Anticipating litigation in contract design.

INTRODUCTION

In recent years, spurred by theoretical developments in the economics of contract, scholars have focused attention on the problems of incomplete contracting: What prevents parties from writing complete contracts that achieve the dual objectives of efficient reliance and efficient trade? (1) Contract theorists have identified two primary reasons for why parties may agree to contracts that fail to provide for the optimal obligations in each contingency (or state of the world) that might materialize during the term of the contract. (2) First, the front-end transaction costs of anticipating all future states of the world, calculating the efficient outcome in each state, and providing specifically for low-probability states may exceed the resulting gains in contractual surplus. (3) Second, the back-end costs of enforcing contracts may exceed all gains, owing to the difficulty of observing and then verifying to a court private information known only to the parties. (4) Unfortunately, this focus on problems of incompleteness has led scholars to neglect a related, and equally important, question: How do parties manage the costs of creating and enforcing the incomplete contracts that they write ?

Despite its theoretical advances, therefore, the theory of incomplete contracts has yet to yield predictions that are borne out by the realities of commercial practice. This gap between theory and practice is due to a number of limitations in the literature. First, scholars often neglect to weigh contracting costs, at either the front or back end, against the incentive gains that they produce--what we refer to as the incentive bang for the contracting-cost buck.

Second, scholars tend to focus on either front-end or back-end obstacles to complete contracts and assume the absence of friction at the other end. (5) For example, theorists concerned about back-end verification or uncertainty costs assert that parties will tend to avoid vague contract terms such as "best efforts" or "commercial reasonableness." (6) Yet these provisions are commonplace in commercial contracting because they reduce front-end transaction costs. Indeed, the mix of precise and vague terms that characterize the typical commercial contract can be framed as the product of a tradeoff that the parties have made in investing in the front end or back end of the contracting process, based on their particular circumstances. By reaching the optimal combination of front-end and back-end costs, parties can minimize the aggregate contracting costs of achieving a particular gain in contractual incentives. Conversely, for any given expenditure of contracting costs, the parties can reach the highest possible incentive gains by optimizing the allocation of their investment between the front and back ends.

Third, contract theorists assume a highly stylized enforcement mechanism in which the court verifies information and then orders the parties to execute the trade or not to execute it. As noted above, these scholars postulate that some contract provisions are too costly to verify and yield excessively uncertain enforcement outcomes. Their analysis adopts a binary approach in which these terms are labeled nonverifiable, while the remaining provisions can be verified without error at no cost. When parties enter into a legally binding contract, however, they invoke an adversarial enforcement mechanism that is governed by an elaborate set of procedural rules. The parties bear their own evidentiary costs, and a wide range of institutional features constrains the cost of litigation so that the back-end costs are lower than the verification costs envisaged by contract theorists. Moreover, when parties settle or renegotiate rather than litigate, they avoid verification costs entirely. Accordingly, parties may well find it desirable to accept these back-end costs in order to reap savings at the front end. Although the uncertainty in judicial factfinding might undermine contract incentives, the effect is context-dependent, and it is simply one factor to be taken into account in resolving the tradeoff.

Finally, contract theorists focus on substantive contract terms and not on attempts by the parties to regulate the enforcement process. Yet some of the rules governing litigation are default rules that the parties can vary or manipulate in their ex ante contract. By doing so, the parties can further reduce the cost of litigation and improve the ex ante incentive gains from enforcement. This has repercussions on the choice between precise and vague terms. A reduction in back-end enforcement costs should lead the parties to substitute more back-end for front-end investment by replacing precise provisions with vague terms. (7)

In this Article, we explore how the choice between precise and vague terms shifts investment between the front and back end of the contracting process and thereby improves efficiency. In designing their contract, parties choose contract terms based on the expected mechanism of enforcement. We offer a theory of contract design that anticipates the enforcement of contracts by adversarial litigation. Courts do not verify facts by direct investigation, but rather rely on the self-interested evidence presented by the parties. The enforcement of vague terms entails additional layers of evidence production. For example, a promisor would first propose to the court the activities that constitute "reasonable care" and then provide evidence that she performed them. We refer to the intermediate determination as the selection of "proxies" for reasonable care. (8) The choice between precise terms and vague terms thus reduces to who chooses the relevant evidentiary proxies and when they are chosen: the parties at the time of contracting or the court at trial. To illustrate this distinction, we might compare an obligation to deliver a widget weighing ten pounds and an obligation to deliver a widget of merchantable quality.

There are various bits of evidence that can establish the weight of the delivered widget in the first case. For example, compare the testimony of the seller's agent as to the widget's weight immediately before delivery against the testimony of the buyer's agent as to its weight the day after delivery. By specifying the proxy ex ante (a widget weighing ten pounds), the parties delegate to the court the relatively simple task of choosing between these evidentiary bits before deciding whether to find a breach. When the contract requires instead a merchantable widget, the weight of the widget competes with other proxies in establishing merchantability. In this case, the litigation process determines which proxies are relevant and the weight to be assigned to each. The back-end cost is borne only with respect to the contingency that actually materializes, and it might be avoided entirely if the parties settle or renegotiate.

The parties choose between front- and back-end proxy determination by comparing the informational advantage the parties may have at the time of contracting against the hindsight advantage of determining proxies in later litigation. Damages for contract breach provide a familiar illustration of this choice. Suppose that contracting parties wish to set damages so that the breacher internalizes the expectation loss inflicted on the promisee. The parties have a choice between a liquidated damages term and a broad standard of expectation damages (which also happens to be the legal default). The parties might choose liquidated damages that are fixed or otherwise based on fairly specific pieces of evidence, such as market prices. If the parties adopt instead the default of expectation damages, the court will invite the parties to propose proxies for the value of the promisee's lost expectation. Courts regularly require the parties to present market evidence of costs and values, which they then use to measure damages. (9) The court thus chooses among more or less efficient proxies for the promisee's expected losses from breach, in light of the information it enjoys ex post. Efficient proxies are those that maximize the gains in contractual incentives net of expected litigation costs. The parties may agree to liquidated damages, therefore, because they determine that their private information at the time of contracting is superior even to the court's market information ex post.

Our analysis of the tradeoff between front-end transaction costs and back-end enforcement costs owes an intellectual debt to the work of legal scholars who have analyzed the choice between rules and standards in legislation and administrative regulation. (10) They frame the choice between rules and standards to focus on the stage at which content is given to regulation: Either a rule is promulgated before the regulated behavior occurs, or a standard is enforced after the behavior occurs. (11) In a similar manner, we frame the choice between precise terms (rules) and vague terms (standards) as the decision to give content to legal obligations either on the front end or back end of the contracting process.

We build on this analysis in several important respects, however. First, we unpack the enforcement process to represent more accurately how content is injected at the back end. In particular, we treat the back end as an evidentiary process in which the court chooses proxies with which to judge whether the promisor has complied with a vague contract obligation.

Second, in public lawmaking, promulgators are typically legislatures or administrative bodies, whose lawmaking process is complicated by problems of collective decisionmaking and agency relationships. These problems impede the efficient choice between rules and standards, as promulgators may be more or less willing to delegate to a future court. (12) The front-end agency and bargaining relationships in a commercial contract are far more straightforward. Each lawyer represents a single party and is likely to be better informed about the relevant contracting parameters. Thus, we can be more confident that the parties will agree to an efficient mix of rules and standards in their contract.

Third, our analysis recognizes that the parties have some discretion in choosing their mode of enforcement (e.g., arbitration or litigation) or varying some of the rules (e.g., burdens of proof) in order to reduce enforcement costs. Their decisions in this regard bear on the choice between rules and standards.

Finally, we draw an explicit connection between the choice of rules or standards and the complexity or "completeness" of the contract--that is, the degree to which a contract separately addresses different contingencies that call for different obligations. (13) By efficiently choosing between vague and precise terms, the parties can lower the cost of writing a more complete contract. Indeed, by improving the cost-effectiveness of litigation, the parties can incorporate more standards in their contract, and reduce the cost of writing a more complete contract even further.

This Article is organized as follows: Part I examines the determinants of front-end transaction costs and back-end enforcement costs. We focus primarily on the back-end factors contributing to the direct costs of litigation and on the effect of uncertainty and the risk of legal error on contract incentives. The rules of evidence and procedure significantly constrain ex post litigation costs and, in some cases, may thereby expand the opportunities for parties to trade off front-end and back-end costs. We then show how the choice between precise and vague terms implicitly allocates costs between the front end and back end of the contracting process.

Part II explores how parties use precise and vague terms to lower contracting costs by assigning proxy choice either to the parties on the front end or the court at the back end. We set out a general theory of proxy choice and then describe guidelines by which parties select the "chooser." The parties use precise contractual rules to specify proxies whose accuracy is less likely to be affected by the future state of the world, while vague contractual standards delegate to the court the later choice of proxies that are more likely to be state contingent. We then discuss how contracting parties can further improve efficiency by combining precise and vague terms to define the space within which the court, with the aid of interpretative maxims, can select appropriate proxies.

In Part III, we examine how parties can further enhance the benefits of trading off front-end and back-end costs by varying some of the procedural rules that will govern the enforcement of their contract. We examine mechanisms by which the parties tailor burdens of proof to their circumstances. By doing so, the parties reduce enforcement costs, which permits them to achieve even greater incentive gains (or lower contracting costs) by shifting more activity to the back end. Parties shift activity to the back end by substituting vague for precise terms or, more generally, by expanding the proxy space available to the litigation process.

This Article argues that contract design can be improved by anticipating carefully the effect of the course of litigation on contract terms. We provide examples from commercial contracts and judicial opinions suggesting that the prospect of litigation does in fact influence contract design. It is difficult, however, to assess the degree to which parties consciously anticipate litigation in their choice between rules and standards, and the degree to which there remain unrealized gains in doing so. Rather than resolve disputes through litigation, parties may settle or renegotiate their contract. A more complete theory of contract design would anticipate all possible back-end processes and the interaction among them. (14) Our analysis thus calls for further research into the interaction between contract and litigation, as well as future investigation into the effect of other back-end processes, such as arbitration, renegotiation, and settlement. (15)

I. THE FRONT-END AND BACK-END COSTS OF CONTRACTING

Contracts scholarship identifies a wide variety of obstacles that limit the completeness of contracts. As we will describe in greater detail, these contracting costs arise mostly from the fact that information is costly, and they can be divided between costs incurred at the front end and back end of the contracting process. We will refer to the front-end costs as transaction costs and the back-end costs as enforcement costs. The important distinction between the front and back ends is that they are separated by the resolution of uncertainty. For example, the front end is drafting the contract and the back end is litigating disputes that arise when the contract turns out to be a losing proposition for one party.

The goal of contracting parties is to maximize the incentive bang for the contracting-cost buck. (16) Parties thus incur contracting costs to improve the efficiency of incentives in their relationship, particularly the incentive to perform when it is efficient to do so and the incentive to make efficient investments that enhance the value of their exchange. Investment in contracting costs enables parties to write a more complete contract that provides for efficient obligations in a large number of possible states of the world. Parties would wish to minimize contracting costs if the number of possible states provided for in the contract is held constant. However, the parties may wish also to increase contracting costs if that yields a greater gain in the incentives to invest and perform efficiently. Accordingly, the parties should continue to invest in contracting costs until the marginal cost of further investment exceeds the marginal benefit in incentive gains. If circumstances change so as to lower contracting costs or increase the incentive gains at the margin, the parties should increase their investment (and vice versa). For convenience, we will refer to changes in contracting cost per incentive bang to include both changes in cost and changes in incentive effects that stem from incremental investments in making the contract more complete. (17)

Front-end (transaction) costs are relatively straightforward and well documented in the literature. The parties invest in foreseeing possible future contingencies, determining the efficient obligations that should be enforced in each contingency, bargaining over the share of the contracting surplus, and drafting the contract language that communicates their intent to courts. (18)

Contracts scholars also include in the category of transaction costs the observation that information asymmetry between the parties at the front end may impede efficient contract terms. (19) We set these obstacles aside in this Article by assuming that the parties are symmetrically informed. This Part focuses principally on back-end transaction costs because they are less well understood among contract theorists. The reason is that they largely stem from the process of litigation, a distinct game played between the parties under relatively complex evidentiary and procedural rules. Our Article attempts to bring more detail and sophistication to the representation of back-end costs. Part II then shows how the parties can manipulate the tradeoff between back-end and front-end costs to improve the bang for their contracting-cost buck.

