therefore, only by showing that a goal neutral interpretive theory better supports the case for mandatory rules than the case for defaults; or by showing that parties would reject the defaults that we identify.

We begin here by reviewing and extending the case for interpretive defaults. (41) Note first that no interpretive theory can entirely ignore costs. A multi-year trial to resolve a dispute over the meaning of a contract is unacceptable to all. Three cost categories are relevant to interpretation: (a) the court's opportunity cost-the more time the court spends on a particular interpretive issue, the less time it can spend on other issues or other cases; (b) the parties' contract-writing cost, which partly is a function of the interpretive rules; and (c) the parties' enforcement cost. We focus on the latter two cost categories for two reasons. First, the state does not appear to take court costs into account when it chooses interpretive rules. Second, as we argue below, parties rarely prefer more evidence than courts prefer, and usually prefer less. Hence, the danger that parties will overwhelm courts with interpretive tasks is slight.

There would be no need for defaults or interpretive instructions (from parties to courts) if two premises were satisfied: first, the state's goal is to maximize the parties' expected profits; and, second, the state is as capable of doing so as the parties. Then the state could adopt optimal interpretive rules. These rules would direct the court to broaden the evidentiary base when this would maximize the parties' expected utility and to narrow the base when that would be best. (42) Parties would anticipate that courts would interpret disputes according to these rules and thus would not issue interpretive instructions (there is no need to instruct a perfect agent). The case for defaults (and instructions) thus begins with the recognition that neither of these premises holds.

Assume, initially, that the state does share the parties' surplus-maximizing goal. (43) Even so, the second premise cannot be satisfied: the state is less capable than parties at identifying optimal interpretive rules because the state is less well informed than parties about contract-writing and enforcement costs and the benefits that correct interpretations produce. Thus, even a state that shares the parties' goal will sometimes mistake the optimal tradeoff between accuracy and cost. The interpretive rules thus should tell the court to follow party instructions regarding the evidentiary base when the contract contains instructions. The rules, that is, must be defaults. (44) But does this conclusion change when we consider, as noted above, that the state is goal neutral and thus does not share the parties' surplus-maximizing goal? In other words, can interpretive rules that require greater accuracy in interpretation than the parties prefer be justified under a goal neutral interpretive theory? The apparent answer is no. Goal neutrality gives the parties control over the substantive terms of the contract. It takes an argument to reject the obvious implication that the parties should also have control over the rules that determine how those terms are identified and understood.

A number of courts that pursue goal neutrality actually do put more weight on accuracy in recovering the parties' intended purposes than do typical parties. This judicial preference for accuracy largely rests on libertarian grounds. Some courts are reluctant to impose the state's coercive power on a reluctant party unless the court is relatively certain that the party failed to do what the contract required. Courts implement this libertarian value without serious constraint because judges do not internalize the costs that implementing this value causes: the judges do not observe the parties' contract-writing costs and only imperfectly observe their litigation costs. In addition, judges do not bear the consequences of rules that require parties to bear costs the parties would prefer to avoid.

This observed preference of some courts for more accuracy than parties prefer is not sustainable under the norm of goal neutrality. A court's reluctance to coerce parties is misplaced if parties are not reluctant to impose coercion on themselves. Sophisticated parties now can waive the right to a jury trial, or even the right to a trial in court, so they seemingly also should be able to waive the protection of exhaustive interpretative hearings. And, while it is true that courts do not internalize the costs of contract writing and evidence production, that is not a reason for the court to override party preferences. Because parties are more sensitive to costs than courts, parties rarely prefer courts to admit more evidence than courts want to hear, and usually prefer courts to admit less. (45)

We can now summarize our claim that interpretation rules should be defaults. A major task of any interpretive theory is to recover the parties' contracting intentions. No defensible theory supports requiring courts and parties to devote infinite resources to this task. Therefore, any theory "on the ground" must trade off the gains from increased accuracy in interpretation against the costs. Contracting parties have a comparative advantage over the state in choosing the interpretive rules that best make this tradeoff because the parties are better informed than the state about the relevant costs and benefits. It follows that courts should yield to the parties' preferences over the rules unless the state is pursuing an interpretive theory that is animated by a goal that requires greater accuracy than parties prefer. We have just argued that the case for greater accuracy cannot rest on the efficiency norm or on goal neutrality. So far as can be known, the state is not pursuing a goal other than these. It follows that there is no good reason to deny parties as much control over the interpretive terms of a contract as parties now have over the substantive terms.

2. Parties Prefer Textualist Defaults

We summarize here a complex argument with a simple example. (46) Assume that an agreement requires a contractor to construct a particular structure. The parties later do not contest what the contractor actually did--how many rooms the completed structure had, the thickness of the walls, and so forth. The issue is whether the contractor created the structure that the contract required.

We denote a conforming structure with the letter [S.sup.*]: this structure has the attributes the parties intended. The buyer's payoff, if the seller complied, is represented as the value of [S.sup.*]; that is, v(S*). Assume now that the buyer claims, correctly, that the completed structure has fewer attributes--is less desirable--than [S.sup.*] would have been. We denote such a nonconforming structure as [S.sup.-], so that the structure yields the buyer a value of v([S.sup.-]). The seller concedes that she erected the structure [S.sup.-], but argues that this performance was all that the contract required. In the seller's view, S- actually is the [S.sup.*] the contract described. If the court accepts the seller's erroneous interpretation, the buyer must pay the price and so he realizes less value (v([S.sup.*]) - v([S.sup.-])) than he contracted for. Alternatively, suppose that the seller actually complied by delivering the structure [S.sup.*], but the buyer claims that the contract called for an even better structure. We denote this structure as [S.sup.*], which, had it been delivered, would have yielded the buyer a value of v([S.sup.+]). If the court accepts the buyer's erroneous interpretation, the buyer can reduce the price by his claimed "loss": v( [S.sup.+]) - v([S.sup.*]).

We motivate this example with two assumptions. First, we assume that parties contract in such a way that a court will make the correct interpretation more often than the court will err when the court sees the contract, the seller's performance, and the pleadings. Correct interpretations are more likely than not for two reasons. Parties prefer correct interpretations to incorrect ones. Further, parties know the interpretive rules when they contract; hence, parties know what evidence a court will later see. A contract's character thus is partly a function of the legal rules. For example, if the rules restrict a court to a narrow evidentiary base, parties likely will write a more detailed contract than if the rules permit the court to see everything.

Second, we assume the court is unbiased. Judicial errors are as likely to be on the high side (the contract requires the seller to produce the S+ structure) as they are to be on the low side (the contract requires the seller to produce the Sstructure). The court is unbiased because, in commercial cases, courts are unlikely to have a systematic preference for particular litigant types. Rather, courts are goal neutral, wanting only to recover the parties' intentions.

These assumptions imply that parties expect the court's interpretation to be correct (the contract requires the seller to erect the [S.sup.*] structure), but they also know that interpretive errors are possible. (47) We add, for clarity, that the probability of a correct interpretation likely increases with the width of the evidentiary base. In the illustration above, then, [S.sup.-] and [S.sup.+] are likely to be closer to [S.sup.*] for a contextualist court than for a textualist court. Using the technical term, more evidence shrinks the variance around the true interpretation. An infinite amount of evidence reduces the variance to zero.

This analysis should clarify the accuracy/cost tradeoff we described above. When parties contract they know that they can affect the probability of a correct interpretation through the contract they write and the evidence they later can introduce. We claim that business firms put major weight on controlling litigation costs. It is less costly to write a contract than to try a case. Firms also are assumed to be risk neutral. A risk neutral firm wants the mean of a distribution of future values to be "true," but the firm is less concerned with variance around the true mean. Thus, we argue that firms write contracts that are sufficiently detailed to ensure that the most likely interpretations of their contracts are correct when courts interpret those contracts using the minimum evidentiary base we specify. In the illustration here, then, the court requires the seller to deliver the [S.sup.*] structure with a greater probability than either of the other two possible structures. There remains a real possibility of error, however, because the parties want to avoid a trial whose result would be materially to shrink this possibility. Trials are too costly.

To be sure, this argument has limits. When a correct interpretation is particularly important-say the interpretation of a new contract that is expected to be widely used--a party may prefer a court to hear all the available evidence. Similarly, parties dislike infinite variance and will spend some money to keep interpretations in the (sometimes wide) range of the plausible. Thus, we argue for defaults under which the court makes interpretations on a minimum evidentiary base. Parties who want more evidence than this can contract out. In sum, we claim that interpretive rules should be defaults, and that the defaults should be textualist. We next defend this second claim against the most important objections.


In Part I we discussed what may be called abstract or general criticisms of our theory. This Part discusses four concrete objections: (1) firms are not risk neutral; (2) unbiased interpretations can encourage strategic behavior; (3) the evidentiary base we claim that parties prefer is too narrow to sustain accurate interpretations, so parties free to choose would reject the rules we propose; and (4) there is no evidence that actual parties possess the preferences we attribute to them. We discuss each objection in turn.

A. Risk Neutrality

Antiformalist critics contest our claim that firms are risk neutral.(48) This criticism is misplaced because these critics do not contest our claim that firms maximize expected profits. Profit maximization implies risk neutrality. (49) A risk neutral firm acts as if it is risk averse only when it is willing to incur costs to reduce variance in the space of contractual interpretations. We noted above, and explained more fully in our paper, that firms act in this way when a correct interpretation is particularly important to them. (50) Few business contracts have this "bet the ranch" character, however. Thus, we make risk neutrality the premise when analyzing the preferences of firms.