Before the parties can decide how much to invest in the back end, they must determine the expected net value of the incentive gain that they would secure with their back-end (enforcement) buck. This requires them to anticipate the course of their litigation and its outcome. Contracts scholars have focused on two back-end obstacles to efficiency: (1) the direct costs of enforcing contracts--namely the costs of communicating information to the court--and (2) the uncertainty and error costs of enforcement. Recognizing these obstacles, scholars postulate that parties avoid contract terms that are prohibitively costly for a court to verify or terms that are vague. (20) However, their predictions are at odds with commercial contracting practice that, for instance, frequently adopts vague terms such as "best efforts" or "commercial reasonableness." (21) One reason for this gap between theory and practice is that the scholarly conception of verification costs is based on a highly stylized understanding of the litigation process. In the following sections, we discuss the reasons why the current literature is at odds with empirical realities.

A. Back-End (Enforcement) Costs

1. Direct Costs

Contract theorists identify verification costs as one of the principal obstacles to complete contracts. They postulate that parties will not condition their contract obligations on factors that are not verifiable (that is, when the cost of verification exceeds a notional threshold). (22) For example, they assert that parties to an agency contract would not condition payment on effort because effort is nonverifiable. (23) In doing so, these scholars mischaracterize judicial enforcement as an investigatory rather than adversarial process. In particular, they neglect three important features in the judicial enforcement of contracts: (1) information comes to the court by way of self-interested parties bringing costly evidence to the court's attention; (2) the court makes its judgment based on a relative rather than absolute assessment of its confidence in factfinding ("preponderance of the evidence" or "balance of probabilities"); and (3) the parties have considerable influence, either by their contract or later agreement, on the course of the future litigation. In light of these factors, the verifiability of a contract obligation or contingency is context-specific and endogenous. Moreover, verification costs are likely to be substantially lower than economists implicitly assume, so that a contract might well try to regulate effort, for example.

Courts do not observe facts directly; rather, they make factual determinations by relying on proxies for the truth. The performance of a contractual obligation is proved or disproved by the presentation of evidence rather than by the court's direct observation. Suppose, for example, that a contract requires delivery of a widget that is exactly 0.0025 inches wide. The promisor's compliance with even this very precise contract term is not established directly by a court undertaking to measure the widget. Rather, compliance is proven indirectly by, for example, expert testimony on the width of the widget--testimony that is subject to cross-examination concerning the accuracy of the expert opinion. The same is true if the contract calls for a widget of merchantable quality. The court selects proxies for merchantable quality and then examines the evidence to determine whether or not those proxies are satisfied. The cost of proof therefore depends on what proxies are considered and what evidence is invoked to establish the presence or absence of the proxies.

Significant institutional forces and incentives constrain the costs of litigation, regardless of the substantive contract provisions. In the adversarial litigation system, the court chooses between the self-interested evidence presented by the parties. The parties present only the evidence that is in their respective self-interest, and the parties also bear most of the cost of their respective evidence production. Given the evidentiary and procedural rules of litigation, each party decides how much to invest in evidence production. They stop presenting when the marginal cost exceeds their marginal private benefit, which is a product of the probability of winning and the amount at stake. A significant decrease in the probative value of evidence, for instance, might therefore result in a relatively inexpensive trial. Moreover, the parties' evidentiary decisions are interactive, in the sense that the marginal benefit of one party's evidence is affected by the other's evidentiary strategy. One party's evidence may well discourage the other party from further investing in the litigation. (24)

In civil cases, such as contract disputes, courts make factual determinations with substantially less than complete confidence in their fact-finding. Indeed, unlike criminal cases in which the facts must establish guilt beyond a reasonable doubt, civil disputes are decided using a relative rather than an absolute standard: the preponderance of evidence or the balance of probabilities. Moreover, where proof is particularly difficult, trials may be abbreviated by several well-known procedural mechanisms. Even before the parties present their evidence, the court might award summary judgment to one party if the other is unable to show that there are genuinely contested issues of material fact. (25)

The factfinding process in litigation is governed by burdens of proof and presumptions that tend to curtail litigation costs. The burden of proof consists of two distinct burdens--the burden of production and the burden of persuasion--that carry distinct standards of proof. The party with the burden of production must produce sufficient evidence such that, in the eyes of the judge, a reasonable jury could infer the fact. (26) If that party fails to carry that burden, the court will order a directed verdict or judgment as a matter of law in favor of the other party. (27) Some or much of the cost of a full-blown trial might thereby be avoided. (28) The burden of persuasion follows if the burdens of production are met and both parties have presented all their evidence. The court instructs the jury that one party carries the burden of persuasion and that, unless this burden is met, the jury must return a verdict for the other party. In a civil case, such as an action for breach of contract, the burden is satisfied if the party with the burden establishes that the alleged fact is more likely than not to be true. This underscores the relative character of the adversarial process. One party's evidentiary production need not be any higher than that which is necessary to pass the burden threshold, given her opponent's evidence. At the same time, a party carrying the burden may retreat in the face of additional evidence presented by her opponent. (29)

Legal presumptions shift burdens from one party to the other and, in so doing, might further limit litigation costs. (30) Under a presumption, the satisfaction of a burden with respect to fact X satisfies the burden of production on fact Y, and it also shifts the burden to the other party to establish the nonexistence of fact Y or face a directed verdict against it. For example, suppose a shipper can show that it delivered goods to Carrier A in good condition and received them from Carrier B at the ultimate destination, but in defective condition. In an action brought against Carrier B, there is a presumption that the damage occurred while the goods were in the control of Carrier B. (31) Such presumptions are sometimes justified on the ground that fact Y is highly correlated with fact X or that the other party has superior knowledge about fact Y. (32)

Despite these various mechanisms that curtail the evidence that is presented at trial, litigation costs may be inefficiently high because of the litigants' incentives. Each party's investment is a function not only of its probability of winning, but also the amount at stake--for example, the damages award being sought by the plaintiff. At the time of the trial, the parties are engaged in splitting a fixed gain or loss with little, if any, prospective efficiency value. The value of the litigation outcome derives from its effect on ex ante incentives, which are of no interest to the parties at the time of trial. (33) The litigants continue to invest until the marginal cost of additional evidence exceeds the marginal increase in the expected litigation outcome, rather than in the improvement in incentives. From this perspective, the economists' concern with verification costs may be restated as the prospect that the parties will overinvest in litigation, relative to the gains in ex ante incentives. Even in this more refined frame, however, the concern with back-end costs is overstated because it ignores the ability of the court or the parties themselves to address these inefficient incentives.

Judges have self-interested motivations to abbreviate the duration and cost of trials. A judge's prestige and influence may well be enhanced by presiding over more rather than fewer cases, while holding her personal effort constant. And, within a case, the judge may reduce the demands on her time and effort by limiting the amount of evidence. In light of the public spotlight on litigation costs, some courts have enjoyed a positive reputation for putting in place mechanisms that speed trials. (34) The rules of procedure and evidence provide judges with tools for doing so. Accordingly, judges have discretion to constrain pretrial discovery, to accelerate trial dates, to limit the length of trial, and to exclude evidence. (35) Several commentators assert that rules of evidence and procedure are designed to drive a wedge between the lower cost of evidence supporting the truth and the higher cost of inaccurate (or fabricated) evidence. (36) This improves the efficiency of litigation in several ways. First, given this cost differential, Rubinfeld and Sappington suggest that the effort invested by each party in litigation may be a signal of the truth. (37) If the court can observe effort, litigation may yield a second-best equilibrium in light of the court's inability to verify the truth directly. The nonperforming defendants invest nothing in litigation and are found liable, and the performing defendants spend until their private marginal benefit of investment (in reducing their expected liability) equals the marginal cost of additional evidence. Despite judicial pronouncements to the contrary, courts often do draw inferences in civil cases from the failure of a party to present evidence that might exonerate it. This separation between nonperforming and performing defendants ameliorates the concern with excessive litigation cost.

Second, if parties can reduce their evidentiary costs by performing their contractual obligations, this saving may have an ex ante incentive effect by inducing performance. (38) Therefore, contracting parties may wish to contract over factors that might entail prospectively high litigation costs if there is a significant discrepancy between the evidentiary cost that would be incurred by the performing party (who would tell the truth at trial) and that of the nonperforming party (who would lie). The promisee's evidentiary costs, in contrast, have no beneficial effect on performance incentives other than by raising the likelihood of a finding of liability. (39) Nevertheless, as before, our point is simply to suggest that the focus on verification costs alone is far too simplistic to explain contract design.

Finally, the parties themselves may further reduce litigation costs by consent. They can do so narrowly, by stipulating facts or agreeing to limited discovery, or more broadly, by settling the case altogether. Indeed, the prospect of settlement provides another illustration of how the concern with verification costs is misleading. Settlement is more likely, all other things being equal, the higher the anticipated litigation costs. For any given difference in the parties' expectation regarding the likely judgment, the likelihood of settlement increases with the expected aggregate cost of trial. (40)

2. Uncertainty and Error Costs

In light of the various constraints on evidence production and the modest confidence threshold for judicial factfinding (balance of probabilities), there is uncertainty and the prospect of judicial error in contract enforcement. As some contracts scholars argue, this uncertainty can undermine performance incentives. (41) The precise effect on incentives depends, however, on the nature of the uncertainty, which in turn is a function of context-specific variables. (42) In particular, the prospect of legal error is more likely to undermine incentives if the factfinding lies within a discrete rather than continuous set of possible alternatives. Consider a contract that requires an agent to make a specific investment in a venture. Suppose that the agent's share of profits from the venture and her reputational concerns are such that her self-interested strategy would be to invest $30, which the parties know will be suboptimal. The contract, therefore, requires the agent to invest $60. The parties agree to a clause requiring the agent to pay liquidated damages if the agent fails to invest exactly as promised. However, the actual investment of the agent is not verifiable: The court observes only a noisy signal of her investment and therefore may assess it incorrectly. Under these conditions, if the agent's choice is binary (she can invest either $30 or $60), then the risk of error will undermine incentives by raising the possibilities that an agent investing $60 may nevertheless be found liable (Type I error) and that an agent investing $30 may not (Type II error). But as long as the probability of liability is higher if the seller breaches by investing $30 than if she invests $60, the enforcement of the obligation will improve her incentive somewhat. The important question is whether the gain warrants the enforcement cost.

Suppose now that the contract specifies performance that lies on a continuous set of decision points. In this case, the effect on incentives is less clear and depends on the distribution of error. (43) The risk of inaccurate assessment may ameliorate the degree of underinvestment in effort (a good thing) or it may overshoot and cause excessive effort (which may be either a good or a bad thing relative to the agent's contractual incentives in the absence of the contract term requiring additional effort). Assume again that the contract requires the agent to invest $60 and that the agent would invest $30 in the absence of the investment clause in the contract. Here, however, the parties do not agree to liquidated damages if the investment falls below $60. Rather, under the default of expectation damages, the agent's liability increases continuously the more the investment falls below $60. The parties do not have sufficient incentive to produce evidence at trial that leads the court to pinpoint the agent's effort in fact (for example, because the marginal cost of evidence rises steeply beyond some point). This risk of legal error creates a wedge between the social and private return from investment above $30. The social return in this case is the value created within the venture, while the private return to the agent is the incremental reduction in expected liability caused by additional investment. The expected liability would fall because of the lower probability that the court would find a breach.

Unless the court declines to adjudicate this dispute, it will arrive at an assessment of the agent's investment. At the time the agent makes her investment decision, however, the court's prospective determination is a probability distribution. If the court is extremely uninformed, the distribution may be uniform across effort levels regardless of the actual effort expended. The agent would then enjoy no private return (in the way of lower expected liability) from increasing her investment and it will remain at $30. If, however, the distribution is normal and peaks at the agent's actual level, then the agent would reduce her expected liability by moving from $30 to $60. In fact, Craswell and Calfee demonstrate that where the variance is sufficiently low, the agent may overinvest in order to further reduce expected liability. (44) For example, by expending $61, the agent may achieve a safety margin that reduces her expected liability by an amount that justifies the incremental investment. Although this may be privately profitable for the agent, it would not be jointly optimal: The extra investment is not justified by the increase in the contracting surplus, only by the expected reduction in the agent's liability. (45) Thus, the clause requiring the agent to invest $60 might correct the incentive of the seller to underinvest, but it might also overshoot its target and lead to overinvestment. Some overinvestment may nevertheless yield a reduction in efficiency losses compared to the underinvestment that would occur in the absence of the investment clause. This improvement may justify the clause, despite the uncertain enforcement. (46)

We will return to the consequences of uncertainty and legal error in our discussion of the choice between precise and vague terms in Part II. For now, we note that the effect on incentives is ambiguous: It may be in either direction and of varying magnitude. The impact is context-specific, thus suggesting that contracting parties may differ in their willingness to face uncertainty in legal enforcement. (47) Moreover, there are likely measures that the parties can take in their contract design to predetermine the distribution of factfinding, (48) but we leave that inquiry for future research.