B. Strategic Behavior

Critics fault our interpretive theory for encouraging parties to engage in strategic behavior, either by acting opportunistically (51) or by "chiseling" on their obligations. (52) An unbiased court nevertheless may issue an interpretation that is wide of the mark, and the possibility of large errors, the claim goes, creates an incentive for parties to cheat. Professor Bowers concisely summarizes this claim: "If a party thinks there is a good chance that its chiseling behavior will be protected by an erroneous contract interpretation, the more chiseling will tend to occur." (53)

This is a significant objection. As we have said, parties write contracts to give themselves incentives to take maximizing actions. If parties behave strategically when courts are only right on average, the incentive function of a contract may be vitiated; that is, a party may take the chiseling rather than the maximizing path. If this danger is real, parties would prefer contextualist interpretations, which shrink variance and thus reduce the strategic behavior risk.

The objection is unpersuasive, however, because the incentive of a party to behave strategically is small when courts are right on average. To see why, we use the example that involved contracting for a particular structure. Consider the seller's incentives after the contract is signed. The seller knows that if there is a dispute and if interpretation is textualist, the court will later see the contract, the pleadings, and the performance the seller rendered. The court will evaluate this data using its experience. (54) For the reasons given above, we assume that the parties' contract is written such that the court is likely to make the correct interpretation more than half the time. On this view, "probability mass" encourages compliant behavior. The seller knows that the court will make a correct interpretation with probability greater than one half, and that it will incorrectly find that the contract required the seller to tender a performance of lower quality than the contract required with a lesser probability. In both cases, a seller who complies--who supplies the [S.sup.*] structure--will receive her expectation (the court believes that the seller either complied or overcomplied). The seller will erroneously have to pay damages only if the court finds that the contract required a greater quality level than the compliant quality level the seller supplied. Given symmetry--the court is as likely to err on the high side as the low side--the court's high and low side errors cancel out. Hence, the compliant seller receives her expectation with probability that exceeds one half. (55)

However, the seller may have an incentive not to comply. Her expectation is price less cost, and the cost of a cheating performance is less than the cost of a compliant performance. But now the probabilities disadvantage the seller. She will have to pay damages to the buyer with a probability that is the sum of the probability that the court renders a correct interpretation and the probability that the court makes a high side error (it finds that the contract requires a performance that exceeds the true contract quality). The seller profits from cheating only if the court finds that the contract required the noncompliant performance that the seller actually tendered. This probability is relatively small, however. Hence, the seller will not cheat unless the gain to cheating is very large (cheating can be done at very low cost) or the gain to compliance is very small (a compliant performance generates little surplus for the parties to share). In the common case, cheating will not pay. (56)

To make this argument more vivid, consider a numerical example. Suppose the buyer values a conforming structure at v([S.sup.*]) = 120 and this structure costs 80 to build. The structure is worth more to the buyer than to anyone else. The contractual surplus is 40 (value less cost). Thus, making the standard equal bargaining power assumption, the price is 100 (the 80 cost pius one half of 40). (57) The court in this example is just barely more accurate than not. The probability that it will correctly construe the contract to require the seller to deliver a structure of quality [S.sup.*] is .55. With probability .225, the court will interpret the contract to require the seller to deliver the lesser quality [S.sup.-]; and with probability .225, the court will require the greater quality S+. In this illustration, the seller tenders a structure whose quality the court can observe. The interpretive issue is whether the contract required the seller to deliver that quality or some other quality. If the structure is found not to conform to the contract, the seller expects to sell it on the market at cost. (58)

Assume initially that the seller produces the conforming structure [S.sup.*]. With probability .225, the court finds that the seller overcomplied--only [S.sup.-] was required--but she cannot get more than the price. Hence, she earns price less cost, which is 20. With probability .55, the court correctly finds that the seller complied, and again she gets the price and earns 20. With probability .225, the court erroneously finds that the seller breached: she should have delivered [S.sup.+]. In this case, the seller is liable for expectation damages of v([S.sup.*]) - price = 120 - 100 = 20. The seller's expected payoff from complying with the contract thus is .225(20) + .55(20) + .225(-20) = 11.

Now assume the seller cheats by actually producing the [S.sup.-] structure at the lower cost of 40, one half of the compliance cost. With probability .225, the court finds that the contract required the seller to produce only an S- structure, so the buyer must pay the full price for a degraded performance. The seller then earns 60--the 100 price less her 40 cost. With probability .55, the court finds that the contract required the seller to deliver an [S.sup.*] structure, so the seller is in breach. The seller will then sell the structure at her cost of 40 and pay damages of 20 to the buyer. With probability .225, the court erroneously finds that the contract required the seller to produce the [S.sup.+] structure, so she again is in breach and must pay damages of 20. The seller's expected return if she cheats thus is .225(60) + .55(-20) + .225(-20) = -2. The seller thus will not cheat when she expects to face a barely accurate court.

We can summarize the intuition that supports the example. A seller who cheats earns a positive return when the court construes the contract to require only the performance that she tendered. However, the court is more likely to find that the contract required a conforming performance. Moreover, a court may also find that the contract required a greater performance than the contract actually requires. The cheating seller benefits from errors on the low side but loses both from correct interpretations and from errors on the high side. Thus, the seller commonly realizes a higher expected payoff when she attempts to comply with the contract than when she cheats. In Section II.D. below, we review the evidence that shows parties prefer textualist interpretations. The analysis here shows that these interpretations are good enough to support efficient contracting. (59)

C. Accuracy, Context, and Party Preferences

Almost all of our critics erroneously claim that we urge courts to interpret contracts on narrow evidentiary bases. (60) On the contrary, we urge courts to follow party preferences regarding the interpretive rules. It is the parties, we claim, who commonly prefer textualist interpretations.

These critics also argue that parties prefer contextual interpretations because they are more accurate. As we point out above, this response does not engage our claim because it assumes that judicial accuracy comes at zero cost. We engage the accuracy issue more deeply here. The ground of the objection is summarized in the claim that words are meaningless unless they are read in context. We are taken to prefer courts to interpret contracts without reference to context.

To the contrary, we agree that context is crucial to interpretation, though not exactly for the reasons our critics espouse. Rather, context is necessary for finding whether a performance conforms to the contract. Recall again the illustration regarding the construction of a structure of specified dimensions. Assume that the buyer is a chemical company that plans to use the structure to store dangerous chemicals. A structure that complies with federal regulations regarding chemical storage would be conforming while a structure that resembles a gazebo would clearly be deficient. Hence, a court cannot interpret a contract that requires the seller to deliver "a structure" (even one whose dimensions are specified) without knowing the context--the buyer's business and his plan.

A minimum evidentiary basis ordinarily will convey sufficient contextual information. For example, the pleadings and supporting briefs, evidence as to what the seller delivered, the contract, a recent 10-K SEC filing by the buyer (if any), (61) and the life experience of the judge, together are likely to permit a court correctly to decide that an agreement between a professional construction company and a chemical company is unlikely to be satisfied by the delivery of a gazebo.

This view is made more plausible when one remembers that the court sees what the parties want it to see. (62) Recall the timing of the interpretation game described above in Section I.A. The state moves first by choosing the rules. The parties then move by creating the contract. (63) At this stage, the parties know their context, so they can decide whether to include context evidence in the contract. The court next moves by resolving any interpretive dispute. The parties thus can anticipate, when they create the contract, what the court will later see. Suppose then that the court is expected to use a minimum evidentiary base. The parties' best response may be to make the contract more illuminating. For example, the "whereas clause" in the contract to deliver a structure can describe the buyer's business and his contemplated use of the structure, and appendices to the contract can contain the specifications that the structure is to meet and any memoranda the parties want an interpreting court to use. (64)

An argument that parties prefer courts to use broad evidentiary bases thus loses force when it is recognized that parties have substantial control over the width of the base. To be sure, ex post, one party may want the court to see particular bits of evidence that the other wants excluded. The issue, however, is whether, ex ante, the party wanted the opportunity to introduce such evidence. We argue that business firms commonly reject that opportunity because exercising the opportunity at trial is costly and they expect to get satisfactory interpretations without it.

There is a need for such interpretations because contracts seldom conclusively settle things. Professor Bayern apparently believes, to the contrary, that parties do settle everything that they consider. In his words, it is "wildly unlikely" for parties to consider a subject to the point of writing down a solution, but not have the solution resolve the issue. (65) This reasoning leads him to claim that we erroneously distinguish gap filling from interpretation. If the written words settle everything, any dispute must involve an issue that the writing does not treat. Hence, the interpretation function collapses into the gap filling function.

This claim is incorrect. To see why, consider again the contract to deliver a structure of specified dimensions. It may be that the chemical company actually wanted a gazebo to decorate the lawn in front of its office. Parties assist courts by communicating context information in their contracts. Thus, the whereas clause in the contractor/chemical company contract could recite either that the parties want a structure for decoration or that they want a structure for storage. To see the problem with Bayern's claim, imagine that each piece of information that describes a part of the full context is a bit. The contract may require a number of bits to describe the subject of sale: the size of the structure is a bit, its weather resistant properties another, and so forth. The industry in which the parties function would require additional bits to describe.

Writing context bits is costly. Each additional bit, however, increases the probability that the court will make the correct interpretation, and so each bit increases the parties' expected payoff under the contract. Now make two standard assumptions. First, the expected gain from including bits is concave; that is, the first few bits greatly increase the probability that the court will make the correct interpretation, while the last few bits only slightly increase that probability. Second, contract-writing costs are convex: the contract becomes more costly to write as it adds contextual bits, and contract-writing costs increase at an increasing rate. On these assumptions, there is commonly an "internal solution" to the parties' contract-writing problem: parties will include contextual bits until the marginal gain--the increased expected contractual payoff--from further bits equals the marginal cost of writing them. In the usual case, the internal solution yields a contract that contains fewer than all the bits required to describe context perfectly. (66) Put more simply, it is actually "wildly unlikely" for the optimal solution to the typical parties' contract-writing problem to require the parties to state every conceivably relevant thing. When things are left out, however, the contract may require interpretation. Therefore, the interpretation function both is significant and exists independently of the gap filling function. (67)

D. The Evidence

Our earlier analysis, and the reexamination here, are theoretical. The sketchy evidence that exists is consistent with our account, however, and offers no support to our critics. To begin, parties explicitly attempt to alter common judicial interpretive practices. A common provision found in alliance agreements thus recites:
      The Parties' legal obligations under this Alliance Agreement
   are to be determined from the precise and literal language of this
   Alliance Agreement and not from the imposition of state laws
   attempting to impose additional duties of good faith, fair dealing
   or fiduciary obligations that were not the express basis of the
   bargain at the time this Agreement was made.

      The Parties are sophisticated business entities with legal counsel
   that have been retained to review the terms of this Alliance
   Agreement and the Parties represent that they have fully read this
   Alliance Agreement, and understand and accept its terms. (68)

In addition, parties often direct courts to ignore prior negotiations, oral understandings, and course of conduct and instead to limit their consideration to the "appropriate documents in final form ... executed and delivered by each of the parties." (69) In the same vein, parties routinely write merger or integration clauses that attempt, though sometimes unsuccessfully, to restrict the evidence courts admit when interpreting their contracts. (70) The data go beyond anecdotal evidence from individual contracts. Lisa Bernstein's work shows that parties who are members of trade associations--and who thus rely on both informal or relational enforcement as well as third party enforcement--preserve formal interpretive rules and reject contextualized standards. (71)

Our analysis also generates an empirical prediction. As noted above, New York courts are textualist when using the common law while California courts are contextualist. Thus, we predict that parties will contract into New York law when they can, but will not contract into California law. Recent work by Ted Eisenberg and GeoffMiller studying choice of law and choice of forum clauses in a data set of 2,865 contracts is consistent with this prediction. (72) Their study shows that parties chose New York law in forty-six percent of the contracts and New York as the forum state in forty-one percent of the contracts. (73) California was chosen for its contract law in less than eight percent of the contracts, though its commercial activity, as measured by the place of business of the contracting parties, was second only to New York.

Geoff Miller uses the differences in contract law between New York and California to explain this large difference in party preferences. (74) His analysis confirms the conventional wisdom: New York strictly enforces bargains, retains a hard parol evidence rule and a plain meaning rule, and frequently declines to consider context evidence in resolving interpretive disputes. California, in contrast, has a contextualist interpretive regime and is predisposed to consider all material context evidence that litigating parties seek to introduce. Miller concludes that "[t]he revealed preferences of sophisticated parties support arguments by Schwartz, Scott and others that formalistic rules offer superior value for the interpretation and enforcement of commercial contracts." (75)

We claim that interpretive rules should be defaults in a goal neutral interpretive regime. This claim implies that the rules should reflect the preferences of the parties who must live under them. Their preferences, we argue, most commonly direct courts to interpret contracts on narrow evidentiary bases. This argument is supported by theory and is consistent with the evidence.


The interpretation debate has been theoretical, but interpretation issues arise in litigation contexts. These contexts require courts to choose between two arguments about what the contract requires. We sketch out the doctrinal framework that controls how courts make this choice. Our object is to make salient an underappreciated virtue of formalism: it permits parties to design contracts so as optimally to trade off contract-writing costs and expected litigation costs. (77) Situating the interpretation debate in its institutional environment thus further clarifies the formalist case.

Interpretive disputes require courts to identify the contract's terms and then to interpret those terms. The answers to identification and interpretation questions do not turn on the parties' subjective understandings but rather on the objective manifestations of their intent at the time of contract. (78) We next show how the doctrine, ideally applied and in connection with the contract, helps parties to control the evidentiary base on which courts will interpret their agreements.

A. Recovering the Contract Terms: The Parol Evidence Issue

The parol evidence rule functions to permit parties to control the admissibility of certain kinds of evidence. The rule holds that when parties fully integrate a final written agreement, they cannot later prove understandings that the integrated writing did not contain. Parties may want to exclude evidence of possible understandings reached during the contracting process for three reasons: (a) tentative agreements may later be abandoned; (b) agents may misrepresent their principal's commitments during the course of negotiations while the final agreement likely will set things straight; or (c) parties may prefer to exclude a portion of their agreement from legal enforcement. The common law consistently defers to these reasons, treating the decision to integrate an agreement as in the parties' discretion. (79)

American courts, however, divide over the appropriate test for determining whether or not parties intended to integrate part or all of their agreement into an exclusive, legally enforceable writing. (80) Formalist courts, such as common law courts in New York, use a hard parol evidence rule. (81) Derived from the traditional rule at common law, a hard parol evidence rule gives presumptively conclusive effect to merger clauses, (82) and, in the absence of such a clause, determines whether the written agreement is fully integrated by applying the four corners presumption. (83) This presumption holds that the contract is fully integrated if it appears final and complete on its face. A hard parol evidence rule thus maximizes party discretion over the content of the legally enforceable contract: it functions as a procedural default rule that shrinks the evidentiary base on which the court finds the terms of the contract, but the rule permits parties to expand that base as they see fit. When parties fully integrate the agreement and use a merger clause, an interpretation dispute over contract terms may be resolved on summary judgment.

Antiformalist jurisdictions, such as California, favor a soft parol evidence rule. (84) The test for integration in these states admits extrinsic evidence of meaning, even if the writing has an unambiguous merger clause or would appear final and complete on its face under the four corners presumption. (85) Antiformalist courts believe that a merger clause raises a rebuttable presumption of integration that may be overridden by extrinsic evidence that the parties lacked the intent to integrate. (86) When a merger clause is absent, soft rule courts reject the four corners presumption in favor of admitting any extrinsic evidence that may aid in determining the parties' intent to integrate their writing. (87) A court that uses a soft parol evidence rule is likely to deny a motion for summary judgment, thus increasing the expected litigation costs of both parties. Moreover, a soft parol evidence rule functions, in effect if not formally, as an open gate: any extrinsic evidence that a party sees as advantageous ex post will likely be considered. Thus, contrary to the stated purposes of the rule, the soft version withdraws from parties the ability to determine the evidentiary base that the court will later see.

B. Resolving Disputes over the Meaning of Contract Terms

The parol evidence question is logically prior to the question of what legal "meaning" should be attributed to the contract terms. Contrary to standard antiformalist claims, courts do not convert meaning inquiries into ventures in the philosophy of language. (88) This is not to say that context is irrelevant--any inquiry into the intended meaning of words is necessarily contextual--but rather to make clear that the parties can importantly determine the context within which any meaning inquiry is conducted. When parties write a fully integrated contract, they are attempting to embed much of the context for determining meaning within the written agreement itself. A variety of contract clauses perform this function. These include (a) "whereas" or "purpose" clauses that describe the parties' business plan and the transaction; (89) (b) definition clauses that ascribe particular meanings to words and terms that may vary from their plain meaning; (90) and (c) appendices that provide more precise specifications governing performance as well as any memoranda the parties want an interpreting court to use. (91)

Contests over the meaning of contract terms thus follow a predictable pattern: one party claims that the words in a disputed term should be given their standard dictionary meaning, as read in light of the contract as a whole, the pleadings, and so forth. The counterparty argues either that the contract term in question is ambiguous and extrinsic evidence will resolve the ambiguity, or that extrinsic evidence will show that the parties intended the words to be given a specialized or idiosyncratic meaning that varies from the meaning in the standard language. As with the division over hard and soft parol evidence rules, courts have divided on the question whether express contract terms should be given a "contextual" or a "plain meaning" interpretation. (92) Under the latter practice, when words or phrases appear to be unambiguous, extrinsic evidence of a possible contrary meaning is inadmissible.