B. Tradeoffs Between Front-End and Back-End Costs

Contracts scholars have a split view of vague terms, depending on their perspective. When contracts scholarship is concerned with front-end (transaction) costs, such as the cost of negotiating and writing contracts, vague terms reduce these costs by letting the enforcing court complete the contract. (49) This argument, however, assumes costless enforcement. When contracts scholarship is concerned with back-end costs, including verification costs and uncertainty, the authors prefer precise to vague terms. They argue that courts should refrain from enforcing vague terms that entail prohibitively high verification costs. (50) These arguments tend to set aside the front-end costs of precise provisions.

In fact, however, contracts include both precise and vague terms, and the courts seem to actively interpret and enforce vague terms. (51) Commercial contracts regularly invoke factors such as "best efforts," "reasonable expenses," and "reasonable withholding of consent." (52) Not only are explicit "best efforts" obligations common, they are also the subject of extended negotiations, including negotiation over seemingly minor linguistic variations. Indeed, many contracts reflect a highly nuanced approach to the specification of vague clauses. For example, "best efforts" may be replaced by "commercially reasonable efforts," "reasonable efforts," or "reasonable best efforts." (53) While some courts interpret "best efforts" as the equivalent of good faith, (54) others impose a higher standard of reasonable diligence, (55) and some even require the level of effort that would be exerted by a similarly situated integrated firm. (56)

Contracts scholarship therefore needs a theory to explain the common use of both vague and precise terms, as well as to predict when each type is more likely to be used. This Article addresses these questions by examining the important but neglected tradeoff between front-end and back-end costs. The resolution of this tradeoff in each contracting instance determines the parties' optimal choice between precise and vague terms. For those readers who think in terms of indifference curves, we would draw a graph with front-end and back-end costs on each axis and iso-incentive curves, each of which trades off the costs while maintaining the same level of contract completeness or incentive efficiency. The point on the curve that hits the lowest budget line is the optimal combination of front-end and back-end investments that will achieve the incentive efficiency of the curve. Conversely, the point on the curve that hits the lowest budget line represents the cheapest combination of front-end and back-end costs for the curve's incentive efficiency.

The tradeoff between front-end and back-end costs is never an all-or-nothing choice. The parties will make some effort to describe their obligations on the front end; indeed, the courts require this as a precondition to an enforceable contract and will decline enforcement on account of excessive vagueness. (57) At the other extreme, it is prohibitively costly to draft a contract that entails no back-end costs and creates no enforcement uncertainty. Therefore, it is analytically helpful to invoke a concept that represents what is being traded between the front and back ends (for example, as one moves along the iso-incentive curve described above). We draw on the fact, mentioned earlier, that courts do not directly observe the materialization of contingencies or the performance of obligations, but instead rely on evidence or proxies. Parties can constrain the space from which these proxies may be drawn in litigation by agreeing to more or less precise terms. The more vague the term, the broader the space, and the more work the parties leave for the back end. Conversely, the parties can invest in defining proxies at the front end and identifying them through precise terms.

Contracts essentially provide for pairs of contingencies and performance obligations. For example, when X occurs, the promisor must pay $Y. For our expositional purposes, the parties might define X at any of three levels. First, X might be the production of a specific bit of evidence, such as a signed document or testimony by a specific witness. (58) Second, X may be a relatively specific event, such as the delivery of a widget with a specified weight. In this category, the parties delegate to the court the determination of which bits of evidence are sufficient to satisfy X and trigger the promisor's payment obligation. Third, X may be a vague term, such as the delivery of a widget in excellent or merchantable condition. In this category, the court must determine not only what evidence is sufficient to establish the weight of the widget, but also the degree to which weight is relevant in the determination of whether the standard has been met. For convenience, we refer to this latter determination as the choice of proxy for the vague term or standard. Although there are factual bases for choosing among proxies, the selection is generally regarded as a question of law for the judge. In some cases, the proxy choice becomes fixed as a legal default rule. This is the case with expectation damages, for example, where market damages are regarded as the default mechanism for establishing the promisor's contractual expectancy in the case of goods or services traded in established markets. (59) In other cases, the judge identifies the appropriate proxies for the jury. For example, in Empire Gas Corp. v. American Bakeries Co., (60) Judge Posner ruled that the trial judge had failed to give appropriate instructions to the jury when he did not specify what evidence would support a finding of bad faith. (61)

Thus, precise and vague contract terms (or rules and standards) may be distinguished by the manner in which proxies for a particular contingency or obligation are chosen. The parties may either choose the proxy directly by a rule in their contract or delegate the choice to the court through a contractual standard. In either case, the court determines whether the relevant proxies have been satisfied by screening bits of evidence presented by the parties. (62) If the proxy is determined by contract, the parties incur front-end (transaction) costs. If the parties agree to a vague term (standard), they accept higher expected back-end (enforcement) costs in return for lower front-end costs.

As an example of a court-selected proxy under a vague term, consider the familiar contracts case of Bloor v. Falstaff Brewing Corp. (63) Falstaff purchased most of the distribution network and related assets of a brewer named Ballantine. Part of the compensation consisted of royalty payments of $ 0.50 per barrel of beer sold during the six years following the sale. The parties designed this component of the sale price to reflect the value of the Ballantine distribution assets to Falstaff. (64) However, the royalty threatened to induce suboptimal effort by Falstaff by effectively taxing the marginal product from sales of beer. (65) To deal with this problem of underinvestment, the parties included a provision requiring Falstaff to "'use its best efforts to promote and maintain a high volume of sales.'" (66) When the seller sued and claimed that Falstaff had breached its "best efforts" obligation, the trial judge faced the dual tasks of verifying whether the defendant had breached and determining the appropriate measure of damages. Judge Brieant chose a market proxy for the performance of best efforts: the sales of two integrated firms (Rheingold and Schaefer) that both produced and distributed the same product and that were roughly comparable in size and locale to the contracting parties. (67) The integrated firms provided an appropriate benchmark for efficient best efforts because they did not suffer from the skewed incentives of sharing revenues with separate organizations and because the relevant sales data was readily available. The parties no doubt incurred litigation costs in proposing and arguing over the appropriate proxy. Nevertheless, they avoided front-end costs by contracting for "best efforts" instead of specifying proxies at that stage.

Other parties shift contracting costs and proxy selection to the front end. Many commercial contracts include explicit benchmarks similar to the ones the court adopted in Bloor. (68) Franchisors promote sales efforts by their franchisees by requiring them to maintain sales volume comparable to other similarly situated franchisees or franchisor-owned outlets. (69) These proxies are established by evidentiary bits at a back-end cost that is significantly lower than if the parties were to argue in court about the appropriate proxy. (70) Thus, some contracting parties elect to incur front-end costs in specifying proxies by contract while others leave the proxy choice to the back-end process. More generally, the choice is not simply between specifying proxies or not. Rather, contracts define a broader or narrower space within which the court selects proxies. (71) The size of the space determines the discretion over proxy choice that is assigned to the court instead of the parties, as well as the extent to which the proxies are chosen at trial rather than the time of the contract. Accordingly, the determination of proxy choice implicitly allocates costs between the front end and back end of the contracting process. We turn in the next Part to examining more closely the factors governing the parties' strategies in making this allocation.

II. THE CHOICE BETWEEN RULES AND STANDARDS IN CONTRACTS

In this Part, we explore how parties choose their mix of precise and vague contract terms (or rules and standards) to optimize the selection of efficient evidentiary proxies over two dimensions: when the choice of proxy is made and who makes the choice. We describe the means by which the parties define the domain or space within which the court selects proxies at litigation. A precise term defines a very narrow space--at the limit, a single proxy. It therefore entails larger front-end transaction costs, but lower back-end enforcement costs than a vague term that leaves the court with a broader space. Contracts that combine rules and standards in defining a contingency or obligation offer additional flexibility in setting boundaries for the court's discretion, but they respond to the same tradeoff.

A. Efficient Proxies and Efficient Choosers: Rules or Standards in Contracts

1. Determinants of an Efficient Proxy

It is helpful to describe the features of an efficient proxy before turning to examine how contracting parties would choose between contracting directly on proxies and delegating the choice of proxies to the court. An efficient proxy provides a greater incentive bang for the buck incurred in expected litigation costs than its alternatives (holding the costs of choosing constant). Incentive benefits can be achieved very inexpensively if the enforcement threat alone induces performance, so that litigation is in fact avoided. The same is true if the contract is renegotiated or the disputes are settled before going to court, which is a likely outcome when the parties are symmetrically informed. In this Article, however, we are contemplating a worse scenario in which litigation (or verification) costs are in fact incurred. Even then, they may be efficient in yielding a positive bang for the buck.

Consider the following simple example: Suppose a seller and buyer enter into a contract for the sale of a widget that may be produced either with an ordinary veneer or a premium polished veneer. The buyer values the widget at $1000 with the premium veneer and $600 with the ordinary veneer. The seller's corresponding production costs are $700 for the premium veneer and $500 for the ordinary veneer. Thus, the surplus from trade is $300 if the widget has a premium veneer and $100 if it does not. But the seller would produce an ordinary widget unless otherwise obligated under the contract. The parties consider whether to contract for the premium veneer, in which case they would provide for liquidated damages of $400, a sum equal to the expectation loss of the buyer if the seller uses ordinary veneer. Assume that if the buyer could be confident of the seller's performance, the parties would split the $300 surplus and set a contract price of $850 for the premium widget.

Enforcement, however, might be costly and uncertain. Specifically, the condition of the veneer might be different at trial than at the time of delivery because of the buyer's use of the widget. The parties are likely to offer conflicting expert testimony at significant combined cost. We continue to exclude the possibility of dispute settlement or renegotiation, and therefore assume that all contracts are litigated. To illustrate the bang for the buck concept, suppose that the parties would invest a total of $X in litigation and that the court would thereby detect without error whether the veneer was premium or ordinary at the time of delivery. Would it be efficient for the parties to contract for premium veneer? The gain in joint surplus from the premium veneer is $200 ($300 surplus from premium veneer less $100 surplus from ordinary veneer). So, as long as X is less than $200, the premium veneer term is efficient: It yields a net bang for the buck by inducing the seller to perform. And the seller does perform in this example, because the incremental production cost of premium veneer is $200, which is less than the seller's $400 damages liability (not to mention his litigation costs).

To incorporate a further element of uncertainty, suppose instead that the parties invest a combined $100 in evidence, but the probability that the seller will be found liable is 75% if she has produced an ordinary widget. The $100 is worthwhile if it induces the higher level of performance. As long as the probability of liability is lower than 25% when she produces the premium widget, the seller has the incentive to perform the contract. The seller's expected liability is $300 (0.75 x 400) if she makes the ordinary widget and less than $100 (0.25 x 400) if she makes the premium quality. Thus, the seller will invest the $200 to make the premium widget and avoid litigation, so that the contract achieves an incentive gain greater than its enforcement cost.

2. Determinants of an Efficient Chooser

The parties in the foregoing example have the choice between specifying the obligation to provide premium veneer at the time of contracting or contracting for a widget of "high quality" (or similar vague term) under the expectation that the court will require premium veneer if appropriate in the circumstances. By assigning the proxy selection to the better chooser, the parties can either reduce their contracting costs or improve the efficiency of the proxy, or both. The best information as to proxy choice is held by the parties themselves after the resolution of uncertainty, but the parties have divergent private interests in the choice of proxies at trial. In selecting a chooser, therefore, the parties have only two options: The choice of proxies will be made either at the time of the contract by the parties, who enjoy private information, or after the resolution of uncertainty by the court, which enjoys the benefit of hindsight. The superior decisionmaker is a function of the relative incentives and information of the parties and courts; rarely are either the parties or the court ideally situated. Barring significant asymmetries in sophistication and information, the parties at the time of contracting should have superior incentives; after all, they share in the benefits of efficient contracting. A court presumably has no bias in favor of one party over another in a dispute, but it also does not have much of a stake in the efficient ex ante outcome. As noted earlier, the court may have incentives to contain litigation costs, but its ex post perspective is likely to weigh litigation costs against accuracy in fact-finding, rather than against ex ante efficiency.