The plain meaning rule operates in tandem with a hard parol evidence rule to reduce expected adjudication costs. If the contract is fully integrated, and if contractual terms are facially clear, then the dispute can be resolved at summary judgment. Parties may realize this cost advantage whether or not contextual inquiries into meaning exacerbate the risk of strategic behavior in litigation. (93) Parties who want the court to see additional evidence, but avoid trials, can (and must) embed the evidence in the contract itself. In this way, the plain meaning rule functions as a useful aid to contract design by offering parties the opportunity either to expand or to contract the context on which meaning questions are decided. A contextual meaning rule, on the other hand, limits parties' freedom to narrow the interpretive context even when doing so would maximize the expected value of the contract. (94)

The preceding discussion illustrates why it is important for parties to control by contract the rules that determine the evidentiary base for resolving interpretive disputes. Sophisticated commercial parties incur costs to cast obligations expressly in written and unconditional forms to permit a party to stand on its rights under the written contract, to improve party incentives to invest in the deal, and to reduce litigation costs. Parties know better than courts how best to trade off these front-end and back-end contracting costs. Contextualist courts and commentators prefer to withdraw from parties the ability to use these instruments for contract design. The contextualists, however, cannot justify rules that so significantly restrict contractual freedom in the name of contractual freedom.


Interpretative rules follow from interpretive theories. The regnant theory is goal neutrality: the interpreter's task is to recover the parties' objectively manifested intentions, whatever those intentions may have been. We argued in Contract Theory, (95) and argue again here, that the current interpretive rules cannot be derived from the current interpretive theory. The rules of the Restatement, the UCC, and many jurisdictions are mandatory and require courts to use broad evidentiary bases when interpreting merchant-to-merchant contracts. Goal neutrality, in our view, best supports the conclusion that the interpretive rules that govern these contracts should be majoritarian defaults. This view raises the question: what are the majoritarian preferences? We attempted to show that business parties commonly prefer courts to make interpretations on a narrow evidentiary base that includes the contract, the pleadings, evidence that does not involve the interpretation dispute, and the life experience of the court. Extrinsic evidence is outside this base.

Our views have generated both interest and criticism, but much of the criticism does not engage our arguments. The central objection to our theory of interpretation is that contextual interpretation--the use of extrinsic evidence--yields more accurate results than acontextual interpretation. More evidence, that is, is better than less. Critics have not focused on the claim that goal neutrality implies default interpretive rules. If that argument is conceded, however, the critics' view reduces to a non sequitur. The issue is not which interpretive rules yield more accuracy. The issue, rather, is which interpretive defaults best reflect the parties' preferred tradeoffs among accuracy in interpretation, contract-writing costs, and contract enforcement costs. Our critics largely ignore costs, and so they raise no effective challenge to our view regarding what are good defaults.

We end by acknowledging that much is still to be learned about contract interpretation. We attempt to contribute to this quest by identifying the fight questions. Should the interpretive theory be goal neutrality or some substantive theory? What interpretive rules follow from any theory an analyst can plausibly defend? If those rules are defaults, what are the best reconstructions of party intent regarding interpretation? Identifying the rules that may permit a court best to recover the parties' intentions is just a step toward answering the right questions.

(1.) A strong majority of U.S. courts continue to follow the traditional, "formalist" approach to contract interpretation. A state-by-state survey of recent court decisions shows that thirty-eight states follow the textualist approach to interpretation. Nine states, joined by the Uniform Commercial Code for sales cases (UCC) and the Restatement (Second) of Contracts, have adopted a contextualist or "antiformalist" interpretive regime. The remaining states' doctrines are indeterminate. See U.C.C. 55 1-205, 2-202, 2-208 (2003); RESTATEMENT (SECOND) OF CONTRACTS [section] [section] 200, 209 (1981); Robert E. Scott, State by State Survey (Oct. 7, 2009) (on file with authors).

(2.) See infra text accompanying note 17.

(3.) Judge Richard Posner has estimated that many, if not a majority, of the contract cases he sees present interpretation disputes. Richard A. Posner, American Law and Economics Association Paper Presentation (May 6, 2005); see Richard A. Posner, The Law and Economics of Contract Interpretation, 83 TEX. L. REV. 1581, 1582 (2005) [hereinafter Posner, Law and Economics]. An early empirical study found that twenty-six percent of a sample of five hundred cases raised interpretation and parol evidence issues. Harold Shepherd, Contracts in a Prosperity Year, 6 STAN. L. REV. 208, 223 (1954); see also David A. Dilts, Of Words and Contracts: Arbitration and Lexicology, DISP. RESOL. J., May-July 2005, at 41, 43 ("The construction of contract language is the controversy most evident in contract disputes."); John P. Tomaszewski, The Pandora's Box of Cyberspace: State Regulation of Digital Signatures and the Dormant Commerce Clause, 33 GONZ. L. REV. 417, 432 (1997) ("Most contract litigation involves disputes over construction of the terms in a contract.").

(4.) Alan Schwartz & Robert E. Scott, Contract Theory and the Limits of Contract Law, 113 YALE. L.J. 541 (2003).

(5.) As appears below, an intelligent pursuit of this goal satisfies such second order goals as predictability and stability.

(6.) See, e.g., Shawn J. Bayern, Ratonal Ignorance, Rational Closed-Mindedness, and Modern Economic Formalism in Contract Law, 97 CAL. L. REV. 943 (2009); James W. Bowers, Murphy's Law and the Elementary Theory of Contract Interpretation: A Response to Schwartz and Scott, 57 RUTGERS L. REV. 587 (22005); Juliet P. Kostritsky, Plain Meaning vs. Broad Interpretation: How the Risk of Opportunism Dei:eats a Unitary Default Rule for Interpretation, 96 KY. L.J. 43 (2007); see also Jeffrey M. Lipshaw, The Bewitchment of Intelligence: Language and Ex Post Illusions of Intention, 78 TEMV. L. REV. 99 (2005) [hereinafter Lipshaw, Bewitchment of Intelligence] ; Jeffrey M. Lipshaw, Models and Games: The Difference Between Explanation and Understanding for Lawyers and Ethicists, 56 CLEV. ST. L. REV. 613 (2008) [hereinafter Lipshaw, Explanation and Understanding] (criticizing Schwartz & Scott, supra note 4, but supporting formalism on noneconomic grounds).


(8.) Id. at 30, 46, 215-2o.

(9.) Id. at 1-6.

(10.) Id. at 78-81.

(11.) An accurate interpretation recovers the parties' objectively manifested intentions concerning both the objectives or "ends" of their agreement and the "means" they may have chosen to determine those ends should they later dispute the meaning of the agreement. Courts and scholars typically identify accurate interpretation only with a court recovering the parties' contractual objectives. But accurately recovering the parties' shared objectives (or ends) is only one aspect of determining their shared intention. The parties also can (and often do) specify ex ante the particular sources that will serve as the means for determining their shared ends ex post. In other words, parties can have a shared intention about how their ultimate intentions are to be determined. Avery Weiner Katz, The Economics of Form and Substance in Contract Interpretation, 1O4 COLUM. L. REV. 496, 521 (2004) (referencing Eric A. Posner, The Parol Evidence Rule, the Plain Meaning Rule, and the Principles of Contractual Interpretation, 146 U. PA. L. REV. 533, 57o-71 (1998)). The chosen contractual means for achieving the parties' ultimate ends may ask the court to limit its consideration of relevant evidence to the contract itself (for example, by electing to integrate the contract with a merger clause). We argue below that business parties choose various means by which their contractual ends are to be determined and that courts should respect those preferences. For discussion of the importance of respecting the intended contractual means of sophisticated parties, see Jody S. Kraus & Robert E. Scott, Contract Design and the Structure of Contractual Intent, 84 N.Y.U.L. REV. 1023 (2009).

(12.) Ambiguity is well understood, but the language issue requires some elaboration. Assume, for example, that a contract requires the buyer to take "a minimum of 2000 units." Under the standard language interpretation, the buyer would be in breach if he refused to accept more than 1000 units. The buyer claims, however, that, in the parties' industry, the phrase "minimum" is an estimate, so that parties expect to renegotiate the contract quantity if circumstances materially change after the contract is concluded. The phrase "minimum" is clear in the standard language, and may be clear (to the parties) in the industry's language, but a dispute exists: Should the court restrict its interpretive base to the standard language? Or should it admit evidence that the relevant term was written in the industry language?

(13.) This straightforward claim has been misunderstood by some critics. To clarify: by "the contract" we mean just that--an examination of the contested words or phrases in light of a reading of the contract as a whole--together with the pleadings and briefs describing the parties' performance under the contract. See infra text accompanying notes 29-30.

(14.) Schwartz & Scott, supra note 4, at 584-90. Parties routinely define technical terms and specify particular meanings in extensive definition provisions in their written agreement. See infra text accompanying note 90.

(15.) For empirical support for the predictions of the theory, see infra notes 71-74 and accompanying text.

(16.) A merger or integration clause recites that all prior party understandings are merged into the final written agreement. See infra note 7o and accompanying text.

(17.) See, e.g., Kallman v. Radioshack Corp., 315 F.3d 731, 735 (7th Cir. 2002) ("[I]nterpretafion of the terms of an unambiguous contract is traditionally a question of law and is particularly suited to disposition on summary judgment."); Thrower v. Anson, 752 N.W.2d 555, 560-61 (Neb. 2008); Drake v. Hance, 673 S.E.2d 411, 413 (N.C. Ct. App. 2009); see also William C. Whifford, The Role of the Jury (and the Fact/Law Distinction) in the Interpretation of Written Contracts, 2001 WIS. L. REV. 931, 938 n.18 ("[M]ost courts find that if the written terms have a clear or plain meaning, summary judgment is appropriate, and jury consideration of the meaning of extrinsic evidence is not necessary."); /d. at 952 n.57 ("The combination of a plain meaning rule and a hard [parol evidence rule] keep some cases out of the jury's hands altogether (i.e., permit disposition on summary judgment) in circumstances where alternative interpretive rules would require submission of the case to a jury.").

(18.) An illuminating article taking this line is Victoria Nourse & Gregory Shaffer, Varieties of New Legal Realism: Can a New World Order Prompt a New Legal Theory?, 95 CORNELL L. REV. 61 (2009).

(19.) Professor Burton, in particular, claims that we are literalists: "The reason for neglecting [our and Judge Posner's 'economic' theories] is that the leading analyst's conclusions center strikingly on literalism as the preferred theory of contract interpretation." BURTON, supra note 7, at 219 (footnote omitted). Professor Burton has taken the criticism of our formalism to an extreme. He claims that, under our view, "only the dictionary and the governing contract words in the document may be taken into account [by a court]." ld. at 14. Literalism, as he describes it, urges courts to ignore both context and "the document as a whole"; rather, a literalist theory can restrict a court's attention to "a single word." ld. at 28, 37. We actually said: "[A] minimum evidentiary base is required for any coherent interpretation. This minimum base.., is composed of the parties' contract, a narrative concerning whether the parties performed the obligations that the contract appears to require, [and] a standard English language dictionary.. " Schwartz & Scott, supra note 4, at 572. We also argued that parties would "invest resources in drafting until the writing is sufficiently clear" for a court to understand. Id. at 577. Neither we, nor any theorist we know of, urges interpreting courts to restrict themselves to particular words or phrases in a contract.

(20.) The default evidentiary base that parties prefer courts to exclude is constituted by 0) practice under prior agreements; (2) practice under the litigated agreement; (3) precontractual negotiations; (4) precontractual memoranda and other documents that are not attached to, or referred to, in the contract; and (5) custom. Schwartz & Scott, supra note 4, at 572.

(21.) Professor Burton, for example, rarely mentions cost in a book that has 226 pages of text. Other recent critiques also ignore or dismiss the relevance of costs. See sources cited supra note 6. An important exception is Posner, Law and Economics, supra note 3. Professor Bayern briefly considers and rejects our cost concern in the settlement context. See Bayern, supra note 6, at 969-70. He remarks that parties should settle contractual disputes with the same frequency whether courts use textual or contextual interpretive rules. However, if "at least one of the parties is irrationally optimistic about his chances of getting a favorable (biased) interpretation from the court, then we would expect a rule that allows the maximal evidentiary base to be systematically better .... because, as the variance in the court's expected result decreases, there is simply less to be unduly optimistic about." ld. at 970. Hence, he argues, litigation costs are lower under a contextualist regime. It is difficult to know whether this argument is sound or not because neither Bayern nor we model the behavior of "irrationally optimistic" firms. Rather, we claim that "it is a plausible working assumption that firms rationally pursue" profit maximization. See Schwartz & Scott, supra note 4, at 551 & n.18. Bayern does not seriously contest this assumption.

(22.) See, e.g., BURTON, supra note 7, at 214-18.

(23.) Id. at 216-17.

(24.) For discussion of our views on interpretation and procedural values, see Schwartz & Scott, supra note 4, at 584-90. For default rules, see/d, at 594-609.

(25.) See Alan Schwartz, The Default Rule Paradigm and the Limits of Contract Law, 3 S. CAL. INTERDISC. L.]. 389, 4o2-o3 (1993); Schwartz & Scott, supra note 4, at 596.

(26.) This view often is expressed at conferences whose members are methodologically diverse. An unusually candid written statement is BURTON, supra note 7, at 16. More commonly, articles proceed as if law and economics analyses do not exist.

(27.) When individual persons are parties to transactions, cognitive psychology and behavioral economics are useful in explaining contracting behavior. For recent efforts by one or the other of us along these lines, see Alan Schwartz, How Much Irrationality Does the Market Permit?, 37 J. LEGAL STUD. 131 (2008); Robert E. Scott, A Theory of Self-Enforcing Indefinite Agreements, 103 COLUM. L. REV. 1641 (2003); Robert E. Scott, The Limits of Behavioral Theories of Law and Social Norms, 86 Va. L. REV. 16o3 (2000); and M. Keith Chen & Alan Schwartz, Intertemporal Choice and Legal Constraints (John M. Olin Ctr. for Studies in Law, Econ., and Pub. Policy, Research Paper No. 381, 2009), available at For an effort to combine economic and philosophical insights, see Daniel M. Markovits & Alan Schwartz, The Dual Performance Hypothesis and the Myth of Efficient Breach (Feb. 9, 2010) (unpublished manuscript, on file with authors).

(28.) A first order preference implies the choice of an interpretive rule; a second order preference resolves ties. For example, one state may prefer efficient interpretive rules. Another state may prefer distributionally fair rules. When it is unclear which interpretation of a contract is efficient, the former state would prefer the distributionally fair interpretation if it has a second order preference for distributional fairness. Similarly, the latter state may have a second order preference for efficiency, and so choose the efficient interpretation when it is unclear which interpretation is fair.

(29.) See RESTATEMENT (SECOND) OF CONTRACTS [section] 201(2) (1981) ("Where the parties have attached different meanings to a promise or agreement or a term thereof, it is interpreted in accordance with the meaning attached by one of them if at the time the agreement was made ... (b) that party had no reason to know of any different meaning attached by the other, and the other had reason to know the meaning attached by the first party." (emphasis added)).

(30.) The search for intention, even if defined objectively and prospectively, is more difficult than is commonly believed. Courts typically identify contractual intent with the parties' contractual objectives--the contractual ends of their agreement. But reaching agreement on a shared objective is only one aspect of a shared intention. The parties must also create the particular rights and duties that will serve as their contractual means for achieving their shared ends. Courts frequently conflate the parties' contractual means with their contractual ends. By doing so, they can inadvertently reduce the range of contractual arrangements to which contract law gives effect, thereby potentially depriving commercial parties of essential tools for contract design. For discussion, see Kraus & Scott, supra note 11.

(31.) Those who argue for mandatory contextualist interpretations often justify such rules as necessary to prevent exploitation of unsophisticated individuals, susceptible to cognitive biases, who enter into written contracts with sophisticated parties who supply written contract terms that alter previously settled understandings. See, e.g., Masterson v. Sine, 436 P.2d 561, 564 (Cal. 1968) ("IT]he party urging the spoken as against the written word is most often the economic underdog, threatened by severe hardship if the writing is enforced."). Even assuming that these concerns are valid reasons for imposing a mandatory contextulist regime in contracts between individuals and firms, they do not apply to the firm-to-firm negotiated contracts that are the focus of our theory.

(32.) Recall that in parol evidence contests the issue turns on whether the parties intended the written agreement to contain all the terms of the contract or whether the parties intended the contract to consist of other terms, such as prior understandings or relevant usages in addition to those in the writing. For discussion, see ROBERT E. SCOTT & JODY S. KRAUS, CONTRACT LAW AND THEORY 541-46 (4th ed. 2007).

(33.) This claim could be based on the argument that other sources of evidence-such as the parties' prior negotiations-show that the written term is ambiguous, or that those other sources show that the parties attached a specialized or private meaning to the term in question. See, e.g., Pac. Gas & Elec. Co. v. G.W. Thomas Drayage & Rigging Co., 442 P.2d 641 (Cal. 1968).

(34.) See, e.g., BURTON, supra note 7, at 193-202.

(35.) Everyone agrees that the parol evidence rule should permit parties to exercise control over the evidentiary base by either expanding or contracting the sources of evidence used to identify the contract terms. Courts that apply a soft parol evidence rule, however, deny presumptive effect to merger clauses and reject the four comers rule under which a written contract that appears complete on its face is held presumptively fully integrated with respect to its terms. In consequence, soft parol evidence rules are effectively contextualist. The same point holds for the rules governing the meaning of terms. We expand these points in Part III.

(36.) Recall that our theory holds only for the interpretation of contracts between business firms. Schwartz & Scott, supra note 4, at 544-47. Throughout this Review, we refer to the firm-tofirm contractors in this "subworld" as "parties" without further elaboration.

(37.) SCOTT & KRAUS, supra note 32, at 543-44- Courts generally do not distinguish between "vague" and "ambiguous" terms and we follow that convention here. A typical judicial definition of ambiguity includes, for example, any term or word that "has no definite significance or... is capable of more than one sensible and reasonable interpretation." Ross Bros. Constr. Co. v. State ex tel. Transp. Comm'n, Highway Div., 650 P.Ed 1080, 1082 (Or. Ct. App. 1982). For discussion of the variety of contract-writing errors that parties make, see Charles J. Goetz & Robert E. Scott, The Limits of Expanded Choice: An Analysis of the Interactions Between Express and Implied Contract Terms, 73 CAL. L. REV. 261, 265-73 (1985).

(38.) Support for an argument of this type can be found in Ronald J. Gilson & Alan Schwartz, Understanding MACs: Moral Hazard in Acquisitions, 21 J.L. ECON. & ORG. 330 (2005).