The comparison of informational advantages is a closer call, and this is what leads to the diversity in the use of precise and vague contract terms. At any given time, the parties have information superior to that which they can communicate to the court. Yet, as we have seen, the selection of the proxy chooser is between the parties at the time of the contract and the court at the later time of litigation. The efficiency of proxies (their incentive bang and enforcement buck) is often determined by the surrounding circumstances. Therefore, front-end proxy choice must contemplate the operation of the proxy in various possible future states of the world either by identifying pairs of proxies and states or by more crudely lumping states in groups. In contrast, back-end proxy selection can be fine-tuned to the materialized state, albeit at the cost of identifying which state has in fact occurred. (72) In other words, the court has the benefit of hindsight and can focus its attention on the materialized state alone.

The parties may view the court's hindsight as an advantage or disadvantage depending on how much uncertainty has been resolved by the time contract performance is due. Where the enforcement cost of proxies varies with the materialized state of the world (for example, the availability of a market indicator), the court has a systematic advantage. However, the incentive benefit of a proxy depends on whether the relevant contingencies have occurred by the time that performance is due, not at the time of trial. Thus, if uncertainty is resolved before performance is due, the court's proxy selection may be predictable so as to influence the promisor's behavior. Conversely, if the promisor must perform before much uncertainty has dissipated-for example, by investing early in the relationship-then the court's hindsight may in fact be a liability.

In sum, the parties will choose a specific proxy when the parties' private information is more important than the effect of contingencies on the choice of proxy. When the efficient proxies are highly state-contingent and less dependent on private information of the parties, the parties will be more inclined to use standards to delegate proxy choice to the courts, particularly if uncertainty is expected to resolve itself by the time the relevant performance is due.

The case of Eastern Air Lines, Inc. v. Gulf Oil Corp. (73) provides an instructive example. The parties entered into a long-term contract for the sale of jet fuel at designated locations. They wished to set a price for the jet fuel in order to allocate the risk of exogenous changes in the input price of crude oil to Eastern Air Lines and the risk of fluctuations in production cost to Gulf. They selected a contract proxy that adjusted the contract price according to an easily verifiable indicator of crude oil price-West Texas Sour crude "as listed ... in Platts Oilgram Service." (74) Subsequently, as a result of governmental regulation following the oil crisis in the 1970s, Platt's Oilgram no longer tracked the market price of crude oil, and this proxy failed. The court declined to choose a substitute proxy. The parties might have anticipated the failure of the indicator by stating explicitly in the contract that the price either would be adjusted to the price of crude oil (a standard) or that it would be tied to Platt's or "any other appropriate index." In general, a contract might adopt a blended strategy by providing for a specific proxy and delegating to the court the choice of a replacement if the specific proxy should fail. As we discuss in the next Section, the inclusion of such a standard is an invitation to the court to choose a new proxy. (75)

The classic contrast to the conservative approach of the court in Eastern Air Lines is the decision in Aluminum Co. of America v. Essex Group, Inc., (76) in which the court reformed the parties' price adjustment proxy in the absence of an explicit delegation by the parties. The usual critique of this opinion is not that parties would never choose to delegate proxy choice to the courts, but rather that there was no evidence in the contract or otherwise of the parties' intention to do so here. Indeed, the fact that they had invested a great deal of resources up front to provide a specific proxy might have been evidence to the contrary. Alcoa is also unique because the contract reformation was at the court's initiative; neither party suggested an alternative proxy even at the time of trial. (77)

Parties thus delegate proxy choice to the courts through the language of the contractual standard and its combination with the precise rules in the contract. In some cases, the parties might rely exclusively on precise rules and forego standards. In hindsight, a court may be tempted in these cases (perhaps with the encouragement of one of the parties) to see gaps between the discrete rules and therefore to read into the contract implied standards, such as "best efforts" or "commercial reasonableness." Yet commercial parties can include standards in their contract at relatively low cost, and they enjoy superior knowledge of the context of their contractual relationship to determine the optimal allocation of proxy choice. Consequently, the courts are wise to interpret the absence of vague standards in commercial contracts as instructions from the parties to abstain from proxy choice and to limit their construction to the precise terms of the contract. (78)

3. The Benefits and Costs of Uncertainty in the Enforcement of Standards

In Part I, we identified the potential effects of uncertainty and the risk of error in factfinding on performance incentives. Where the promisor's decisions are binary in nature, Type I and Type II errors undermine her incentives to perform. The prospect of legal error is compounded when a court enforces a vague term instead of a precise provision because the court's task is broader: It must choose proxies as well as the evidentiary bits that support each proxy. Indeed, some contracts scholars indicate that precise terms should be preferred to vague terms for this reason. (79) The danger is that the promisor may exploit the uncertainty as to the correct proxy by shirking her obligations and then proposing an alternative proxy. To the degree that the court might adopt the promisor's opportunistic suggestion, the vague term compromises the efficiency of her incentives.

This opportunity for exploitation is constrained, however, by the nature of the adversarial enforcement process and the consequent uncertainty in proxy selection. Both parties propose proxies to the court, and there is no a priori reason why promisors would be systematically more likely to prevail in litigation to determine which proxy should be selected. Moreover, the important question is not whether vague terms are perfect, but whether there are conditions under which they are superior to a contract with a corresponding precise obligation or even no obligation at all. Suppose, for example, that the court has superior information at the time of trial but there is uncertainty as to which proxy it will choose between two alternatives, such as the relevance of weight and of color to the merchantability of a widget. Given the court's superior information, the parties can expect that one or both of the proxies will be less noisy under the circumstances than the one that the parties would pick ex ante. Therefore, even when discounted by the relevant probabilities of judicial choice, either alternative would improve performance incentives over a certain, but inferior, specific contract proxy. (80)

A further virtue of delegating the proxy choice to the court via a vague standard is that the uncertainty as to which proxy will be selected might help to reduce the incentives of promisors to game precise rules once an adverse risk has materialized (a problem familiar to the design of tax rules). (81) The uncertainty in judicial factfinding discussed in Subsection I.A.2 concerned the court's error in determining the dollar amount of investment by the promisor, a unidimensional variable. In contrast, vague terms are often used when the performance in question is multidimensional, such as effort, and uncertainty raises different considerations in this case. Compare the incentives of an agent faced with a specific proxy for effort (in the form of a contract rule) and another agent whose behavior is governed by a broad standard of effort. The first agent has the incentive to direct her attention to satisfying the proxy alone and to ignore all other dimensions of the desired performance. (82) When faced with a standard, however, the agent has many proxies that might bear probabilistically on litigation outcomes. Her optimal strategy may therefore be to focus on effort rather than on any single proxy, thereby improving her position vis-a-vis all proxies.

As an illustration of when vague terms are superior to precise terms, consider the following example offered by Canice Prendergast:</p> <pre> It is difficult to imagine an occupation for which there are more measures of performance [than baseball]. Despite this, it is not common for players to have contracts where pay is directly related to specific performance measures. Part of the reason for this is that teams are reluctant to offer a contract that rewards a player for home runs, say, because the player may have an incentive to hit home runs even when it is not in the interest of the team for him to do so. By contrast, the more common cases where players are offered explicit bonuses are for aggregate measures of performance, such as making the All Star Team or being the league's Most Valuable Player. Since these are more holistic measures of performance, they suffer less from the multi-tasking dilemma. (83) </pre> <p>The parties to such a contract are using a standard in order to delegate to a third party the evaluation of the player's performance. Part of the motivation is the challenge of specifying all the relevant facets of a player's contribution to his team. But as discussed in the excerpt above, a distinct difficulty is the problem of specifying the desirable state-contingent proxies as they differ among possible future states of the world. Hitting or fielding may be relatively more important depending on the course of the season. A retrospective determination of performance can economize on having to specify state-contingent performance measures and compensate for the parties' bounded rationality. The baseball contract contemplated above delegates the proxy choice to experts in the industry. (84) If experts are not available, however, one can imagine that even a delegation to a court may be superior to the parties' attempt to list the relevant proxies ex ante. Moreover, the passage quoted above also suggests that aggregate measures mitigate multitasking problems. Our argument is that vague standards can achieve the same effect probabilistically, as long as the range of individual proxies is correlated with the desired performance.

A similar, more cynical, argument may be made about the agent's incentives under a specific proxy. (85) As an alternative to performance, an agent has the option to invest in persuading the court that she has satisfied a specific proxy. For example, she may tamper with a testing mechanism or misrepresent accounting results. (86) In light of opportunities to manipulate or manage evidence, contractual sanctions for nonperformance might increase the incentives to perform, but they may also raise the payoffs from investing in evidence management. When the cost of successfully fabricating evidence is lower than the cost of performance, the agent has the incentive to invest in socially wasteful evidence management rather than in performance. Given that evidence management is socially unproductive, the parties have a joint interest at the time of contracting to deter this activity by delegating the proxy choice to a court under a vague standard. (87) If the proxy is uncertain because it is within the discretion of a future court, the uncertainty discourages evidence management by blurring the target. (88) The agent must discount the benefit from evidence investment with respect to any given proxy by the probability that the court will choose that proxy. As a result, the expected benefit from evidence management with respect to that proxy is lower under a standard than a rule. (89) Thus, within some margin, the agent may be better off simply performing under the standard, given that performance is correlated with all the possible proxies. The benefit from shifting incentives from evidence management to performance may, however, be offset somewhat by a factor we have largely excluded from our analysis: Under standards, promisors may invest in predicting how a future court will interpret their vague obligation. (90)

B. Maxims of Interpretation and the Scope of the Proxy Space

The choice between party-selected proxies (rules) and court-selected proxies (standards) is not an exclusive binary choice based on relative informational advantages and the cost efficiency of proxy choice. The parties can, and regularly do, include both types in their contract. The combination of vague and precise terms is widely used in commercial contracting. One conventional explanation for vague terms in this context is that they act as "catch-alls" that compensate for the underinclusiveness of precise terms. Yet this raises the question of why parties do not simply agree to a broad standard alone (the catchall without the precise terms) that invites the court to choose the proxies invoked by the contract rules. In the discussion that follows, we reframe this explanation in terms of the efficient delegation of proxy choice. The parties may choose to give the court a defined space within which to select some proxies, while specifying other proxies in contract rules. Although precise and vague terms provide useful benchmarks for narrow and broad spaces, the parties have the range of intermediate options to choose from.

Consider the court's choice of proxy under a contract containing a combination of precise rules and vague standards. In the adversarial system, the choice of proxy is likely to be a choice between the proxies offered by the litigating parties. The court's task is (1) to ensure that a given proxy falls within the space contemplated by the parties in their agreement and (2) within this space, to choose the appropriate proxy or proxies. The former is a matter of contract interpretation and, in this task, the courts are guided by interpretation maxims. These principles are followed with sufficient regularity that the parties can anticipate them at the time of contracting. We first outline the most relevant interpretive principles and then demonstrate how they are reflected in the patterns of rules and standards in commercial contracts.

As a general matter, the canons and maxims of contract interpretation do not depend on a finding that a contract term is ambiguous. Rather, they are used both in determining what meanings are reasonably possible as well as in choosing among divergent interpretations. (91) These maxims first instruct the court to view the agreement ex ante-that is, to put itself in the position the parties occupied at the time of contracting, (92) and to interpret provisions in light of the purpose of the contract. (93) Consistent with the notion of purposive interpretation, a contract must be read as a whole, and each part must be interpreted in light of all provisions. (94)

For our purposes, it is useful to examine the interpretive effect of the choice of combined rules and standards, as compared to stand-alone rules or standards. Three well-known maxims are particularly relevant: ejusdem generis, noscitur a sociis, and expressio unius est exclusio alterius. If a contract through its exclusive use of precise terms provides only for specific proxies, the maxims of interpretation caution the court against considering other proxies at the time of trial. Under the expressio unius maxim, the expression in the contract of one or more things of a class implies exclusion of all that is not expressed. (95) The inference is that all omissions should be understood as exclusions, and the specification of particular items impliedly excludes other items relating to the same general matter. (96) Moreover, when a contract provides that a thing should be done in a certain way, it is presumed to be exclusive. (97)

A standard on its own gives the court a relatively large space within which to choose proxies. When the parties combine standards and rules that relate to the same subject matter, the ejusdem generis canon applies, whether the general language is preceded or followed by the enumerated precise terms. The meaning of the general language is then limited to matters similar in kind or classification to the enumerated precise terms. (98) But the parties must be careful when using combinations of standards and rules to use words that signal to the court a desire to have new proxies created at trial. In a recent case, a lease contract provided that the lessor could terminate "for good cause," and this general language was then followed by enumerated items such as nonpayment of rent, serious or repeated damage to the premises, or the creation of physical hazards. The appeals court held that the general phrase "for good cause" did not include other violations of the lease, such as keeping a dog on the property. (99) Contracting parties can avoid a restrictive interpretation under the ejusdem generis rule by providing that the general language includes but is not limited to the precise enumerated items that either precede or follow it. (100)

Under noscitur a sociis, which means "it is known by its associates," the court determines the meaning of vague phrases by reference to their relationship with other associated words and phrases. Under this maxim, the coupling of words or phrases indicates that they should be understood in the same general sense. (101) As noted above, when the parties provide for specific proxies but no standard, expressio unius might prevent the court from reading a general purpose. Moreover, even under noscitur a sociis, a series of specific proxies may not have enough in common to indicate to the court the general objective that associates them. But when a broad standard is added to a listing of precise terms, it communicates the underlying objective and helps the court interpret the precise terms in light of the general purpose. The noscitur maxim requires that the general and the specific words must be considered together in determining the contract's meaning, so as to "giv[e] effect to both the particular and the general words." (102) Thus, the general term informs the interpretation of the specific proxies as well and might allow the court to fine-tune a specific proxy in light of its information advantage in hindsight.