(39.) We illustrate our claims here using the ambiguity question, on which most of our critics focus. Regarding the language issue, we argued earlier that the default should take parties to have written in the majority language because this would reduce strategic behavior. Under the alternative default, a party who wants to exit what has turned out to be a bad deal can easily create an interpretive dispute by claiming that a clear term has a different meaning in the parties' private language. Parties wishing to avoid this possibility must contract to exclude all of the possible private languages from the evidentiary base if there is a dispute. It is more efficient to suppose that parties wrote in the majority language, so that they need only opt into the particular technical language they want to use (supposing them to prefer any such language). Schwartz & Scott, supra note 4, at 584-90.

(40.) See Robert E. Scott & George G. Triantis, Anticipating Litigation in Contract Design, 115 YALE L.J. 814, 857 (2006) (discussing the lack of knowledge about judicial treatment of default rules).

(41.) In Part II we attempt to refute criticisms of the defaults we identify.

(42.) See generally Steven Shavell, On the Writing and Interpretation of Contracts, 22 J.L. ECON. & OPG. 289 (2006) (describing optimal interpretation rules for efficiency minded courts).

(43.) We urge the state to adopt this goal for business contracts.

(44.) It may be helpful to expand this point. The terms in a commercial contract perform two functions. They describe what the parties intend to trade, and they create incentives for parties to maximize expected surplus. The price divides this surplus. As interpretive accuracy increases, the incentive effect of the terms increases. An efficient interpretive rule thus encourages parties to describe their intentions dearly, and to introduce context evidence at trial when this would better improve judicial accuracy. As said above, however, contract-writing cost is positive and litigation cost is positive. Hence, an efficient interpretive rule would maximize an expression such as the following:

[max.sub.t,c] [[infinity]summation over (k=1)][u(k)-pr(l(t))-kc].

Here k is an individual contract; t is trial cost and c is contract-writing cost. The first term in the summation is the benefit of contracting in creating good incentives for the parties; the second term is the expected trial cost (the probability that the contract will raise an interpretation issue that requires more than the minimum evidentiary base to resolve); the third term is the cost of writing a contract. We argue that parties commonly can solve this maximization problem better than courts because parties are better informed about the benefit term and the two cost terms. Hence, while an ideal adjudicator can be trusted to solve this problem for society, efficiency minded courts should solve the problem only when parties do not instruct them. A more formal analysis of interpretation along these lines is in Alan Schwartz & Joel Watson, Contract Interpretation and Contextual Asymmetry (Jan. 27, 2010) (unpublished manuscript, on file with authors).

(45.) We refer here to parties' ex ante preferences. Conflicts can arise in litigation because parties have strategic reasons to disagree on evidence production.

(46.) A complete statement is in Schwartz & Scott, supra note 4, at 573-84.

(47.) Professor Bayern rejects our argument on this point, but he is mistaken. See Bayern, supra note 6, at 959-61. He considers an arbitrary distribution of numbers (from -3 to 3) from which a court can pick. The midpoint is zero. He then takes us to claim that the expected value of the court's pick is zero because there is no more reason to believe that the court will pick a higher number than zero than that the court will pick a lower number. But, he says, there is no more reason to believe that the court will pick a number higher than two than lower than two either. Bayern's argument holds for numbers because he implicitly assumes that his number distribution is uniform. The probability of any particular outcome materializing from a uniform distribution is the same as any other outcome materializing. For example, the midpoint of a uniform distribution ranging from zero to one is .5. Beginning at zero, the probability that an outcome equals .1 is the same as the probability that the outcome equals .2, and so on. Now consider a normal distribution (think of a bell shaped curve). The midpoint of this distribution is the most likely probability. The curve describing the distribution is rising until the curve reaches a maximum because the relevant graph plots probabilities on the vertical axis and values on the horizontal. Therefore, starting with zero, the probability that an outcome equals, say, .1 is less than the probability that the outcome equals .2. Similarly, the probability of outcomes declines as one moves past the maximum. We argued that the space of interpretive outcomes is normally distributed. Normality is implied by our two assumptions. If the court is more likely to be right than to be wrong, then the curve describing the interpretation space has a unique maximum at the correct interpretation. If the court is unbiased, then this curve is symmetrical. A normal distribution is symmetrical with a unique maximum. Moreover, our assumptions are supported by the observation that (a) parties want to maximize the probability that the court will draw the correct interpretation from the space of interpretations; and (b) parties have some control over the outcome because they write the contract that the court later will interpret. Thus, we say:
   It is optimal for risk-neutral firms to invest resources in
   drafting until the writing is sufficiently clear.., so that the
   mean of the distribution of possible judicial interpretations is
   the correct interpretation .... [F]irms will attempt to write
   contracts with sufficient clarity to permit courts to find correct
   answers, though with error.

Schwartz & Scott, supra note 4, at 577. It would be irrational of parties to prefer the space of interpretive outcomes to be uniform. To see why, suppose the space of outcomes is uniformly distributed between zero and one, and the correct interpretation is described by the number .4. Then, the probability that the court will interpret the contract to realize the outcome .4 is only 10% because every outcome is equally likely. The court is effectively random. Parties who prefer random courts would not incur the costs of putting description information in their contracts, but real parties always do. Bayern's example thus is irrelevant to the interpretation debate because it implicitly and erroneously assumes that the space of interpretive outcomes is uniformly distributed.

(48.) See, e.g., Bayern, supra note 6, at 953 n.31, 962 n.60; Bowers, supra note 6, at 598.

(49.) This note is for generalists. The declining marginal utility of money theory holds that individuals have strictly concave utility functions. In lay language, this means that the first dollars a person receives will generate a great deal of utility; these dollars will purchase food and shelter. More dollars than this satisfy ever less urgent needs, so that a well-off person ultimately is choosing among brands of champagne. That the marginal utility of a dollar declines as the person has more dollars implies that individuals weigh losses more heavily than gains. A gain adds to wealth while a loss detracts from it. Individuals thus are not "neutral," say, between a .5 chance to earn a dollar and a .5 chance to lose a dollar. The gain would yield less utility than the loss would take away. People thus must be paid to bear risk. It is otherwise with firms that maximize profits. These firms have linear utility functions. Each dollar of additional profit counts as much for the firm as does the prior dollar. Since firms care only about money, a firm is "neutral" between a .5 chance to earn a dollar and a .5 chance to lose a dollar. Such a firm also is neutral between a .5 chance to earn $10,000 and a .5 chance to lose $10,000. The latter gamble has a much greater variance than the former gamble. This is why we say that firms care much more about the mean--they prefer higher expected values to lower--than they care about the variance.

(50.) See Schwartz & Scott, supra note 4, at 575-77.

(51.) Bayern, supra note 6, at 97o n.81; Kostritsky, supra note 6, at 58.

(52.) Bowers, supra note 6, at 601.

(53.) Id. To "chisel" is to supply less than the performance the contract required, but to claim that the performance is compliant, so that the promisor is entitled to the full price.

(54.) Professor Burton believes that the "interpreter's experience and understanding of the world ... should be excluded in principle" because the rule of law virtues require exclusion and because the parties do not "want the resolution of their disputes to turn on" the interpreter's identity. BURTON, supra note 7, at 37. The former claim requires more argument than Burton provides. For example, the rule of law virtues are thought not to be violated even though the fate of a criminal defendant is importantly a function of whether the trial judge is a former prosecutor or a former public defender. The latter claim is questionable as well because parties choose the enforcement institution. They can agree to a court, or choose arbitration and sometimes choose the actual arbitrator. A contrary view that the court's experience is helpful for interpretation--is found in Posner, Law and Economics, supra note 3, at 1603-06. For example, Judge Posner asserts:
   But often, when the parties' intentions are not readily inferable
   from the written contract, the best, the most cost-efficient, way
   to resolve their dispute is not to take testimony and conduct a
   trial; it is to use commercial or economic common sense to figure
   out how, in all likelihood, the parties would have provided for the
   contingency that has arisen had they foreseen it.

Id. at 1605.

(55.) In this and the following note, we add a formal version of the argument we advance in the text. Readers for whom the text is sufficient can skip these notes. Denote a compliant performance as [i.sup.*]. The probability of a compliant performance, on our assumptions, is p([i.sup.*]) [epsilon] [1/2, 1]. The seller's gain from complying and being found to comply is her expectation. The seller also receives her expectation if the court errs on the low side. The damages the seller pays if she is found not to comply are the buyer's expectation. If the parties have equal bargaining power, which is a standard assumption, the parties' expectations are equal: they split the expected surplus fifty-fifty. Denote the seller's expectation, which thus equals the buyer's damages, as e*. Then the seller's expected gain from complying is:

[[pi].sub.c] = [e.sup.*][p([i.sup.*])+ 1 - p([i.sup.*])/2] -[e.sup.*][1 - p([i.sup.*])/2] =[e.sup.*](p([i.sup.*])).

(56.) Denote the seller's gain from cheating as G = k - q where k is the contract price and q is the cost of a noncompliant performance. Then q < c, where c is the cost of the performance that complies with the contract. The seller's expected gain from cheating is:

[[pi].sub.B] = G [1 - p(i.sup.*]/2] -[e.sup.*][p([i.sup.*]) + 1 - p([i.sup.*])/2]. =[e.sup.*](p([i.sup.*])).