A contract standard thus presents the court with two tasks. The first is to define the space for proxies allowed by the standard, in light also of the specific proxies specified in rules of the contract. This application of the interpretative maxims is a question of law. (103) The second is to choose the most appropriate proxy, or set of proxies, within that space. The court will weigh the incentive gains from the proxy, and the verification costs. The goal, as we have previously noted, is to find the proxy with the biggest incentive bang for enforcement buck. Both the bang and the buck are likely to depend somewhat on extrinsic facts. At least with respect to evidentiary costs, however, the judge would have an advantage over the jury in comparing the verification costs of alternative proxies.

C. Rules-Standards Combinations

In this Section, we provide and explain various common illustrations in which the contract's use of combinations of precise and vague terms can guide the court's future interpretation of the standard itself, as well as the accompanying rules. (104)

1. Acceleration Rights in Loan Agreements

Loan agreements provide a useful example of an effective combination of rules with a standard. The lender typically is entitled to accelerate the maturity of the loan and enforce collection of the principal and accrued interest upon an event of default. Failure to make a scheduled payment is an event of default, but so are the violations of specific covenants such as the debtor's promise to maintain insurance on important assets or to refrain from issuing future secured debt. In addition, many agreements provide that the lender may accelerate the loan if it deems itself insecure or believes in good faith that the prospect of repayment is impaired. (105) These acceleration rights are designed to permit the lender to exit upon evidence of borrower misbehavior or a higher risk of such misbehavior. At the same time, the parties wish to limit the ability of the lender to trigger default for ulterior purposes, such as calling back and relending the funds at higher market rates of interest.

The parties find it desirable to list specific proxies for inefficient debtor behavior, but front-end costs prevent them from including a comprehensive list. Therefore, they agree to a vague good faith standard that would catch the residual behavior not covered by the specific covenants. Why then do the parties not cover all suspect behavior with the insecurity standard alone? After all, the specific concerns of the failure to insure or the issuance of more debt all fall within the scope of events that would impair the prospect of repayment. The reason is that the parties wish to contain the proxy-choosing discretion of the court under the vague standard.

In this insecurity standard (as well as elsewhere in commercial law), "good faith" is interpreted by the law as meaning honesty in fact and the observance of reasonable commercial standards of fair dealing. (106) We have suggested that the parties agree to such a standard when they wish to harness the benefit of a court's hindsight and to address the risk that the debtor will game specific events of default. It is tempting to argue, nonetheless, that this vague standard of good faith--standing alone-is--simply not verifiable or is too uncertain. (107) Loan agreements, however, combine the vague standard with specific proxies, such as the promise to insure or to refrain from additional indebtedness, that are not significantly context-dependent and do not benefit from the ex post information advantage of a court. Relying on the application of standard maxims of interpretation, the parties combine their description of the standard with precise terms so as to define the space within which the court can choose proxies for good faith ex post. Courts will select proxies under the good faith standard only to the extent that they are similar in kind or classification to the enumerated precise terms. (108) Not only do the precise rules help to define the meaning of good faith insecurity, but the standard also aids in interpreting the meaning of the precise terms. When a broad standard is added to a listing of precise terms, it communicates the underlying objective and helps the court interpret the precise terms in light of the general purpose. (109)

2. Franchisee Obligations

Similar combinations of rules and standards are commonly found in franchise and distributorship contracts. These contracts typically provide that the agent both satisfy specific requirements and generally exercise best efforts. (110) The Taco Bell franchise contract is a good illustration. It provides that "It] he Franchisee shall devote his or her full time, best efforts and constant personal attention to the day to day operation of the Restaurant" and "[i]n addition, and without limiting the generality of the foregoing ... the Franchisee shall ... [d]iligently promote and make every reasonable effort to increase the business of the Restaurant." (111) The same section states that the franchisee may not have any financial stake or contractual relationship with any similar business (a noncompetition covenant). (112) The agreement also requires that the franchisee attend a training course and refresher courses offered by the franchisor; comply with the methods, techniques, and material taught at these courses; and instruct employees in the same material. (113) The franchisee must keep the restaurant open for the business hours specified in the company manual. (114) And, as a final illustration, the agreement requires the franchisee to maintain and repair the restaurant, including signage and landscaping. (115)

The performance obligations in these franchise and distributorship agreements address two distinct incentive problems. The first stems from the distortion in incentives caused by the sharing of the profits of the franchise outlet. We noted this effect in our earlier discussion of the court's opinion in Bloor v. Falstaff Brewing. In the Taco Bell Agreement, the monthly franchise fee is a percentage (5.5%) of gross restaurant sales. (116) The franchisee must deliver annual reports to the franchisor that are prepared in accordance with specified accounting standards and accompanied by the signed opinion of a certified public accountant. (117) The combination of a "best efforts" standard and associated precise terms is intended to address the distortion of incentives caused by this marginal tax on receipts.

The second incentive concern addressed by a "best efforts" clause is that, despite the tax on sales, the franchisee has an incentive to take actions that would raise its own profits but impair the value of the Taco Bell trademark and reputation (a cost that the franchisee externalizes to the franchisor and other franchisees). The agreement appears to address this concern within the "best efforts" provision through qualifying language stating that "without limiting the generality of the foregoing [best efforts]" the franchisee shall operate the restaurant "in a clean, safe and orderly manner, providing courteous, first-class service to the public." (118) Later, the agreement provides that the franchisee must also sell only products authorized in the company manual, and it must prevent the use of the restaurant for any immoral or illegal purpose or for any other use not expressly authorized in the agreement or in the company manual. (119)

Both incentive concerns are thus addressed in this contract by combining a vague "best efforts" standard with precise terms. The combination confines the domain of best efforts to those alternative proxies offered by the parties ex post that similarly improve the franchisee's incentives to invest and to protect the trademark. Alternatively, the franchisee is protected against an opportunistic claim by the franchisor that nonperformance of any specific covenant justifies termination of the franchise. The franchisee is free to offer alternative proxies to show that the broad concerns with protecting investment and reputation have, in fact, been met.

3. Force Majeure and Liquidated Damages

Force majeure clauses typically provide that performance is excused in the event of specific contingencies (such as war, labor strikes, supply shortages, and government regulation that hinders performance). But these clauses also identify excusing contingencies that fall within a vaguely stated category of factors beyond the control of the parties. (120) The combination serves the dual purposes noted in earlier examples. The domain of the vague standard of excuse for factors beyond the control of the parties is confined by the ejusdem generis canon to excuses that are similar in kind to the enumerated contingencies. Under noscitur a sociis, however, the broad standard of events beyond the parties' control communicates the underlying objective and would help a court interpret the application of a particular contingency, such as a labor strike.

Liquidated damages clauses similarly provide for a calculation of damages based on a laundry list of specific market factors together with a general reference to "any similar valuation." For instance, a recent gas and power supply agreement provided that liquidated damages should be determined by comparing the contract price to the relevant market prices either quoted by a bona fide third-party offer or that were reasonably expected to be available in the market under a replacement contract. To ascertain the market prices of a replacement contract, the contract permitted the promisee to consider, "among other valuations, any or all of the settlement prices of NYMEX Energy futures contracts, quotations from leading dealers in Energy and Gas swap contracts and other bona fide third party offers, all adjusted for the length of the remaining Contract Term ... and differences in locational basis." (121) Here again, the combination of the general standard "among other valuations" with the specific measures of market prices for replacement contracts confines the space within which a court can select an alternative proxy other than those specified by the parties.

The preceding examples show how parties to commercial contracts deploy precise terms alongside a standard. These combinations determine the boundaries of judicial proxy choice. The existence of precise terms constrains the court's choice of proxies under the standard. While the existence of the vague term affects the application of the specific contractual proxies. We turn now to explore the interaction between the substantive and procedural provisions of contracts.

III.HARNESSING LITIGATION BY CONTRACT

A. Contractual Design of Litigation Procedure

In some circumstances, contracting parties can agree to the procedural rules that will govern the enforcement of their contract. It is now common for parties to agree to have disputes resolved by arbitration rather than by litigation or by the court of a specified venue. (22) In many of these cases, the parties' ex ante agreement as to procedure improves the cost-effectiveness of their prospective enforcement mechanism. (23) In other words, the procedural provisions may increase the incentive bang for the enforcement buck (or lower the enforcement buck per bang). The ability of the parties to effect such improvement has further repercussions in our analysis. If the parties reduce the back-end cost-per-incentive effect, they should then substitute more back-end for front-end contracting investment. This substitution leads to further reduction in the cost-per-incentive effect and allows the parties to achieve even more efficient contract incentives. As noted above, the parties can substitute back-end for front-end costs by including more vague terms and leaving proxy choice to the enforcement process.

Although arbitration and venue clauses are common in contracts and widely discussed in the literature, the fact that parties can vary the rules of litigation in their ex ante contract is relatively unexplored. The ability of contracting parties to preclude factfinding by agreeing to confession of judgments (or cognovit notes) is well established. (124) Most courts allow parties to waive the right to jury trial, (125) and a majority of states permit contracting parties to agree to shorten the time provided by statutes of limitations to bring breach of contract actions, as long as the period is reasonable. (126) We have been hard pressed, however, to find scholarly treatises on procedure or evidence that identify the subset of these rules that are default rather than mandatory provisions. And, as we have already noted, contracts scholars focus principally on the substantive terms and not on the ability of the parties to regulate the procedural course of their future enforcement. This is a rich avenue for future research, and we take a preliminary step in this Part by examining the ways in which the parties can vary one important feature of judicial factfinding: the allocation of burdens of proof and standards of proof. (127) A threshold question is whether burdens and standards of proof are regarded as mandatory background rules or as defaults subject to alteration by individual parties. While we have not found direct authority, we believe that courts would enforce reasonable contractual burden of proof provisions. (128) And, we have found ample evidence that many contracts in fact contain such provisions. (129)

Burdens of proof illustrate the important connection identified above between the rules governing litigation and the rule/standard choice. When the parties delegate proxy choice to the court, the court typically chooses among the conflicting and self-interested proxies that the parties propose at trial. The lower the cost of resolving this dispute over competing proxies and the more efficient the expected outcome, the more likely the parties are to use vague terms ex ante. To illustrate, suppose that an agent such as a fast-food franchisee is bound by a vague contractual promise not to injure the reputation of the franchise. The agent coaches little league baseball but is also known to drink excessively. Each activity is a candidate proxy that might be selected under the contract standard. In many (if not most) cases, the factual issues are not whether the proxy is or is not satisfied (e.g., did the agent coach little league and drink?), but rather the choice of (and weight assigned to) the proxy. Under the default rules of litigation, a principal (or franchisor) who seeks to prove a breach of promise by its agent typically will be allocated both the burden of production and the burden of persuasion. Unless the principal satisfies its burden, the result will be as if the agent's proxy were chosen because the agent will be found not to have breached its promise. As a formal matter, the court determines the proxy, but the burden effectively assigns the advantage to the agent by reducing her evidentiary costs and raising the likelihood that her proxy (coaching) will be adopted over the franchisor's proxy (drinking). As we suggest below, it may be efficient in some circumstances to place the burden on the agent and favor the principal's proxy. The parties can shift the burden by contract, and thereby enhance the incentive bang for buck extracted from the vague reference to protecting the franchise's reputation.

The contrast between two classic contracts cases illustrates the significance of burden allocation. (130) Consider Raffles v. Wichelhaus (131) and Frigaliment Importing Co. v. B.N.S. International Sales Corp. (132) In Raffles, the parties entered into a contract to buy and sell cotton. Their contract called for the delivery of cotton by way of a ship named "Peerless" sailing from Bombay to Liverpool, when in fact there turned out to be two ships named "Peerless" sailing from Bombay to Liverpool within three months of each other. The buyer believed "Peerless" referred to a ship departing Bombay in October, while the seller believed "Peerless" referred to a ship departing Bombay in December. The defendant buyer refused to accept and pay for that cotton, and the court agreed, holding that "there was no consensus ... and therefore no binding contract." (133) In Frigaliment, the buyer accepted the goods and thereafter sued the seller for selling it "fowl" (lower-grade chicken) instead of "broiler" (higher-grade chicken). The seller argued that the term "chicken" in the contract included all types of chicken, while the buyer contended it meant only broiler chicken. The court found that the meaning of "chicken" was vague and ruled for the seller on the grounds that the buyer, as plaintiff, had not carried its burden of proving which of the two plausible meanings the parties intended. (134)

Allan Farnsworth has explained the importance of the burden allocation in both Raffles and Frigaliment:</p> <pre> For the seller to prevail in a suit against the buyer [in Raffles], it would seem that the seller would have to sustain "the burden"--as the court in Frigaliment put it--of showing that the word Peerless was used to refer to the ship that sailed in December. This the seller did not do. But if the buyer had sued the seller, it would seem that the buyer would have had to sustain the burden of showing that the word Peerless was used to refer to the ship that sailed in October. This, presumably, the buyer would not do. The explanation, then, for the judgment for the seller is not that there is no contract, but that neither party can sustain the burden of showing that its meaning should prevail.... If the buyer in Frigaliment had rejected the chickens and the seller had sued for the price, the same court might have found for the buyer on the ground that the seller had not sustained the burden of showing that chicken was used in the broader sense. (135) </pre> <p>In sum, because the buyer in Raffles had rejected the goods, the seller had the burden of establishing that the parties had agreed to the delivery of cotton via the December "Peerless" and was unable to do so. In Frigaliment, the buyer had the burden of establishing a narrower interpretation of "chicken" because it had accepted the goods but failed to satisfy that burden. (136) The contrasting effects of burden allocation raise two questions that have yet to be addressed in contracts scholarship: First, which is the more efficient allocation? Second, if the common law does not provide for such efficient allocation, how might the parties themselves do so by contract? The first question is complex and context-dependent. We set out below some of the factors that may affect the optimal allocation in any given case, without attempting to resolve the conditions. Indeed, we do not exclude the possibility that contracts vary procedural rules for ulterior, inefficient purposes that favor one party over the other. (137) Our contribution instead is to draw attention to the contractual mechanisms by which parties might assign burdens in their ex ante agreement.

B. Burdens of Proof

1. Efficient Burden Allocation

The nascent scholarship on the efficiency of burdens of proof falls into two groups: One is concerned with the cost-effectiveness of truthfinding, and the other with the deterrence effect on the primary behavior being regulated. (138) The first approach examines the effect of burden allocation on the cost of communicating information to the court. This scholarship is in the spirit of the seminal work of Landes and Posner, who evaluated rules of procedure as devices for minimizing the sum of the direct costs of litigation and of the costs of erroneous factfinding. (139) The advantage of the adversarial system is that the factfinder can choose between two sources of information with different cost schedules. Bruce Hay and Kathy Spier suggest that the burden of proof ought to be assigned to the party with superior knowledge of the facts in dispute or to the party asserting the more unusual version of the facts. (140) This allocation reduces the expected evidentiary costs of trial. In a similar vein, several authors suggest that burden allocation enhances the informativeness of negative evidence-the failure of a party to present evidence favorable to its case. (140) Such negative evidence is costless. However, negative evidence is also noisy when a party might be uninformed, because the court cannot infer whether the failure to present favorable positive evidence is due to the fact that it does not exist or simply that the party is unaware of it. Thus, the burden should be placed on the more informed party, or the party more likely to have access to the evidence if it is available. (142) Finally, we note that the party with superior access to information often can misrepresent the truth at lower cost than her opponent. As we observed earlier, commentators have argued that evidentiary rules pry a wedge between the costs of telling the truth and lying. Placing the burden on the party with the best information may magnify the effect of the wedge by forcing that party to present more evidence.

These theories, however, do not accurately predict the allocation of burdens in practice. For example, plaintiffs tend to bear the burden of establishing the facts necessary to plead their case, even though these facts are typically more accessible to the defendant. As Chris Sanchirico points out in this respect, tort plaintiffs carry the burden of proving their defendant's negligence, while the defendant has the burden of proving that the plaintiff was contributorily negligent. (143) Modern discovery practices may be one explanation for the insensitivity of burdens to presumed informational advantages. Discovery attenuates the informational advantages that one party might have over the other. In this light, Hay and Spier suggest that the reason for placing burdens on plaintiffs is that they typically assert the more unusual facts because people tend to comply with the law. (144) But this claim does not account for an important selection effect: the fact that the plaintiff has decided to bear the cost of initiating a lawsuit, which suggests that the defendant is more likely than average to have done wrong. Yet another theory proposes that burdens of proof follow pleading burdens by falling on the party with the more specific allegation. For example, the plaintiff pleads a specific type of negligence while a defendant asserts the absence of any negligence. While that approach may justify the pleading responsibility, it does not explain the burden of proof. Once the plaintiff has alleged the facts necessary to support its claim, the burden of proof could fall on the defendant to show that those allegations are untrue. (145)

The second line of scholarly analysis of burdens focuses on the effect of burden allocation on deterrence--in our analysis, on contract performance. (146) Burdens affect the evidentiary strategies and costs of plaintiffs and defendants. Each effect bears on ex ante incentives in two respects. First, as the expected evidentiary cost of the plaintiff rises, plaintiffs are less likely to sue, all other things equal. However, the lower incidence of litigation may lead to a string of consequences that complicates the analysis. The reduction in lawsuits may undermine the performance incentives of the defendant, causing a rise in nonperformance and a consequent increase in the expected recovery of plaintiffs. (147) It might thereby result in an offsetting increase in the number of cases filed, which might restore the performance incentive. The ultimate effect on incentives is unclear and context-dependent. (148)

Second, the evidentiary costs of defendants have a direct impact on incentives because, like the ultimate determination of liability, they impose a sanction. This sanction improves deterrence (or contract performance) if the evidentiary-cost sanction on complying defendants is lower than that on noncomplying defendants. At first blush, it may appear that the allocation of burdens does not affect this process because burdens assigned to defendants fall indiscriminately on complying and noncomplying actors alike. However, the following simple example suggests otherwise by taking into account the plaintiffs incentive to sue. (149)

Suppose that a principal-agent contract requires an action that will cost the agent $100 and provides that the agent must pay liquidated damages of $105 if she fails to perform. In order for the principal to enforce the provision, the court must determine whether the agent performed or not. Suppose that the evidentiary cost to the agent of proving performance is $10. The net gain to the agent from performance is $105-$100 = $5, less whatever evidence cost the agent would have to pay to exonerate herself. If the burden is on the principal, the principal simply will not sue if the agent performs, and the agent would enjoy the full gain of $5 from performing. If the burden is on the agent, however, the agent would suffer a net loss of $5 compared to nonperformance because of the $10 it would have to pay to satisfy the evidentiary burden. Thus, the litigation burden imposes a prospective tax on the defendant agent that discourages performance. In this example, therefore, the burden is more appropriately placed on the principal, the potential plaintiff. (150)

For the purposes of this Article, it is particularly important to note that whether the objective is to reduce evidentiary costs or to improve contractual incentives, the effects of burden allocation are highly context-dependent. These advantages are not very susceptible to general rules of allocation. At best, the law can provide default allocations from which the parties may contract away if they wish. Thus, parties may tailor burdens to accommodate their particular circumstances. For example, the contract might shift the burden to the defendant if the defendant has access to a key exculpatory document and if discovery is costly or imperfect in enabling the other party to obtain the document. In a similar vein, parties who seek to use burdens to sanction nonperformance and reward performance must also be sensitive to such context-specific factors. Harnessing burdens by contract, however, requires first an appreciation of the default rules by which the law allocates burdens of proof.

2. Default Burden Allocations

The default scheme of proof burdens allocates burdens of production and burdens of persuasion, and is overlaid by the operation of presumptions that shift burdens between the parties as they present their evidence. As noted above, the default burdens of proof track the pleading burdens. The general rule is that the pleading responsibility rests on the party who invokes the intervention of the court to change the status quo. (151) There are exceptions, however, for affirmative defenses. In breach of contract actions, the plaintiff must plead and prove a claim of nonperformance. The plaintiff must allege and prove the making of the contract, its consideration, and the satisfaction of all conditions precedent (whether express or implied) to the defendant's reciprocal obligation to perform. (152) The defendant may respond that the obligation to perform has been discharged on any of a number of substantive grounds, including novation, accord and satisfaction, cancellation and termination, impossibility, mutual mistake, release, alteration, merger, and the failure of a condition subsequent to performance. As affirmative defenses, all of these must be pleaded and the burden of production carried by the defendant. (153) Once all the burdens of production are met, the overall burden of persuasion, however, remains on the plaintiff. Thus, the burdens with respect to most facts in contract-breach cases- particularly the performance standards with which we are most concerned-fall on the plaintiff, the party who is seeking to change the status quo. (154)

A doctrinal explanation of the general allocation to the plaintiff and the exception of affirmative defenses eludes commentators. (155) Courts and commentators typically offer three justifications for affirmative defenses. (156) First, the defendant may have the comparative advantage in information production. (157) Second, the defendant may be assigned the burden with respect to a fact that is particularly unusual. (158) Third, commentators sometimes mention a category comprising defenses that are normatively disfavored, such as contributory negligence or statutes of limitations. (159) Unfortunately, however, given their generality and the inconsistency of their application, none of these supposed policies is reliable as a working rule. (160) In particular, we noted in the previous Subsection that the comparative advantage in information production fails to predict the allocation of burdens in practice. (161) For example, in breach of contract claims the plaintiff has the burden of proving that the defendant's conduct constituted a breach notwithstanding the fact that the defendant has better access to the facts in question. (162)

Substantive contract law often determines which party will be the plaintiff in disputes. For example, the identity of the plaintiff may depend on the course of the parties' actions. Consider section 2-607(4) of the Uniform Commercial Code. This provision assigns the burden of proving that a delivered good does not comply with the contract to the buyer if the buyer accepts the good. If instead the buyer simply rejects the good, the burden falls on the seller who sues for breach of contract. This burden includes the burden of establishing facts as to the condition of the goods upon delivery. (163) The identity of the plaintiff in any dispute-and the consequent allocation of the burden of proof-thus may rest on factors having little to do with either informational advantages or self-interested behavior. (164) In the case of an allegedly defective good, the burden hinges on whether the plaintiff has accepted or rejected delivery. In short, the parties' ex post actions can affect the burdens in litigation. Consequently, the default allocation of burdens is neither predictable at the time of contracting nor based on factors that seem to have clear efficiency consequences. More pertinent to our project, therefore, is the ability of the parties to determine burden allocation (and the proof standards) by their ex ante contract, a subject to which we now turn.

C. Contractual Allocations of Burdens and Standards of Proof

Like other commentators, we are hard pressed to rationalize the procedural rules for burdens of proof. It is therefore not surprising that contracting parties might wish to fashion their own rules. Even if the legal scheme can be justified, it is highly unlikely that it yields the efficient burden allocation for every contract. The parties may therefore wish to clarify, reverse, or fine-tune the default allocation in their contract. In this Section, we identify three ways by which the parties might do so, and we provide examples from commercial practice: The first approach is by direct allocation of burden; the second is by predesignating whom the plaintiff will be in the event of a dispute; and the third is by framing the substantive provisions governing, for example, the right to assign or terminate a contract. We also observe that the parties' flexibility extends beyond simple binary burden allocation between the parties. They may aLso provide for shifting burdens based on explicit or implied presumptions.

I. Direct Burden Allocation

The most straightforward way for parties to reallocate burdens or alter the standard of proof is for them to do so directly through an explicit term in the contract. Indemnity agreements, for example, commonly reallocate burdens and elevate standards of proof. (165) Consider the standard indemnification agreement between DAOU Systems and its directors and officers. The contract provides in relevant part:</p> <pre> Presumptions and Effect of Certain Proceedings. (a) Upon making a request for indemnification, Indemnitee shall be presumed to be entitled to indemnification under this Agreement and the Company shall have the burden of proof to overcome that presumption in reaching any contrary determination.... (b) Indemnitee shall be deemed to have acted in good faith if Indemnitee's action is based on the records or books of account of the Company, including financial statements, or on information supplied to Indemnitee by the officers of the Company in the course of their duties, or on the advice of legal counsel for the Company or on information or records given or reports made to the Company by an independent certified public accountant or by an appraiser or other expert selected with reasonable care by the Company. In addition, the knowledge and/or actions, or failure to act, of any director, officer, agent or employee of the Company shall not be imputed to Indemnitee for purposes of determining the right to indemnification under this Agreement. Whether or not the foregoing provisions of this Section 7(b) are satisfied, it shall in any event be presumed that Indemnitee has at all times acted in good faith and in a manner he reasonably believed to be in or not opposed to the best interests of the Company. Anyone seeking to overcome this presumption shall have the burden of proof and the burden of persuasion by clear and convincing evidence. (166) </pre> <p>Consistent with the corporate statute of Delaware, where DAOU is incorporated, this provision conditions the firm's obligation to indemnify on the good faith of the director or officer. (167) Litigation over this provision is likely to be brought by the director or officer seeking indemnification. In the absence of a contract term to the contrary, this party would carry the burden of showing that she acted throughout in good faith by introducing proxies supporting this claim. Yet the DAOU Systems standard form (like most agreements of its kind) shifts the burden to the firm, which is typically the defendant, and also elevates the standard of proof from the default "balance of probabilities" to "clear and convincing evidence." (168) It shifts the burden by way of a presumption that is triggered when the agent presents the minimal evidence that her actions were based on the company's records or books, on the advice of legal counsel, or on information supplied by an independent certified public accountant.

The parties had the following three options, among others, in drafting their indemnification agreement. The agent might have enjoyed (1) a blanket entitlement to indemnification, (2) an entitlement conditional on a finding of good faith (without varying the default burden of proof), or (3) an entitlement conditional on good faith, together with a presumption of good faith. The first option would protect a risk-averse agent from liability if the firm should fail but also would insulate the agent from bearing the cost of her negligence or self-dealing. The second option would deter the agent from such misbehavior but leave her open to the firm's opportunistic claims that she had not acted in good faith (even though untrue). Although the firm ultimately might fail in court, the agent would bear litigation costs and the risk of legal error. As a result, she would be reluctant to make risky decisions on behalf of the firm, even if they would be profitable. This is particularly true if the agent were to bear the burdens of proof, as she would under the default procedure. The parties might choose between these first two options by weighing the severity of the risk of agent and firm opportunism. If the risk of firm opportunism were relatively more severe, the parties would omit the good faith exception.

The third option permits the parties to use the procedural tools of burdens and standards of proof to reach an intermediate solution that fine-tunes the tradeoff between setting efficient incentives for one party or the other. (169) We also can see how the use of a vague standard such as "good faith" is more likely when the parties can manipulate procedural rules such as burdens of proof. If the parties might have eschewed good faith when limited to the binary choice between the first two options, they might include it if they could fine-tune their agreement with the aid of burden allocation. The example in the next Subsection provides another illustration in which contracting over burden allocation broadens the range of available incentive schemes.

2. Predesignation of Plaintiff

Parties can harness burdens indirectly, without an explicit contract term. As an example, consider once again a simple sales contract between buyer and seller. Recall that under section 2-607(4) of the Uniform Commercial Code, the default burden of proving whether a good is defective or not depends on whether the buyer has accepted or rejected the good. The rejecting buyer sues the seller for damages and carries the burden; the seller sues the accepting buyer for the price and carries the burden. As in the indemnification contract in the previous Subsection, parties to a sales contract could contract directly over which party, the seller or the buyer, would bear the burden of proof as to the condition of the goods in all cases. As an alternative to an explicit contract term, however, the parties can harness the efficiency benefits of burdens indirectly. Although there are default rules defining acceptance and rejection, they may be varied by contract to implicitly assign the burden of proof to the seller or the buyer, depending on the parties' preferences. (170) Deposits are a mechanism by which commercial parties may structure substantive provisions to influence the likely identity of the plaintiff. (171) When the buyer has prepaid or made a deposit, the seller has less to gain by suing for the price. Thus, the buyer is more likely to be the plaintiff whether she has accepted or rejected the goods.

Construction contracts present a variation of this approach. Like the "good faith" requirement of directors and officers in the indemnification contract, the default threshold for contractor performance is also a standard: "substantial performance." (172) And, as a vague term, it raises the prospect of high litigation costs, uncertainty, and opportunistic claims by either side that undermine the efficiency of incentives. The owner may introduce evidence that apparent defects in construction (such as noncomplying piping material) reduce her value substantially, whether or not this is true. Conversely, the builder's opportunistic strategy is to shirk on performance but claim that it nevertheless complied with the standard, by offering an alternative proxy (such as aesthetic appearance).

A construction contract typically requires the property owner to make progress payments to a builder during construction. An important contract design choice, therefore, is whether each payment is made before or after the builder completes the construction to which the payment relates. One might think of this as choosing which party gives value first--essentially, a deposit for each stage. As we now know, this decision determines the party who bears the burden of proof and whose opportunistic arguments at litigation are correspondingly constrained. Assume initially that payments are made in advance, and particularly that the final payment is made prior to the completion of construction. This provision places the default burden on the owner (as plaintiff suing to recover its payments) in litigation over whether the builder has substantially performed its obligation. The burden deters opportunistic suits by the owner and might reduce litigation costs. Yet a reduction in the likelihood of litigation might also undermine the builder's incentives by enabling him to point to self-serving proxies for substantial performance. If this is the net effect of the burden allocation, then the parties must trade off the litigation cost savings against the adverse effect on performance incentives. This is a similar tradeoff to that described above in the context of suits by corporate officers for indemnification. The parties have a procedural as well as substantive decision variable with which they can fine-tune the balance: They can contract for the standard of proof as well as the burden of proof. Shifting the burden to the owner is less significant when the standard remains the preponderance of the evidence than when the standard is raised to clear and convincing evidence, as in the indemnification agreement.

In sum, the contracted order of performance--whether the construction occurs before or after the corresponding payment from the owner--determines who is more likely to be the plaintiff and, accordingly, who will carry the burdens of proof. Given the standard of proof, the burdens may be significant because of the relative costs that they impose on the litigants and their relative likelihood of victory. Thus, the alternative burden allocations have contrasting effects on incentives. Yet the simple choice of placing the burden on one party or the other is unlikely to achieve the first-best incentives for builder performance.

In the construction contract at issue in the famous case of Jacob & Youngs, Inc. v. Kent, (173) the parties adopted a more intricate burden-shifting solution common in construction contracts. Their contract provided that final payment was due upon the issuance of the architect's certificate. (174) Effectively, this provision assigned the choice of proxy to the architect, presumably on the grounds that he enjoys the advantage of industry expertise over the court. Like an arbitrator, the architect's discretion is disciplined by his reputational stake in not appearing to be biased in favor of builders or owners. After all, he would like to be chosen in subsequent similar arrangements. Yet, in most such contracts, the architect's certificate operates as a presumption of substantial performance that can be rebutted by evidence that its issuance was influenced by fraud, bias, or mistake. (175)

3. Framing of Substantive Rights: Contract Assignment and Termination

The following set of examples differs from the foregoing in that the burden allocation results from the manner in which substantive rights are framed. In addition, the examples are interesting because the parties' dispute is not simply over the division of the spoils from a completed relationship. Rather, it occurs in the midst of a potentially ongoing relationship, such as a distributorship or franchise. This is an important difference because it complicates the weighing of possible opportunistic behavior by each party. In the construction and indemnification examples discussed above, the parties generally assess the relative concerns that a promisor would shirk or that the promisee would sue opportunistically. In the cases that follow, as in the earlier example of the loan agreement, the parties are also concerned with the parties' opportunistic attempts to terminate or continue (or assign) the relationship for selfinterested rather than efficient reasons.

a. Assignment Clauses

Assignments of contract rights have mixed efficiency consequences. On the one hand, they can move contract rights from lower- to higher-valued uses. On the other hand, leaving the assignment decision to the promisor may lead to inefficient transfers because she does not internalize the cost of the assignment to the promisee. So, for example, franchise agreements restrict the ability of the franchisee to assign its rights under the contract because the franchisee is interested only in maximizing the proceeds from a purchaser, without regard to the effect of the new franchisee on the franchise's reputation and value. Thus, agreements do not permit assignment but, recognizing that transfers may be efficient, the contracts also do not prohibit all transfers. Banning assignments completely would prevent the exploitation of the franchisee's private knowledge of higher-value franchisees. It is difficult, however, to distinguish between the benign and malign scenarios by precise rules. Consequently, the parties rarely attempt to list requirements that must be met. Instead, they invoke a "reasonableness" standard under which the franchisor's consent to any assignment is required but will not be "unreasonably withheld." (176) The reasonableness requirement is intended to have bite. The parties guide the courts by combining the standard with precise rules or by explicitly stating the objective of the standard. The Taco Bell Franchise Agreement, for example, states:</p>

<pre> The Franchisee acknowledges that the purpose of the aforesaid 

restriction is to protect the Company's trademarks, service marks,

trade secrets and operating procedures as well as the Company's

general, high reputation and image, and is for the mutual benefit

of the Company, the Franchisee and other franchisees of the Company. (177) </pre> <p>The contract further provides that "[i]n considering a request for a transfer, the Company will consider, among other things, the qualifications, apparent ability and credit standing of the proposed transferee as if the same were a prospective, direct franchisee of the Company." (178)

Consider two alternative ways of framing the reasonableness condition that illustrate the parties' anticipation of burdens of proof. Under either alternative, litigation addresses the issue of whether the assignment is reasonable and the parties present alternative proxies. The first approach permits the franchisee to assign its rights only if reasonable. The second permits the franchisee to assign its rights only with consent of the franchisor and provides that such consent will not be unreasonably withheld. Commercial agreements tend to adopt the latter approach to regulating assignments. The choice of the latter version anticipates the assignment of burdens of proof in litigation. In the former case, the franchisor, suing for damages and to prevent the continued use of its trademark, would be required to prove that the transfer was not reasonable. Under the latter version, the franchisor would initially establish that it withheld consent. Then, the burden would shift to the franchisee to show that the franchisor's consent was unreasonably withheld. One might speculate that this allocation may be efficient on grounds of comparative information advantages: The person in contact with the intended transferee is likely to have better information about the qualifications of the new franchisee.

b. Contract Termination 

Explicit termination clauses are common in many different categories of commercial contracts, including employment agreements, service contracts, merger and acquisition agreements, loans, and franchise and distributorship arrangements. (179) Their role is puzzling because even in their absence, either party to an ongoing relationship can terminate by declaring that the other party materially breached its obligations. Under the common law of contracts, material breach entitles the nonbreaching party to withhold performance and seek damages for breach. (180) One reason for explicit termination clauses is to provide for the conditions that trigger termination, rather than relying on the common law requirements for material breach. We suggest in this Subsection that termination clauses also tailor burden allocation. Indeed, we present evidence that the burden design under termination rights may entail burden-shifting similar to that invoked by the provision for the architect's certificate in construction contracts.

Consider in general terms the benign and malign reasons why a party to a long-term contract--such as a lender, employer, or franchisor--might wish to terminate the relationship. For convenience, we refer to that party as the principal and the counterparty as the agent. First, a principal may wish to terminate the contract because the agent failed to exert the level of effort required in the contract (i.e., shirked) and thereby jeopardized the value of the relationship. Shirking may be an efficient justification for termination: It both arrests a relationship that is no longer valuable because of the dealer's shirking and yields an ex ante discipline that might deter shirking. (181) Second, the principal may terminate because the materialization of an exogenous risk, such as changed market conditions, has rendered the contract unprofitable to the principal (but not to the agent). Cancellation for this reason alone would lead to the loss of the relationship's future value and would also undermine the contract's allocation of risks. The principal's incentive to guard against exogenous risks that make its own performance more costly or that make the return performance less valuable is undermined by its ability to escape adverse changes through termination. (182) Third, the principal may threaten termination in order to force a renegotiation of its terms so as to secure a larger share of the contract surplus. This opportunism is an attempt to appropriate the agent's contract-specific investments in the ongoing relationship. The prospect of renegotiation deters the agent from investing in the relationship. As with the second reason, the principal's termination is opportunistic and contrary to the ex ante interests of the parties.

In light of the mixed motivations for termination, the parties might seek to regulate in their contract the principal's substantive right to terminate. Depending on their assessment of the front-end and back-end costs of doing so, they would agree to a combination of vague and precise substantive triggers for the right to terminate. (183) If the parties do not address burdens of proof by contract, either directly or indirectly, the default allocation of burdens governs and determines the impact of the substantive conditions of termination. The plaintiff has the burden of establishing the prima facie case, and the defendant bears the burden of proving affirmative defenses. Either the principal or the agent might be the plaintiff in litigation. For example, a manufacturer who has terminated a distribution arrangement may seek to enjoin further use of the trademark by the dealer and exercise its option to buy the dealer's premises on termination. Alternatively, the dealer may be the one who sues in order to recover damages for unjust termination. Yet the efficient allocation of burdens depends on their impact on the various incentives described above and on litigation costs. These are unlikely to be affected by whether the plaintiff is the manufacturer or the dealer. We made the same observation earlier in connection with the effect of a buyer's acceptance or rejection on the burden of proving the defect or conformity in goods delivered under a sales contract. The failure of the default burden allocation to respond to incentive effects may lead parties to tailor burdens through an express termination provision.

Unfortunately, the parties are unlikely to be any more successful than they would be under the default scheme in conditioning the allocation of the burden of proof on the principal's motivation for terminating. After all, the principal's motivation lies at or at least near the core of the factfinding operation. Assigning the burden of proof in the contract to the principal would deter inefficient termination but also efficient cancellation in response to shirking by the agent. Therefore, if burden allocation is a binary choice, the best available arrangement depends on a comparison of the prospects of efficient and opportunistic cancellation. For example, to the extent that the agent's incentive to shirk is disciplined by reputational constraints, the burden of proof is more appropriately placed on the principal. We encountered a similarly rough determination in the context of the indemnification agreement and the construction contract. To give an example more specific to the termination context, if the principal's exposure to exogenous risks is small or if the agent's specific investment is minor, then the parties might be more likely to allocate the burden to the agent. In any event, our main observation is that the allocation of burdens provides a procedural lever that complements the substantive termination right.

The case of International Harvester Co. v. Calvin (184) demonstrates how a termination clause might yield a more complex shifting burden of proof. International Harvester concerned a long-term franchise contract for the sale of heavy-duty trucks within a designated region. The contract contained a combination of rules and standards governing the distributor's performance under the contract. These provisions committed the distributor, inter alia, to exercise its best efforts to promote the sales of the manufacturer's product, to "'provide and maintain physical facilities commensurate with the sales possibilities and service needs in the Dealer's trade area,'" and to "'achieve a reasonable share of the market for the goods covered by the agreement in the normal trade area served by the Dealer's location."' (185) Two years after the contract was concluded, the manufacturer notified the dealer that it was in violation of its obligations, including its commitment to maintain a reasonable market share. The manufacturer warned that it would terminate unless the dealer corrected the violations. Subsequently, the manufacturer notified the dealer that the agreement was terminated, effective in ninety days. (186) A state regulatory body set aside the termination, however, and the manufacturer sued to reverse that administrative order.

The court interpreted the termination provision to include a tailored allocation of burdens similar to that raised by the architect's certificate in the construction contract discussed earlier. The court effectively treated the termination provision as if it were a delegation of proxy choice to the manufacturer itself. The manufacturer enjoyed a presumption that the termination was justified if it could satisfy two easy requirements at trial: establish by simple affidavit that the dealer had failed to comply with the reasonable market share requirement and that the manufacturer had delivered the required termination notice. (187) The burden then fell to the dealer to prove that it had in fact complied with its contractual obligations.

As in the case of the architect's certificate, however, the court was also receptive to claims of process abuse. The dealer could avoid the burden of proving compliance by showing that the manufacturer had an ulterior motive in terminating--for example, that the manufacturer sought to install another dealership in the adjoining county. Indeed, in International Harvester, the dealer had filed a formal protest with the agency charged with jurisdiction over claims of unfair treatment of dealers. Only then, the dealer contended, did the manufacturer's evaluations of the dealer's sales performance begin to deteriorate. The dealer also testified that the manufacturer attempted to coerce the dealer to expand its facilities and greatly increase its investment in inventory and fixed costs. The notice of termination, the dealer argued, was the result of its reasonable refusal to comply with these demands. The court held that this prima facie showing of bad faith shifted the burden to the manufacturer to show by a preponderance of the evidence that the termination was not motivated by strategic considerations and that it would have terminated even in the absence of the alleged bad faith purpose. (188)

This example of shifting burdens (or presumptions) suggests that the parties have more flexibility in burden design than all-or-nothing allocations to each party. The court in International Harvester adapted the scheme of presumptions and shifting burdens from civil rights case law. (189) Yet we should not miss the important lesson for contract design. The discussion underscores the importance, but also the complexity, of the contracting task of efficient burden assignment, whether by explicit or implicit provisions. On the one hand, the parties must identify and evaluate the relative severity of the agent's incentives to shirk and the principal's incentives to make opportunistic claims of breach, both in terms of their likelihood and their efficiency consequences. On the other hand, the parties must anticipate future litigation, and in particular who is more likely to be suing and for what. Although the parties can undoubtedly improve on the default burden allocations, the tailoring task is also likely to involve substantial upfront transaction costs.

CONCLUSION

In this Article, we analyze the relationship between the front-end and backend stages of contracting by examining (1) the choice between precise and vague terms and (2) the interaction between substantive and procedural contract provisions. We offer a preliminary theory explaining the feedback effect of the adversarial litigation system--and especially the process of proxy selection and proof--on contract terms. In doing so, we hope to set a research agenda for further integrating the litigation mechanism with the theory of contract design. Much can be gained by a sharing of knowledge and insights between procedure and contracts scholars and, in the world of practice, between litigators and transactional lawyers. Indeed, contract design should anticipate not only the effect of litigation, but also other possible back-end processes, such as arbitration, renegotiation, or settlement.

Commercial parties can (and do) design contracts that motivate better contractual performance and do so at lower cost than has been previously understood. By examining how contracts can anticipate and harness the litigation process, we hope to breathe new life into the scholarly acceptance of vague terms by rebutting a persistent skepticism in contracts scholarship about their cost effectiveness. Vague terms can be valuable by deferring proxy selection to the enforcement stage, particularly when the parties can also improve the efficiency of litigation by, for example, manipulating the assignment of burdens of proof. The use of deposits or termination rights in combination with vague terms illustrates this strategy.

The claim that party-created standards can enhance efficiency in harnessing the ex post informational advantage available at litigation does not conflict with a formalist view of contract enforcement. Formalism, in this context, instructs the court to apply the precise terms of the contract unless the parties opt out by clear language that defines a broader interpretive space. Thus, notwithstanding their ex post informational advantage, courts should not imply default standards in the face of precise contract rules. The cost to the parties of writing such vague terms is low and they have better information than public lawmakers in deciding whether and how much to delegate to their court through the combination of rules and standards.

A distinct question concerns the role of the court when the parties fail to make any provision for an obligation or a contingency. (190) Suppose that the contract in Bloor v. Falstaff Brewing (191) made no reference to Falstaff's obligation to sell Balantine beer. Even when contracts are obligationally incomplete in this way, we argue that the courts should not inject a standard as the default. In many such cases, the parties anticipate that extralegal sanctions would fill in such a gap. When this is not true, the absence of a default standard is likely to encourage parties to bargain expressly over the crucial question of contract design: How much discretion should be left to the back-end enforcement process?

AUTHORS. Robert E. Scott is David and Mary Harrison Distinguished Professor, University of Virginia School of Law; Justin W. D'Atri Visiting Professor of Law, Business and Society, Columbia Law School. George G. Triantis is Perre Bowen Professor, University of Virginia School of Law; Visiting Professor, Yale Law School. The authors thank Ian Ayres, Richard Brooks, Albert Choi, Michael Dooley, Lee Fennell, Ron Gilson, Victor Goldberg, Mitu Gulati, John Harrison, Edward Iacobucci, Dan Kahan, Kevin Kordana, Jody Kraus, John Langbein, Anup Malani, Caleb Nelson, Eric Posner, Carol Rose, Chris Sanchirico, Alan Schwartz, Michael Trebilcock, and workshop participants at the 2005 AALS meetings and at Berkeley, Duke, Pennsylvania, Ohio State, Virginia, and Yale Law Schools for helpful comments. We also thank Patrick Dempsey, Ben Doherty, and Kent Olson for research assistance.

(1) Parties trade efficiently when the value of the exchanged performance to the buyer exceeds the cost of performance to the seller; parties rely (or invest) efficiently when their reliance maximizes the contract's expected surplus net of reliance costs. See, e.g., Alan Schwartz & Robert E. Scott, Contract Theory and the Limits of Contract Law, 113 Yale L.J. 541, 545 (2003).

(2) Note that a contract may be obligationally complete even though it is informationally incomplete. An obligationally complete contract might lump together various states and provide for the same obligations across the states of each lumped set. Yet, such a contract is informationally incomplete because it fails to discriminate within each set between states of the world that, optimally, call for different obligations. States of the world reflect both exogenous and endogenous variables. For example, different oil prices produce different states, but so does the decision of a seller to tender or not. Each event changes the state of the world and may be paired in the contract with a different obligation on the buyer.

(3.) E.g., Alan Schwartz, Incomplete Contracts, in 2 THE NEW PALGRAVE DICTIONARY OF Economics and the Law 277 (Peter Newman ed. 1998) [hereinafter New Palgrave Dictionary]; see also infra notes 18-19.

(4.) A nonverifiable factor is one for which the information cost at trial outweighs the incentive benefit of the related contractual provision. The paradigmatic example in agency contracts is a clause that requires a minimum level of effort from the agent when a third party (such as a court) cannot observe directly how hard the agent is working. Economists postulate that parties will not contract on factors that are nonverifiable. E.g., Oliver Hart & John Moore, Incomplete Contracts and Renegotiation, 56 Econometrica 755 (1988). Many legal scholars have adopted this premise as well. E.g., Eric A. Posner, Economic Analysis of Contract Law After Three Decades: Success or Failure?, 112 Yale L.J. 829, 857 (2003) ("The literature stipulates that transaction costs mean that the [reliance] investment is not verifiable by a court, so the parties gain nothing by putting the optimal investment in the contract."); Alan Schwartz, Relational Contracts in the Courts: An Analysis of Incomplete Agreements and Judicial Strategies, 21 J. Legal Stud. 271 (1992).

(5.) A notable exception is Richard A. Posner, The Law and Economics of Contract Interpretation, 83 Tex. L. Rev. 1581, 1583-84 (2005).

(6.) See, e.g., Oliver Hart, Firms, Contracts, and Financial Structure 37-38 n.15 (1995) (arguing that because vague standards such as quality are not verifiable, parties contract on alternatives, such as sales volume); Schwartz, supra note 4, at 304 (suggesting that because "best efforts" obligations are not verifiable, courts will authorize any efforts above zero).

(7.) See Christopher R. Drahozal & Keith N. Hylton, The Economics of Litigation and Arbitration: An Application to Franchise Contracts, 32 J. Legal Stud. 549, 550-51, 554-55 (2003) (explaining that when parties agree to be bound by arbitration, they may prefer vague terms); Posner, supra note 5, at 1594.

(8.) We use the term "proxy" in this Article to describe what proceduralists refer to as "operative facts," which are relevant to establishing compliance with precise and vague contract terms. A precise term narrowly confines the content of the operative facts. Indeed, in the limiting case the term directly specifies the evidentiary proxy. A vague term (or standard) defines a broader space within which a court can select the evidentiary proxy that best establishes compliance with the term.

(9.) See, e.g., U.C.C. [section] 2-7008(1) (2004) (defining seller's market damages); id. [section] 2-713(1) (defining buyer's market damages); id. [section] 2-723(2) (defining the criteria for proof of market damages by stating: "If evidence of a [market] price prevailing at the times or places described in this Article is not readily available the price prevailing within any reasonable time before or after the time described or at any other place which in commercial judgment or under usage of trade would serve as a reasonable substitute for the one described may be used....").

(10.) See Colin s. Diver, The Optimal Precision of Administrative Rules, 93 Yale L.J. 65, 73 (1983) ; Isaac Ehrlich & Richard A. Posner, An Economic Analysis of Legal Rulemaking, 3 J. Legal Stud. 257 (1974) ; Louis Kaplow, Rules Versus Standards: An Economic Analysis, 42 Duke L.J. 557 (1992); Carol M. Rose, Crystals and Mud in Property Law, 40 Stan. L. Rev. 577 (1988).

(11.) E.g., Kaplow, supra note 10, at 559-60. Kaplow contrasts the costs of promulgation and of enforcement of regulation: Standards are more likely to be preferable when the former are larger and the latter are smaller (and vice versa). Promulgation costs are larger if the regulation covers numerous heterogeneous circumstances, so that standards are more appropriate in these cases.

(12.) Kaplow distinguishes between the timing choice and the choice between the institutional choosers (legislator, regulator, or court). Id. at 608-11.

(13.) Kaplow also distinguishes between the timing and complexity of regulation: They are distinct attributes in that rules or standards each can be more or less complex in addressing discretely the different circumstances that might arise. Id. at 586-96.