The seller complies if [[pi].sub.C] > [[pi].sub.B]. To get a sense of how likely compliance is, assume that the court is barely accurate: p([i.sup.*]) = .55. Then, saving the reader a lot of algebra, if cheating is costless (q = o), the seller nevertheless complies if c < .83k. The seller's profit is price less cost. Hence, when cheating is costless, and the court is barely accurate, the seller nevertheless complies unless her profit from compliance would be relatively small (seventeen percent or less of the price in this example).

(57.) We omit for convenience a term that reflects the seller's expected damage liability. When the seller complies, her expected damage liability is zero, as said above.

(58.) Recall that the buyer values the particular structure the contract required more than any other buyer.

(59.) The argument in the text made two important assumptions. First, it assumed a symmetric distribution of possible interpretations, but parties sometimes face skewed distributions. The common practice in these cases is for parties to contract into symmetric distributions. See Schwartz & Scott, supra note 4, at 581-83. Second, in the text's example the probability distribution is invariant to the nature of the seller's performance. This assumption may seem unrealistic because the court sees the performance, and may update its interpretation to reflect that performance. The important feature of the example is that the court does not reduce the probability that the parties intended to trade an S* structure when the seller tenders an S- structure. The motivation for this assumption is as follows. Let the court see the contract and the pleadings, and use its experience to infer that the parties at bar are among the set of contracting types that intend to trade S* structures. Now assume that the seller actually produces one of these structures. When the court sees S*, its belief that the parties at bar are S* types thus increases. Now let the seller produce the noncompliant structure S. A sophisticated court would not update its prior understanding to increase the probability that the parties are among the contracting types that intend to trade [S.sup.-] structures, so that the seller complied by producing one of them. Parties would have a strong incentive to degrade performances were a court to increase its probability that a degraded performance was intended from observing a degraded performance. Rather, the court rationally treats a degraded performance as having no informational content; the performance, that is, will not change the court's prior belief that the parties intended to trade an S* structure. Hence, we assume that a seller who does not comply expects to face a probability distribution of interpretations that is not skewed in her favor by her breach. For further justification of this line of reasoning, see Schwartz & Watson, supra note 44.

(60.) Professors Burton and Bayern make this mistake frequently. See, e.g., BURTON, supra note 7, at 36; Bayern, supra note 6, at 945 ("[E]conomic formalism posits that the law will function better, in view of the costs of adjudication, if courts narrowly limit the field of information or arguments that they permit themselves to consider when hearing cases.").

(61.) A 10-K contains an extensive description of the flier's business.

(62.) This is trivially true in a default world: parties who want courts to see more than we suppose would contract out of our default, and parties who want courts to see less than they now see would contract for our rule if the current rules were defaults.

(63.) See supra Section I.A.

(64.) Courts thus apply the plain meaning or "four corners" rules with considerable knowledge of the agreement's context. Our claim is that in the common case courts do not need the additional contextual insight that evidence bits such as industry custom or prior practice would yield in order to be fight on average.

(65.) Bayern, supra note 6, at 955-58, 967.

(66.) To formalize this point, let N be the number of bits required to describe the parties' context perfectly, so that the court could not mistake what the parties agreed to deliver. Hence, the number of bits the parties use in a contract can be all of N or a subset of N. We can denote the optimal--that is, the cost justified--number of bits for a particular contract as n(k*) [member of] [1, ..., N]. Based on the two assumptions in text, an optimal contract contains less than all N bits: n(k*) < N. Given concavity in the return function, the argument in the text also goes through if contract-writing costs are nondecreasing in the number of bits. An illuminating way to put this conclusion is that there commonly exists "contextual information asymmetry" between parties and courts.

(67.) Contract-writing problems of this type are formally analyzed in Schwartz & Watson, supra note 44. The challenge parties face in writing contract terms that minimize later interpretative disputes over meaning is discussed in Goetz & Scott, supra note 37, at 281-86.

(68.) Alliance Agreement, E.I. du Pont de Nemours & Co.-EarthShell Corp., July 25, 2002, available at

(69.) SCOTT & KRAUS, supra note 32, at 228 n.13. See, for example, the following typical exclusionary clause:
   No legally binding obligations on the parties will be created,
   implied, or inferred until appropriate documents in final form are
   executed and delivered by each of the parties regarding the subject
   matter of this [agreement].... and containing all other essential
   terms of an agreed upon transaction. Without limiting the
   generality of the foregoing, it is the parties['] intent that,
   until that event, no agreement binding on the parties shall exist
   and there shall be [no] obligations whatsoever based on such things
   as parol evidence, extended negotiations, "handshakes," oral
   understandings, or course of conduct (including reliance and
   changes of position)....

Id. at 228-29 (emphasis added).

(70.) Merger clauses are ubiquitous in contracts between firms. For a typical example, see the following:
   This Contract constitutes the entire agreement of the parties
   hereto and supersedes all prior and contemporaneous agreements,
   representations and understandings of the parties. No waiver of the
   provisions of this Contract shall be deemed or shall constitute a
   waiver of any other provisions, nor shall any waiver constitute a
   continuing waiver. This contract may not be supplemented, altered,
   modified or amended or otherwise changed except by an instrument in
   writing signed by the parties hereto. The course of dealing or
   course of performance between the parties hereto shall not commit
   either party to duties or obligations which are not expressly
   stated by this Contract.

Fruit Purchase Contract, Cargill Juice N. Am., Inc.,-S. Ford Meade Land Mgmt., Aug. 31, 2005 (on file with authors).

(71.) See Lisa Bernstein, Merchant Law in a Merchant Court: Rethinking the Code's Search for Immanent Business Norms, 144 U. PA. L. REV. 1765, 1772-77, 1787-95, 2825-20 (1996) (concluding that formalist contract enforcement by a private system of merchant courts established by the National Grain and Feed Association results from recognition of the role of extralegal, reputational sanctions); Lisa Bernstein, Private Commercial Law in the Cotton Industry: Creating Cooperation Through Rules, Norms, and Institutions, 99 MICH. L. REV. 1724, 1745-54, 2762-87 (2001) (discussing the existence, importance, and supporting institutional conditions of formalist interpretation together with reputation-based, nonlegal sanctions in the cotton industry).

(72.) Theodore Eisenberg & Geoffrey P. Miller, The Flight to New York: An Empirical Study of Choice of Law and Choice of Forum Clauses in Publicly-Held Companies' Contracts (N.Y.U. Law & Econ. Research Paper Series, Working Paper No. 08-13, 2008), available at http ://

(73.) Id. at 19, 34. Delaware was a distant second to New York, with about fifteen percent of the parties choosing its law. No other state accounted for even ten percent of the choices of law. Id. at 19.

(74.) Geoffrey P. Miller, Bargaining on the Red-Eye: New Light on Contract Theory (N.Y.U. Law & Econ. Research Paper Series, Working Paper No. 08-21, 2008), available at

(75.) Id. at 1.

(76.) The following description of the sequence in which contract disputes proceed in litigation was developed in conversations with a number of commercial litigators as well as from the experiences that both of the authors have had serving as expert witnesses in contract litigations.

(77.) Albert Choi and George Triantis argue that parties sometimes want to increase their cost of proving outcome variables such as an agent's effort after the contract is concluded. The parties realize this goal by conditioning contractual payoffs on these costly to verify variables. See Albert Choi & George Triantis, Completing Contracts in the Shadow of Costly Verification, 37 J. LEGAL STUD. 503 (2008). The contracts that do this have the virtues of increasing the expected value of the sanction for breach (the breacher bears the high litigation cost) and deterring litigation that has a low probability of success. This interesting theory is not relevant to the case for cost reduction that we make. Choi and Triantis focus on outcome variables, but interpretation issues require courts to recover "context variables": information that details the parties' circumstances when the parties contracted. Thus, in an illustration above, the outcome variable was the quality of the structure the seller delivered; the context variables were the business the parties were in and the purpose for which they dealt. Choi and Triantis do not argue that parties prefer to increase their cost of establishing context variables, and we believe that parties prefer not to do so. The better informed the court is about context, the more likely the court is to recover the parties' intentions. Hence, parties prefer to reduce the cost of communicating context evidence. It is consistent with the Choi and Triantis theory that parties want to reduce the cost of showing what the agent was supposed to do while also wanting to increase the cost of proving what the agent actually did. This is because the cost of proving an outcome variable cannot be relevantly increased unless the court knows what the contract requires the party to prove.

(78.) Professor Bayern apparently assumes that a court that accepts the objective theory of contract is necessarily committed to a formalist theory of interpretation. See Bayern, supra note 6, at 960 n.53 (suggesting that if the question of intent is determined objectively and prospectively "there is little else for critics of the argument to do but reject the [objective] axiom"). His assumption is incorrect. Courts in a contextual regime admit more evidence than courts in a formalist regime, but the evidence in both regimes is used to answer the same question: what could the promisee reasonably take the promisor to have promised?

(79.) Kraus & Scott, supra note 21, at 1057.

(80.) See, e.g., Eric A. Posner, The Parol Evidence Rule, the Plain Meaning Rule, and the Principles of Contractual Interpretation, 146 U. PA. L. REV. 533, 534 (1998) (distinguishing two polar positions: "hard-PER" where "the court generally excludes extrinsic evidence and relies entirely on the writing," and "soft-PER," where "the court gives weight both to the writing and to the extrinsic evidence"); see also Robert A. Hillman, The "New Conservatism" in Contract Law and the Process of Legal Change, 40 B.C.L. REV. 879, 879-80 (1999) (same); Ralph James Mooney, The New Conceptualism in Contract Law, 74 OR. L. REV. 1131, 1133-35 (1995) (identifying the split between hard and soft interpretive regimes). For discussion of the split, see SCOTT & KRAUS, supra note 32, at 578-618.

(81.) See, e.g., Morgan Stanley High Yield Sec., Inc. v. Seven Circle Gaming Corp., 269 F. Supp. 2d 206 (S.D.N.Y. 2003) (holding that the prior agreement is excluded where the writing appears, in view of thoroughness and specificity, to embody a final agreement); Intershoe, Inc. v. Bankers Trust Co., 571 N.E.2d 641, 644 (N.Y. 1991) (same); Mitchill v. Lath, 160 N.E. 646, 646-48 (N.Y. 1928) (upholding the four corners presumption and excluding evidence of collateral agreement to a land sale contract).

(82.) Merger clauses are given virtually conclusive effect in New York. See Tempo Shain Corp. v. Bertek, Inc., 120 F.3d 16, 21 (2d Cir. 1997) ("Ordinarily, a merger clause provision indicates that the subject agreement is completely integrated, and parol evidence is precluded from altering or interpreting the agreement."); Jarecki v. Shung Moo Louie, 745 N.E.2d 1006, 1009 (N.Y. 2001) ("The purpose of a merger clause is to require the full application of the parol evidence rule in order to bar the introduction of extrinsic evidence.... The merger clause accomplishes this purpose by evincing the parties' intent that the agreement 'is to be considered a completely integrated writing.'" (quoting In re Primex Int'l Corp. v. Wal-Mart Stores, 89 N.Y.2d 594, 600 (1997)) (internal citations omitted)); Norman Bobrow & Co. v. Loft Realty Co., 577 N.Y.S.2d 36, 36 (App. Div. 1991) ("Parol evidence is not admissible to vary the terms of a written contract containing a merger clause.").

(83.) Courts have long recognized that a writing can be found to be a total integration even in the absence of a merger clause. See, e.g., Morgan Stanley High Yield Sec., Inc., 269 F. Supp. 2d at 214 ("Under New York law a contract which appears complete on its face is an integrated agreement as a matter of law."); Mitchill, 160 N.E. 646 (using the "natural omission test" to exclude evidence of a separate oral agreement to add to the terms of a deed that appeared complete on its face despite the absence of a merger clause); see also SCOTT & KRAUS, supra note 32, at 541-43 (discussing the tests for total integration in the absence of a merger clause).

(84.) Pac. Gas & Elec. Co. v. G.W. Thomas Drayage & Rigging Co., 442 641, 645 (Cal. 1968) ("[R]ational interpretation requires at least a preliminary consideration of all credible evidence offered to prove the intention of the parties."); Masterson v. Sine, 436 P.2d 561, 564 (Cal. 1968) (admitting parol evidence to vary terms of deed on ground that "[e]vidence of oral collateral agreements should be excluded only when the fact finder is likely to be misled"); see also Susan J. Martin-Davidson, Yes, Judge Kozinski, There Is a Parol Evidence Rule in California--The Lessons of a Pyrrhic Victory, 25 SW. U. L. REV. 1 (1995) (examining California cases raising the issue of admissibility of extrinsic evidence and concluding that parol evidence jurisprudence in California represented one of the most confused and incoherent areas of law in the state); Scott, supra note I (examining which states follow the California model).

(85.) Miller, supra note 74, at 41-42.

(86.) Int'l Milling Co. v. Hachmeister, Inc., 110 A.2d 186 (Pa. 1955) (holding extrinsic evidence of negotiations and antecedent agreements admissible to show that the buyer had not assented to the contract as a complete integration of the contract despite the presence of an express merger clause); see also RESTATEMENT (SECOND) OF CONTRACTS [section] 216 cmt. e (1981) (same); 3 CORBIN ON CONTRACTS: A COMPREHENSIVE TREATISE ON THE RULES OF CONTRACT LAW [section] 578, at 405 (2d. ed 1960) ("The fact that a written document contains one of these express provisions does not prove that the document itself was ever assented to or ever became operative as a contract. Neither does it exclude evidence that the document was not in fact assented to and therefore never became operative.").

(87.) In sales cases that fall under Article 2 of the UCC, courts are instructed as well to follow a soft parol evidence rule and to reject the plain meaning rule. See U.C.C. [section] 2-202 cmt. 1 (2004) ("This section definitely rejects: (a) any assumption that because a writing has been worked out which is final on some matters, it is to be taken as including all the matters agreed upon; (b) the premise that the language used has the meaning attributable to such language by rules of construction existing in the law rather than the meaning which arises out of the commercial context in which it was used....").

(88.) See BURTON, supra note 7, at 220-23; Lipshaw, Bewitchment of Intelligence, supra note 6, at 112-16.

(89.) See, for example, the following "purpose" clause:

   Apple and SCI entered into a Stock Purchase Agreement on April 4,
   1996 (the "Stock Purchase Agreement") pursuant to which SCI will
   purchase Apple's manufacturing facility located at 702 Bandley
   Drive, Fountain, Colorado ("Fountain") and certain related assets.

   The parties desire that Apple engage SCI to assemble, test and
   package certain Products, Service Units and Spare Parts, as defined
   below, on a turnkey basis at Fountain on the terms and conditions
   of this Agreement. This Agreement defines the general terms and
   conditions governing all transactions between them for Products,
   Service Units and Spare Parts manufactured at Fountain. Individual
   "Product Plans" attached as Addenda to this Agreement, and
   incorporated herein by reference, define the specific terms and
   conditions for each Product, Service Unit and/or Spare Part. The
   initial Product Plans are attached to Exhibit A and numbered A-1
   through A-11. Additional Products and Product Plans may be added to
   this Agreement by addenda to Exhibit A signed by both parties. Such
   addenda will be numbered sequentially, A-12, A-13 and so on.

   In consideration of the above and the mutual promises contained
   herein, Apple and SCI agree as follows....

Fountain Manufacturing Agreement, Apple Computer, Inc.-SCI Sys., Inc., May 31, 1996, available at

(90.) See, e.g., Data Management Outsourcing Agreement, Allstate Ins. Co.-Acxiom Corp., art. 2, Mar. 19, 1999, available at allstate-outsource-1999.03.19.shtml (defining thirty-four technical or nonstandard meanings including specialized meanings of "Agreement," "Confidential Information," "Data Integrity," "Current Projects," "Affiliate," "End-User," "Material Default," "Party," "Person," "Problem," "Term," "Work Order," and "Work Product").

(91.) See, e.g., Fountain Manufacturing Agreement, supra note 89.

(92.) See SCOTT & KRAUS, supra note 32, at 543-45, 578-602 (discussing differences between plain meaning and "contextual" modes of interpretation); Schwartz & Scott, supra note 4, at 584-94 (arguing that most sophisticated parties would prefer plain meaning interpretation because it would (a) reduce contracting costs; (b) minimize the opportunities for strategic behavior; (c) reduce the risk of judicial error; and (d) expand the set of efficient contracts parties could write).

(93.) See supra notes 54-59 and accompanying text.

(94.) The discussion in text assumes that there is a "correct answer" to an interpretive dispute over meaning. But what if there is a genuine ambiguity in the written agreement? In such a case, the divide between formalist and antiformalist positions essentially disappears: a court will consider extrinsic evidence to resolve the ambiguity. If the ambiguity cannot be resolved by additional evidence, the contract may be found too indefinite to enforce. See, e.g., Raffles v. Wichelhans, (1864) 159 Eng. Rep. 375 (Exch.) (holding that "there was no consensus ... and therefore no binding contract" over sale of cotton on one of two ships named "Peerless" sailing from Bombay to Liverpool within three months of each other). Alternatively, depending on the burden of proof, there may be a verdict favoring one of the dueling interpretations offered by the parties. See, e.g., Frigaliment Importing Co. v. B.N.S. Int'l Sales, 190 F. Supp. 116 (S.D.N.Y. 1960) (finding that the meaning of "chicken" was vague and holding for seller on the grounds that buyer, as plaintiff, had not carried its burden of proving which of the two plausible meanings the parties intended). For discussion of the significance of the burden of proof in resolving ambiguities in contracts, see Robert E. Scott & George G. Triantis, Anticipating Litigation in Contract Design, 115 YALE L.J. 814, 859-60 (2006).

(95.) Schwartz & Scott, supra note 4.

Alan Schwartz is Sterling Professor of Law, Yale Law School; Professor, Yale School of Management. Robert E. Scott is Alfred McCormack Professor of Law and Director, Center for Contract and Economic Organization, Columbia University. We are grateful to Victor Goldberg, Jody Krans, Daniel Markovits, Jonathan Rose, Elizabeth Scott, and George Triantis for helpful comments.
COPYRIGHT 2010 Yale University, School of Law
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2010 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Schwartz, Alan; Scott, Robert E.
Publication:Yale Law Journal
Date:Mar 1, 2010
Previous Article:Strategic vagueness in contract design: the case of corporate acquisitions.
Next Article:The significance of signatures: why the framers signed the Constitution and what they meant by doing so.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters