This entry addresses two fundamental characteristics of rules. The first concernsthe degree of precision, detail, or complexity they embody: how finely aredifferent sorts of behavior to be distinguished? A second aspect of legalcommands concerns when a given level of detail is provided -- at the time ofpromulgation ("rules") or subsequent to individuals' actions, in the context ofadjudication ("standards"). These aspects of rules are considered from aperspective that focuses upon information costs and dissemination: differentsorts of legal commands involve differing costs of formulation and applicationby private parties (deciding upon their own conduct) and adjudicators, and thecharacter of laws also influences how well parties actually will understand thelaw and conform their conduct accordingly. The discussion encompasses relatedquestions involving the role of precedent, the evolution of the law over time,legal uncertainty, and accuracy in adjudication. This entry also addresses theseparate problem of how changes in legal rules should apply to prior behavioror preexisting investments -- issues of retroactivity and transition.

JEL classification: K40

Keywords: Rules, Precision, Complexity, Transitions, Retroactivity, Standards

1. Introduction

Legal rules serve many functions: channeling behavior (e.g., tort law,environmental law, criminal law), providing background rules (much of contract,commercial, and corporate law), and defining obligations and eligibility forbenefits (tax law, social welfare provisions). This entry, following existingresearch, focuses on the first type of law. The relevance (or irrelevance) of theanalysis to other types of law will usually be apparent, and some commentsabout the differences in contexts will be noted.

The most commonly noted characteristic of rules concerns the degree ofprecision, detail, or complexity they embody: how finely are different sorts ofbehavior to be distinguished? Such matters may arise in defining the scope ofa legal command, providing for exceptions, adjusting sanctions based uponaggravating and mitigating circumstances, and so on. A related aspect of legalcommands concerns when a given level of detail is provided -- at the time ofpromulgation ("rules") or subsequent to individuals' actions, in the context ofadjudication ("standards"). These aspects of rules are considered from aperspective that focuses upon information costs and dissemination: differentsorts of legal commands involve differing costs of formulation and applicationby private parties (deciding upon their own conduct) and adjudicators, and thecharacter of laws also influences how well parties actually will understand themand conform their conduct accordingly. This approach is used to illuminatechoices about the degree of precision (or complexity) with which legalcommands are formulated and between rules and standards. These subjects, inturn, encompass related questions involving the role of precedent, the evolutionof the law over time, legal uncertainty, and accuracy in adjudication.

After outlining the main ideas concerning these topics, this entry addressesthe separate problem of how changes in the law should apply to prior behavioror preexisting investments -- issues of retroactivity and transition -- andconsiders some other aspects of rules that have been addressed in the economicanalysis of law.

A. Precision of the Law

2. Precision of the Law: The Problem

The precision of a legal command is taken here to refer to the degree of detail ordifferentiation involved. For example, an environmental regulation is moreprecise I f more types of pollutants or sources of pollution are distinguished. Anegligence regime is more precise if the standard of care is more finely tailoredto different types of actors in different contexts. A damages scheme is moreprecise the greater the extent to which damages are designed to reflect theparticular harm done to a particular victim.

The greater the degree of precision, the greater will be the costs offormulating legal commands and applying them in adjudication and of partiesinterpreting them for purposes of deciding how to conform behavior to the law.But it does not follow that the costs of precision -- any more than costs oflitigation or of pollution control -- should be minimized: the simplest rules mightpermit all acts, require equal reduction of pollutants regardless of their toxicity,or require the same speed limits on all roads. Rather, the degree of precisionshould be optimized, which makes an economic analysis particularly useful.

Before proceeding, it is worth noting that greater precision in legalcommands -- higher levels of detail, more distinctions recognized in rules andexceptions -- is a primary source of legal complexity, which is broadlycondemned. The subject of the next section, whether a legal command isformulated as a rule or a standard (the latter leaving more open to subsequentadjudication and thus being more difficult to interpret), also may bear on whatis often described as legal complexity. A number of other sources of complexitycould be considered, but they have received little attention in the literature.Most obviously, poorly drafted laws and legal opinions will be more confusing;it will be more costly to extract less information. Although this problem is oftenserious, analysis of it is straightforward. More subtle are problems like thoseconfronted under tax laws wherein parties may devise complicated schemes tocircumvent simple legal commands. In this setting, complexity may beunavoidable. Moreover, it is possible that greater complexity will reduce costs:if a more complex regime successfully discourages transactional manipulations,parties might in the end spend less in learning the intricacies of legal rules andin making convoluted arrangements.

3. Effects of Precision on Legal Costs and on Behavior

The presentation here follows most closely the model in Kaplow (1995a) . Seealso Ehrlich and Posner (1974) , Diver (1983) . First, we can consider the effects ofprecision on legal costs. It is straightforward that greater precision tends toincrease them: more detailed laws tend to be more costly for the government topromulgate, for parties to interpret, and for enforcers to apply. The second factor-- private interpretation costs -- involves some subtlety, because parties willincur such costs only if they find it in their interest to do so. For example,extremely complex details may be ignored by almost everyone.

Second, more precise commands -- those that differentiate behavior moreprecisely -- will generally result in better behavior. If pollutants are differentiatedand subject to particularized sanctions, rather than grouped together, subject toa single sanction (reflecting, say, the average harm they cause), there will be twobeneficial results. Activities that are more harmful will be subject to highersanctions than otherwise, which will efficiently deter additional undesirableactivity. Similarly, less harmful activities will face lower sanctions, avoidinginefficient overdeterrence.

These behavioral benefits are related to private legal costs, because thebenefits can arise only if private actors expend the resources necessary to learnwhether their particular activity is of a more or less harmful type, and thussubject to a higher or lower sanction. If the cost of determining the relevantsanction -- which may involve learning about the content of legal commands orabout the nature of one's own activity (such as by testing the chemical contentof discharges) -- is too high, individuals will not learn about the legal rules.Individuals with rather low benefits may simply forgo activities, even thoughthere is some chance that they may be subject to low sanctions or no sanction,in which case their activities would be worth undertaking. Individuals with highbenefits may simply act, even though there is some chance that they may facea very high sanction, in which case they will regret their decision. Individualswith intermediate benefits, however, will tend to acquire information and then toact accordingly -- that is, to proceed if the sanction is low and abstain if thesanction is high.

Given the nature of this problem, it is natural to ask whether private parties'incentives to acquire information will be socially appropriate. It turns out that inthe simplest case, in which liability is strict and there are no predictable errorsin adjudication, private and social incentives are aligned (for the familiar reasonthat private actors bear all the social costs of their activities, here including thecost of acquiring information). More generally -- if there is legal error, ifindividuals may misestimate the value of information, if information facilitatescircumvention of legal sanctions, or if a negligence regime governs -- privateincentives may be excessive or inadequate. See Shavell (1988a) , Kaplow (1990) , Kaplow and Shavell (1992) .

Having stated how parties will behave and what costs will be incurred whenlaws are more detailed, it is possible to characterize when greater precision isdesirable. At the broadest level, the lower are the information costs and thegreater is the variation in harm caused by different individuals' actions, the largerwill be the net benefit of precision in legal commands. To offer more insight, itis helpful to consider some special cases.

First, suppose that formulation costs are low (perhaps because a single rulewill apply to a large number of actions) and that adjudication costs are low(perhaps because harm occurs only with a low probability or because violationsare enforced with a low probability/high sanction strategy, as suggested by Becker (1968) ). In this case, more precise laws tend to be desirable as long asinformation is not prohibitively costly for actors to obtain. The reason is that theactors have the option of not obtaining information. Those who do makeexpenditures to become informed will be only those for whom the expectedbehavioral benefit is large -- individuals whose acts will be significantlydesirable if they are not in fact very harmful but significantly undesirable if theyare of the more harmful type. For example, consider a complex rule scheme thatregulates routes over which hazardous substances may be transported. Mostdrivers -- those who never or rarely carry significant amounts of suchsubstances -- will ignore the rules. Those who regularly transport hazardousmaterials, however, will undertake the necessary effort to learn the rules andconform their behavior. Their information costs may be high, but the benefitsfrom better controlling their activity will be even larger.

One implication of this analysis is that the existence of high total complianceexpenditures is not an unambiguous sign of excessively detailed or complexrules. Low total expenditures may indicate that few individuals bother to learnthe details of the law, in which case any additional formulation or adjudicationcosts would largely be wasted. High total expenditures, on the other hand,reflect that many individuals contemplate actions the desirability of which iscontingent upon the type of their act; in this case, saving compliance costs byeliminating the law's precision may be a mistake because significant behavioralbenefits will be sacrificed. Another implication of the analysis is that widespreadnoncompliance need not indicate that rules are unduly complex; as long as thefew for whom the rules are most important do learn the details, the benefits ofprecision will be achieved at little cost with regard to others.

As a second case, suppose that private information costs are low (perhapsmost individuals already know the details of their actions and how they mightbe sanctioned) but that the cost of differentiating acts in adjudication would behigh (because it is hard for the state to learn the details of individuals' behavior).In this case, more refined legal commands may or may not be desirable. Forexample, additional costs may be repeatedly incurred in adjudication eventhough few individuals' behavior would be affected as a result of greater detailin the law. The distinction between this case and the foregoing one is that thestate must make additional expenditures on adjudication even when the ex antebenefit of improved behavior would be small, because the state will not knowwhen this is true; by contrast, when only ex ante information costs are high,private parties will incur them selectively.

4. Relationship of the Precision of the Law to the Accuracy of Adjudication

The precision of the law has a direct connection to the accuracy of adjudication,a topic explored in Kaplow (1994a) and Kaplow and Shavell (1994 , 1996a) . Theformer subject can be taken to refer to the precision of the law on the books andthe latter question to the precision with which the law actually is applied. At thetime they act, individuals will be induced to become more informed about thedetails of their situation only if, in a subsequent adjudication, their behavior willin fact be differentiated from that of others. If, instead, individuals' acts will beassigned to different legal categories largely at random or if no effort will bemade to refine the classification of acts, the situation is much as if the law itselfdid not make the refined distinctions in the first place. Many topics of procedure-- discovery, the use of experts, evidentiary rules, summary judgment, juries,appeals, burdens of proof -- and many substantive rules -- notably, concerningwhich categories of damages may be established and whether damages aredetermined by individualized evidence or by standardized tables -- may thus beilluminated by analysis similar to that presented here. A central concern whenassessing these features of adjudication is parties' incentives in litigation tomake expenditures on their own behalf: it often may be the case that parties, iffreely permitted to undertake further efforts in litigation, will spend excessivelybecause the private benefits from favorably influencing the outcome will exceedany social benefits from enhanced precision in applying legal rules.

5. Related Issues and Literature

The preceding discussion may be extended in a number of ways. Spier (1994) incorporates settlement bargaining. The interesting complication is thatapplication of a more precise rule in adjudication may make information that ispossessed by only one party relevant to the outcome. By thereby creating asituation involving asymmetric information, settlement becomes less likely.Thus, greater precision may be accompanied by increased litigation costs.

Kaplow (1995a) briefly considers costly sanctions -- monetary paymentsmade by risk-averse individuals and nonmonetary sanctions. Fewgeneralizations are possible. For risk-averse actors who do not become informed,greater differentiation may increase risk-bearing costs, although, as noted in Kaplow and Shavell (1996a) , more precise damage awards may reduce the risksborne by uninsured victims.

Kaplow (1990) examines the case in which some individuals know how thelaw differentiates among acts but others do not. In this case, there is no generaldeterrence rationale favoring lower sanctions for individuals who mistakenlycommit violations , but reducing their sanctions may be optimal when sanctionsare socially costly, because the lower rate of compliance by mistaken individualsresults in sanctions being imposed more often.

Kaplow (1995a) extends the analysis to regimes of self-reporting , as arecommon with much environmental and safety regulation as well as with manyforms of taxation. Self-reporting tends to be efficient when individuals'information costs are much lower than the government's; increasing precisionin a self-reporting regime has the confounding effect of inducing someindividuals to acquire information not to determine how better to behave butsolely to determine how best to report their conduct.

Most applied work on complexity and compliance costs is in the field of taxation . Empirical work is extensive. See, e.g., Long and Swingen (1987) , ArthurD. Little (1988) , Slemrod (1989) , Blumenthal and Slemrod (1992) . Analysis of thebenefits of precision in the tax context is qualitatively different from thatpresented here because the benefits of fine-tuning taxpayers' payments typicallyinvolve matters of distributive equity and minimizing adverse incentive effectsof taxation rather than of optimally controlling harm-causing behavior. See, e.g., Yitzhaki (1979) , Kaplow (1994c , 1996) . The most obvious simplifying devices inthe United States income tax code are the standard deduction (which taxpayersmay take instead of itemizing particular deductions) and the use of floors (whichforbid small amounts of deductions); they are analyzed in Kaplow (1994b) and Slemrod and Yitzhaki (1994) .

The discussion here, like much analysis in law and economics, emphasizesthe role of legal commands in providing incentives to govern parties' behaviorex ante. Modifications to the analysis would be required when adjudicationconcerns future behavior , as in the granting of licenses, enjoining harm-causingbehavior, incarcerating dangerous individuals, or determining eligibility forgovernment benefits. In particular, the benefit of precision in adjudication willbe greater than otherwise because the outcome will affect future activity directly(perhaps in addition to influencing ex ante incentives), and concerns aboutwhether individuals will learn the relevant details of the law may be less relevant.

Finally, it is worth noting that the benefits of more detailed rules are relevantto wholly private ordering , as in contractual arrangements (contingencies canbe distinguished more precisely, and provisions may provide reasonablycomplete specifications or be more open-ended) or in the internal organizationof firms. As a result, work on the economics of contracting and organizations,wholly apart from analysis of the law, can illuminate issues concerning thedesign of legal rules, and conversely.

B. Rules versus Standards

6. Rules versus Standards: The Problem

The choice between rules and standards has long received attention from legalcommentators and more recently has been addressed from an economicperspective. The presentation here will follow Kaplow (1992a) ; a number of thepoints, particularly concerning the effects of rules and standards on legal costsand on behavior, can also be found in the prior economic analyses by Ehrlichand Posner (1974) and Diver (1983) .

An initial obstacle in analyzing rules and standards involves matters ofdefinition. Usage varies greatly (which is no surprise since most dictionariesoffer definitions of each term that use the other), and logically distinct issues areoften combined or confused. For purposes of economic analysis, it is useful todefine the difference between rules and standards as involving exclusively thedistinction between whether the law is given content ex ante or ex post. Thus,a rule might list hazardous substances that may not be released into the watersupply whereas a standard may only proscribe releases of hazardoussubstances, leaving the determination of which substances are hazardous toadjudication, after releases have occurred.

This distinction between rules and standards is obviously one of degree,although it often is convenient to discuss it as an all-or-nothing choice. A lawis more rule-like the greater the extent to which it provides advance guidance: itmay, for example, provide a partial list or state criteria to be used, either of whichwill ease the application of the legal norm to particular behavior. Moreover, evena given formal specification will have a different character depending upon whatis understood about the mode of adjudication and upon what other informationis available in advance. For example, if everyone knows that an adjudicator islikely to rely upon a recent government study that indicates the degree of dangerposed by potentially hazardous substances, the situation will be almost the sameas that under a rule that incorporated the study's results. For present purposes,such a situation will be viewed as essentially rule-like, as the economic analysisdepends upon the extent to which content -- as understood by private partiesand adjudicators -- has actually been provided in advance, not upon thewording of legal commands. (It thus can be seen that the generation anddissemination of information, even when not explicitly embodied in a legal rule,is an important aspect of the law.)

Finally, it is important to distinguish the question of the degree to whichcontent is determined ex ante from the preceding subject, which concerns thedegree of precision with which a law is formulated or applied. Often,commentators assume that standards are inherently more precise or subtle thanrules because standards leave open the possibility of considering all possiblefactors whereas rules, no matter how detailed, provide a definitive resolution thatis not open to reconsideration. But this assumption implicitly contemplates thatadjudicators will devote virtually unlimited resources in applying standards.Such an outcome is irrational, implausible, and inconsistent with the actualoperation of legal systems. For example, tax codes and many safety regulationsare far more detailed and context-dependent than, say, the results that would beproduced by a jury that was presented with the open-ended question of whattax payment or safety regime was appropriate. As a result of law, custom, or theapplication of common sense, an adjudicator may well consider far less than allconceivably relevant factors. In an ordinary negligence dispute involvingmodest stakes, parties will not commission numerous expert studies; nor will theadjudicator commit substantial time to consider all possibilities.

Thus, it is a matter of choice how much detail will be incorporated in rulesand standards, and it is hardly the case that rule-like systems tend to beparticularly simple and standard-like schemes extremely precise in actualoperation. The main point for present purposes is that the question of how muchrefinement is optimal when formulating and applying the law -- legal precision-- is logically and, to a great extent, practically distinct from the question whethera given degree of precision is best specified in advance (rules) or left toadjudication (standards). In the end, one can combine the following analysis ofrules versus standards with the preceding analysis of the optimal precision ofthe law to determine what sort of legal command is best.

7. Effects of Rules and Standards on Legal Costs and on Behavior

Rules are defined as legal commands that provide greater specification inadvance. Hence, the initial cost of formulating rules will be greater than the costof formulating standards. It will be necessary, for example, to undertake studiesto determine which substances are hazardous and to determine appropriatethresholds of safety if a rule listing hazardous substances is to be promulgated.Such efforts can be avoided (for the moment) under a standard that merelyproscribes the release of hazardous substances and leaves their determinationto subsequent proceedings.

On the other hand, rules will be cheaper for private parties to interpret whendeciding upon their own conduct and for adjudicators to apply to past behavior.Precisely because more has been determined in advance, there will be less needfor parties to conduct their own studies and to predict legal outcomes and foradjudicators to devote effort to determine whether a violation should be deemedto have occurred. If a rule specifies which substances may be discharged, actorsand adjudicators need merely consult the list to apply a rule, whereas substantialinquiry may be necessary under a standard.

The same reasoning concerning the ease of application suggests that ruleswill tend to produce behavior more in conformity with legal norms than wouldequivalent standards. Because parties will be able to learn the governing normsmore cheaply and accurately in advance, they will tend to behave more in accordwith such norms. (This raises the question, explored in Kaplow and Shavell(1992) and noted above, whether private parties' incentives to learn about thelaw will be socially appropriate.) When guidance is easier to obtain, fewer partieswill be deterred from behavior that actually is not objectionable, because theywill have greater confidence that such behavior will not be subject to legalsanctions. More parties will be deterred from truly undesirable behavior becausethey are less likely to proceed in the mistaken belief that the prospect of beingsanctioned is remote. It is important to observe that the desirable effect of ruleson behavior may hold even when standards turn out to be more precise in actualadjudication, for any refinement embodied in the application of standards willinfluence behavior only to the extent parties are able to predict the relevantdetails in advance, and often parties will not make such precise predictions. (Thepoint that details tend to be helpful only when parties find it in their interest tolearn them was emphasized in the preceding section on the precision of legalrules.)

Combining these effects, one can see that rules tend to be preferable whenparticular activities are frequent, and standards do best when behavior varies sogreatly that any particular scenario is relatively rare. The reason is that the addedcost of rules, at the formulation stage, is borne once, whereas the relativebenefits of rules -- lower costs and greater compliance for private actors andlower costs of adjudication -- may arise often or almost never, as the case maybe. Thus, for dry cleaning and automotive fluids, it would be best to specifypermitted disposal techniques in advance, for the many small establishmentswould be unlikely to be able to determine the effects on their own and, even ifthey could, the costs of each obtaining information on the matter would begreat; also, redundant costs would be incurred in inevitable enforcementproceedings. But in the case substances that are rarely used, it may not beworthwhile to specify the range of permitted and prohibited disposal techniquesin advance because most scenarios involving potentially hazardous action maynever arise.

In this regard, it should be noted that the frequency of activity may be greateven if reported adjudications in a field are few. For example, rules might leadcountless individuals to comply with the law at very low cost. Even whencompliance is imperfect (or when harm nonetheless arises and is subject to aregime of strict liability), there may be little adjudication despite frequent activitybecause the probability of harm is low, detection of harm is unlikely, or mostlegal disputes are quickly settled. Conversely, disputes may appear to befrequent but not indicate that conditions are favorable to the use of rules. Forexample, there are many negligence disputes but some types may involve uniquesituations that, ex ante, were unlikely ever to arise.

Finally, the costs and benefits of rules and standards are important fordetermining the optimal effort to devote to formulating rules and to applyingrules and standards in adjudication. All else equal, the greater the frequency ofthe behavior to be governed, the more valuable it will be to formulate the lawwith greater care. Conversely, to the extent that an adjudication is relevant toonly a single case, it will not be very valuable to expend substantial resourcesto consider myriad factors in order to reach a precise determination -- unless aprecedent it to be set, as discussed below. That individuals may have difficultypredicting refined decisions with any accuracy will strengthen this point. Theseconsiderations reinforce the previous observation that, in fact, rules often willbe more precise and subtle than standards, taking into account how the latterwill actually be applied.

8. Precedent, Other Legal Interpretations, and the Evolution of LegalCommands

Precedent may be seen as the (perhaps partial) conversion of a standard into arule. Whether after accumulated experience or as a result of hearing a singlecase, a tribunal may issue a statement that will govern future adjudication. Atthat point, and to that extent, a rule will have been established. Similarly, evenless authoritative legal interpretations, ranging from discussions in judicialopinions to public statements of enforcement agencies, will tend to guide futurebehavior and adjudications to some degree. Once the analysis of standards --the state of affairs before the precedent -- and of rules -- after the precedent -- isunderstood, the study of precedent (and other legal interpretations) isstraightforward. A few particular factors, however, deserve some elaboration.

It is worth asking why the use of precedent is superior to an earlierannouncement of rules. After all, the primary burden of using a rule, incurringadditional formulation costs, is ultimately borne when the precedent isestablished. But in the interim, before the precedent is announced, the benefits-- of cheaper and better guidance of behavior and simpler adjudication -- areforgone. This point is particularly striking when an important, open legalquestion is identified many years or decades before a precedent is finallyannounced. The primary advantage of delay is that more information may becollected in the interim -- information about the optimal legal command or aboutthe importance of resolving an issue. But many precedents are announced withlittle more information than existed initially. Moreover, the most efficient way tocollect information may be to commission a study in the first place rather thanto wait until a case happens to appear before the relevant tribunal (when all thatis then learned is the single data point from that case or perhaps a handful ofdata points). Also, once more information does become available throughwhatever means, it is not clear why the legal system should not make anannouncement immediately rather than waiting still further.

It should be noted that making initial legal commands more rule-like does notimply that they cannot be changed if new information becomes available. Theprospect of change does reduce the benefit of rules, but it does not follow thatthere will be no benefit from providing guidance in the interim. Indeed, manydetailed regimes that are designed by regulatory agencies have a more rule-likecharacter, with occasional revision, and detailed statutes, such as tax codes, aremade precise but often revised later. (And, of course, precedents themselves areextended, narrowed, and reversed.) With standards, change may occur implicitly,but the prospect of change will accordingly increase the cost and difficulty ofprediction. Substantial developments that call for significantly different legalcommands may be apparent, making prediction under standards easier, but alsopresenting a clear case for adjusting rules. It is hardly the case that operativelegal commands in heavily rule-governed domains (like tax and environmentalregulation) and the interpretation of important precedents change slowlycompared, say, to applications of the negligence rule.

For prior economic analyses of precedent, emphasizing that the body ofprecedent can be analogized to a capital stock that depreciates over time, see Landes and Posner (1976) and Posner (1992) . See also Landes and Posner's(1994) discussion of the related subject of anticipatory adjudication, in which aparty may seek a legal decision before undertaking an activity, so as to avoid theneed to wait for a precedent.

9. Related Issues and Literature

Other law and economics work that is addressed directly to the subjectsconsidered herein includes Ogus (1981) (discussing rules formulated inquantitative terms), and Ogus (1992) (emphasizing private actors greater abilityto predict rules than standards).

The choice between rules and standards is closely related to the problem of legal uncertainty . Indeed, if rules are defined as the extent to which legalcommands are given content ex ante and thus the ease with which private partiescan predict how the law will apply to their conduct, legal uncertainty might beviewed as simply a measure of the extent to which a legal command isstandard-like. Uncertainty about the negligence rule has received the greatestattention in the literature. See Calfee and Craswell (1984) , Craswell and Calfee(1986) , and Shavell (1987) . Because the negligence rule has an all-or-nothing(discontinuous) character -- taking slightly more care will exonerate an actor fromwhat otherwise would be significant liability -- uncertainty about the legalstandard may tend to produce excessive care (although it is also possible thatgreat uncertainty will produce too little care). The suggestion that malpracticeliability results in overly defensive medicine is one example. For empiricalevidence, see Kessler and McClellan (1996) . Kahan (1989) shows that theseissues would not arise if negligent actors were not responsible for all the harmthey caused, but only for the incremental increase in harm due to theirnegligence. Also, in considering legal uncertainty, it would be appropriate toextend the present analysis and that in prior work on the negligence rule to takeexplicit account of risk-bearing costs.

As noted when addressing legal precision, the discussion here focuses onthe role of law in providing incentives to control parties' ex ante behavior. When future behavior (licenses, injunctions, incarceration, benefit eligibility) isinvolved, the analysis will differ. Indeed, the difference between ex ante and expost formulation may be moot, as a definitive resolution will in any eventprecede the relevant effect of a legal decision. There are, nonetheless, similarquestions about the benefits of wholesale decision-making, as in stating a rulefor a class of cases rather than adjudicating the same question repeatedly. Andthe foregoing analysis remains applicable to the extent adjudication willinfluence ex ante incentives in addition to regulating future behavior (such asin the case of incarceration).

There are other functions of law as well. Much of contract law and relatedfields is concerned with providing rules of form or background rules that leaveparties free to transact as they wish. See, e.g., Baird and Weisberg (1982) .Because the purpose of such law is different -- law is largely facilitating ratherthan constraining -- the analysis would have to be different. Nonetheless, itseems that in contexts where transactions are frequent, the benefit of greater exante formulation -- through rules -- would be larger.

Yet another concern with some bodies of law is to constrain fraud . In someinstances (notably, tax evasion) greater specificity in advance might beproblematic because parties will better be able to evade legal norms. Bothuncertainty and complexity, discussed previously, may be desirable preciselybecause parties will be less able to predict the law.

Apart from normative analysis, one can inquire into lawyers' and legalinstitutions incentives to formulate legal commands . In this regard, conflictsbetween private and social interest may be noted. For example, lower legal costssave social resources but may reduce the demand for lawyers. See, e.g., White(1992) . And legislation might be designed with an eye to attracting campaigncontributions rather than serving the public interest. See, e.g., Doernberg andMcChesney (1987) . Regulators may be more concerned with their own costs informulating and applying rules than in private costs. See, e.g., Diver (1983) .

Finally, the present discussion may be situated in the large literatures in theareas of legal philosophy and public choice which examine the rule of law andin particular address the use of rules to constrain official discretion. For areasonably comprehensive treatment of rules that attends to functionalconcerns, see Schauer (1991) .

C. Changes in Legal Rules: Transitions and Retroactivity

10. Transitions and Retroactivity: The Problem

Transition rules, including a determination of whether a law is to applyretroactively, concern not the form and content of substantive legal commandsbut instead how the problem of legal change is to be addressed. The mostcommon problem of legal change concerns the fact that investments and otheractions often have effects in the future, at which time the governing legal regimemay be different. Changes in the law will affect, positively or negatively as thecase may be, the returns from prior actions. This raises the question whetherlosses from legal change should be compensated or otherwise mitigated (asthrough grandfather provisions that exempt prior investments or phase-ins thatimplement the change only gradually) and, analogously, whether gains fromlegal change should be subject to windfall taxation or otherwise reduced.

As first analyzed carefully in the context of tax reform by Graetz (1977) , theseproblems arise even when new legislation is nominally prospective. For example,if a tax subsidy is repealed, a prior investment that benefited on a continuingbasis from the subsidy will become less valuable even if none of the priorsubsidy payments must be returned to the government. The possibility ofexplicit retroactivity does, however, expand the potential scope for transitionissues: even completed transactions might subsequently be rewarded orpenalized when laws change.

The general problem of transition, analyzed in Hale (1927) , Samuels (1974) ,and Kaplow (1986 , 1987 , 1992b) , arises much more widely and with more commonelements than is usually appreciated. Basically, whenever there is uncertaintyconcerning government policy -- about statutory change, new regulation orderegulation, court decisions, government spending priorities, takings of privateproperty for public use, tariffs, and so on -- the resolution of the uncertainty willcreate winners and losers. Thus, all government action (unless fully anticipated),government inaction (that is not fully anticipated -- that is, when change had apositive probability that was not realized), and even changes in the prospect foraction in the future will tend to affect the value of prior investments. Thepotentially affected investments include physical and financial assets,investments in human capital, and choices of location and occupation, amongothers.

As this general description suggests, an important aspect of the transitionproblem concerns the ex ante choice of private actors -- individuals and firms --under uncertainty. The prospect of legal change (like that of any other change)will affect both investment incentives and the imposition of risk, and theseincentive effects and risk-bearing costs will in turn be influenced by whethertransition relief is anticipated and provided. The immediately following analysisconsiders the question of optimal government transition policy, taking as giventhe actual change in the law that might occur and its desirability. The focus ison how transition relief affects the behavior and well-being of private partiesand, accordingly, its effect on social welfare. Subsequent discussion will take upconcerns about possible government misbehavior in changing the law and howdifferent transition policies may influence government activity.

11. Effects of Transitions and Transition Relief on Private Actors' Incentivesand on Risk-bearing Costs

Begin by considering the case in which private actors are risk-neutral (whichmay be approximately true for large, widely owned enterprises that are thesubject of much legal regulation). Suppose that there is some prospect, say, afifty percent chance, of an adverse legal change: a firm's product might bebanned if it turns out that the product is dangerous. Anticipating such apossibility -- and assuming for the moment that there will be no compensation,exemption, or other relief to those investing prior to the transition -- privateparties will invest less than they would otherwise. Half the time they will earnsome positive return and the other half of the time they will earn no return (and,indeed, they may lose their initial investment).

Now suppose instead that investors anticipate that there will be completetransition relief, say, full compensation. In this case, investors expect a positivereturn for sure, so they are more likely to invest -- or to invest at a larger scale --than if there were to be no relief. As Blume, Rubinfeld, and Shapiro (1984) firstshowed formally in the takings context, the prospect of such relief inducesinefficient overinvestment. The key point is that a private loss in such instancescorresponds to a real social loss. If a product is too dangerous and must bebanned, sunk investments that are valuable only for producing such a productwill be of no subsequent social value. Hence, insulating investors from suchrisks is inefficient, just as it would be inefficient to insulate investors from theprospect that their products might become obsolete or lose out to betterproducts of competitors. As noted by Hale (1927) , Samuels (1974) , Graetz (1977) ,and Kaplow (1986 , 1987 , 1992b) , from the perspective of investors, the risk offuture government action is similar to various market risks and natural risks(earthquakes, hurricanes). Private decisions are efficient when actors bear thefull costs and benefits of their actions. Transition relief thus does not correct anexternality; it creates one.

This analysis of transition relief may be extended to determine whenretroactive application of legal changes is efficient. See Kaplow (1986 , 1987 , 1992b) . If a product is banned because it is discovered to be dangerous, it isprobably the case that past production of the product resulted in danger as well.Thus, imposing retrospective liability will tend to be efficient, in order forinvestors ex ante to have had the proper incentives to take safety into account.By contrast, if a pollutant is to be regulated because it damages a newlyintroduced crop strain, there is no reason to impose retrospective liability, for thedecision to regulate the pollutant does not suggest that the pollutant previouslycaused some social harm. To generalize, when the government discovers newinformation about conditions that have long been in existence, retroactiveapplication in some form will often be necessary to provide the right ex anteincentives, but when there is a change in circumstances that makes future useundesirable without affecting the desirability of past conduct, retroactivity willbe inefficient.

Similar reasoning suggests that transition relief is inefficient even whenprivate actors are risk averse. See Kaplow (1986 , 1987 , 1992b) . In this his case,relief has the advantage of reducing risk-bearing costs. But private arrangements-- diversified ownership, insurance, contractual risk allocations, and the like --also may be available. Moreover, these mechanisms have systematic advantagesover transition relief. First, they often are implemented in ways that preserveincentives, in full or in part. For example, insurance premiums are greater themore assets are covered, so investors contemplating expansion in the face ofsome risk will bear the expected loss ex ante through higher insurance payments.In other cases, private risk mitigation is partial precisely to preserve incentives;absent other market imperfections (notably, adverse selection), privaterisk-sharing arrangements tend to be efficient. Transition relief generally lacksthese features. Second, even if transition relief were only partial, private partieswould undertake actions to further mitigate risk; such actions may worsen theirincentives (moral hazard), but some of the resulting cost would be borne by thegovernment. Again the analogy to market risks can be made: just as governmentrelief (as in the case of disaster relief) tends to be inefficient for ordinary risks,see Arnott and Stiglitz (1991) and Kaplow (1991) , so government relief againstgovernment-created risks tends to be inefficient with regard to privateinvestment and risk-mitigation incentives.

12. Related Issues

The analysis assumes that transition policy will be anticipated by investors andactually implemented by the government. To the extent such policy is embodiedin constitutional provisions or strong norms that guide legal reform, theassumption is plausible. More generally, a government is unlikely to succeed inthe long run in getting investors to expect one transition policy while oftenimplementing another. Problems of credible commitment may make it difficult toestablish a desired transition policy. To consider a familiar analogy, somegovernments regularly offer ex post relief when natural disasters strike, whichhas the effect of encouraging individuals to, say, invest excessively in buildingon flood plains. A possible solution to this political commitment problem is forthe government either to regulate such activity (limiting investment in high-riskareas) or to compel the purchase of insurance in advance at actuarially fair rates(the insurance premiums would discourage excessive investment and the factthat victims have paid for compensation in advance would eliminate the pressurefor ex post relief).

Most of the discussion here, like most in various literatures that addresstransition issues -- ranging from tax reform to deregulation -- emphasizes thecase in which investors suffer transition losses . But it was noted at the outsetthat transition gains arise with similar frequency. The analysis of bothpossibilities is symmetric: just as providing relief for losses will encourageoverinvestment, imposing windfall taxes on gains (or otherwise reducing them)will produce underinvestment. Few if any windfall gains -- or losses -- are reallypure windfalls. It tends to be best for investors to bear all the gains and lossesthey cause, with any risk mitigation being contractually arranged to provide forthe most effective limitation of adverse incentive effects.

Another question of transition policy concerns the means of relief. One cancompensate losers (and tax winners) in whole or in part, grandfather priorinvestors (that is, provide some exemption from new rules), or curtail the newpolicy (as by delay or phase-in or implementing reform only partially). Withregard to ex ante incentives and mitigation of risk, all of these approaches aresimilar, to the extent that they provide the same degree of effective relief. (Forexample, a phase-in that reduces gains and losses by half has effects like that offifty percent compensation and fifty percent windfall taxation.) With regard tofuture behavior, however, compensation and windfall taxation are generallysuperior to other forms of relief. Grandfathering, for example, might exemptexisting producers of dangerous products, allowing the danger to continue (butperhaps not to grow); delaying implementation would similarly permit somefuture harm. By contrast, immediate elimination of the danger accompanied bycompensation would avoid this difficulty.

As in the discussion of rule formulation, the present analysis is mostapplicable to laws that regulate behavior . Contract rules that serve as defaultsare quite different. If a default rule is changed, it may well be that the prior ruleshould govern preexisting contracts, for parties may have failed to make specialprovision in their contracts precisely because they knew they could rely uponbackground rules. On the other hand, if a default is changed because newanalysis suggests that most parties have a different preference than previouslysuspected, and if many contracts are silent simply because of the cost of makingspecial provision, it could be best to apply the new default to preexistingagreements.

13. Effects on Government Behavior of Requiring Transition Relief

The foregoing analysis takes as given a particular legal change, assumes that itis well conceived, and asks how transition relief affects private behavior. But itcannot be assumed that government always acts appropriately, and transitionrelief may affect the behavior of government.

One familiar argument is that transition relief -- in particular, requiringcompensation -- provides the government proper incentives because thengovernment bears the full costs of its actions. One can make the analogy to tortliability when both injurers' and victims' behavior affects accident costs. Foranalysis of this tort problem, see Shavell (1980) , and for application to thetransition context see Kaplow (1986 , 1987) ; for a similar discussion, see Blume,Rubinfeld, and Shapiro (1984) . No liability -- or schemes that provide nocompensation when injurers behave well, a negligence-type rule, or that denycompensation when victims behave improperly, contributory or comparativenegligence rules -- is best for victims' incentives (like investors' incentives,above), whereas full liability -- strict or otherwise -- is best for injurers'incentives. If one can assume that the government (the "injurer") alwaysbehaves properly, then a regime that denies compensation is best for theincentives of investors (the "victims"). But if one cannot make this assumptionabout government behavior, one may wish to impose liability on the government-- for example, by requiring compensation. Unless some tribunal can determinewhen the government or investors misbehaved, which is likely to be verydifficult in most transition contexts (and which involves courts second-guessinglegislative, executive, agency, or their own actions), it must be recognized thatany solution will be a compromise: if some relief is required, government'sincentives may improve but investors' incentives will be worse.

This argument, however, is incomplete. See Kaplow (1986 , 1987) . The implicitassumption is that the government, like a private investor, receives the benefitsfrom its actions. But the benefits of most government action -- banninghazardous products, improving tort rules, building highways -- are received byprivate citizens. Thus, making the legislature or an agency pay for transitionlosses (in addition to paying for direct costs, as in building a highway) when fewif any of the benefits are directly received might systematically skew decisionsagainst any change. Once benefits as well as costs are considered, it is notobvious that even a highly imperfect government is generally biased in favor ofchange when no compensation or other relief is required. For any bias -- whetherfor or against change -- to exist, there must be some asymmetry in thegovernment's treatment of costs and benefits. Some theories of governmentaction, see Downs (1957) and Olson (1965) , suggest that the most effectivepolitical interest groups are those on whom gains or losses are concentrated.Other theories are concerned with the majority exploiting the minority. Stillothers, see Niskanen (1971) , emphasize bureaucrats desire to enhance theirbudget or prestige. Thus, whether requiring compensation would makegovernment behavior better or worse depends upon which theories ofgovernment behavior are most plausible, whether it is the benefits or the coststhat are more concentrated, and other factors. See also Quinn and Trebilcock(1982) .

Another concern with government behavior involves the problem of pure expropriation . See Kaplow (1987) . It is familiar from public economics that themost efficient government tax is a one-time capital levy, in which capital (morebroadly, savings, assets in any form) is taken and the government promisesnever to repeat the action. As long as the expropriation in wholly unanticipated,there will be no adverse ex ante incentive effects; likewise, if it will never berepeated, there will be no future distortions in behavior. Levmore (1993) offersthis as a basis for a more favorable view toward retroactive taxation. But it haslong been recognized that governments would find it difficult to engage in suchinconsistent behavior. Rather, there needs to be some commitment -- whetherthrough constitutional means or long-established norms -- that the governmentwill not expropriate, and governments that deviate from such a path will beunlikely to be able to provide investors with confidence for the future.

In principle, expropriation can be distinguished from other governmentaction. As noted when addressing private incentives, it is efficient for investorsto anticipate and respond to the prospect that new information or changedcircumstances will affect what government regulation is optimal, just as newdevelopments will influence the market directly in ways that investors shouldtake into account. Anticipating expropriation, however, is inefficient ex ante. Thebenefit arises from transferring resources to the government, not from providingproper incentives for private behavior. To be sure, some taxation is required, tofund public goods and services and to accomplish redistribution. Particularlywith regard to redistribution, it would improve efficiency if taxpayers did notanticipate having to pay taxes, as such anticipation results in distortion ofbehavior. But this problem is largely unavoidable.

Successful economies seem to have succeeded, for the most part, in avoidingexpropriation. The mechanism by which this is accomplished is not that clear;no system prohibits taxation or legal changes, and constitutional or other strongcompensation requirements are limited. (For example, in the United States, thegovernment must compensate a landowner whose building is taken for ahighway, but a high capital income or wealth tax is not legally proscribed.) Thedevelopment of strong norms seems to play a greater role in successfuleconomies, but such norms cannot succeed unless the relevant audience canunderstand which legal reforms are improper in the relevant respect.

A variant of the expropriation problem arises when the government isinvolved with ordinary contractual behavior and related activity. In fact, mostgovernments accomplish many of their purposes -- building roads, procuringmilitary equipment, hiring employees -- through ordinary market transactions. (Insome countries, state enterprises are common; these too, however, often useordinary contractual means of paying workers and purchasing other necessaryinputs.) Moreover, most governments, despite the availability of sovereignimmunity from private suits, voluntarily subject themselves to a legal regimeunder which their contracts having a binding effect similar to that of privatecontracts. This avoids problems of opportunism, both large and small.

Some commentators have analogized legislation and other governmentlaw-making to government contracts. See, e.g., Graetz (1977) , Kaplow (1986 , 1987) , Ramseyer and Nakazato (1989) , Logue (1996) . Thus, sometimes thegovernment uses subsidies to induce certain behavior rather than explicitcontracts under which it pays for a promise to undertake certain activity. If thesubsidy is terminated, this might be viewed as a breach of an implicit contract.Of course, subsidies are routinely terminated. When governments use ongoingsubsidies to encourage an activity rather than entering into a long-term contractto pay for a given amount of the activity, they are choosing to use a moreflexible mechanism, allowing the possibility of change without compensation. Toimpose a norm or constitutional requirement that all law be treated as if it werea binding contract would not add to the government's options; it wouldeliminate one. (Actually, it may have no effect at all, as laws could simply bewritten to expire unless extended or with some other loophole providing thegovernment a way to avoid any binding obligation.)

14. Additional Issues

Transition relief is sometimes favored because it may make it politically feasibleto implement desirable reform. For example, efficient deregulation might beopposed by protected incumbents, who could be bought out withcompensation. See Tullock (1978) ; see also Usher (1995) (addressingcompensation for takings). The problem is that, ex ante, special interest groupswould have a greater incentive to seek inefficient rules, knowing that even if therules were subsequently repealed they would benefit from transition relief. See Kitch (1977) , Quinn and Trebilcock (1982) , McKenzie (1986) , Kaplow (1987 , 1992b) . Thus, it is not obvious that a policy of buying out such politicalopposition would in the long run produce better or worse laws and lead to moreor less resources being wasted on rent-seeking activity.

Another problem concerns changes in political power that may produce legalchanges that do not reflect real changes in information or circumstances(although legal reforms may reflect shifts in perceptions or in value judgments).Small changes in beliefs may produce large legal changes (as when anopposition party gains a few percentage points in support and comes intopower), and such change may be frequent. Ex ante behavior that is desirablefrom the perspective of one party's policies may be undesirable from theperspective of the other's. Use of transition relief will tend to cause investors toput more weight on the status quo, something unlikely to be favored by asubsequent regime seeking change. Analysis of social welfare in such instancesis difficult.

15. Literature

The first careful analysis of many issues concerning legal transitions appears tobe Hale (1927) . For a general analysis of changing legal rules in a variety ofcontexts, see Kaplow (1986 , 1987 , 1992b) .

Most writing by legal academics on the subject focuses on takings or on taxreform. With respect to takings, Michelman (1967) offers an important earlyanalysis, although it does not take the economic approach as far as subsequentwork. Baxter and Altree (1972) advance a "first-in-time" approach to address theproblem of airport noise; they are among the first to pursue directly the problemof the adverse incentive effects of compensation. Blume, Rubinfeld, and Shapiro(1984) analyze incentive effects of compensation in a formal model and Blumeand Rubinfeld (1984 ) consider risk, although these authors do not integrate thetwo analyses or systematically consider private risk mitigation. Kaplow (1986 , 1987) discusses takings in the context of analyzing the transition problem moregenerally. For additional work, with emphasis on the problem of thegovernment's incentives, see, e.g., Fischel and Shapiro (1989) , Hermalin (1995) .For an empirical analysis of the effect of requiring compensation on governmentbehavior, see Cordes and Weisbrod (1979) .

Feldstein (1976a , 1976b) and Graetz (1977) offer the first substantialtreatments of tax transitions, followed by Kaplow (1986 , 1987 , 1992b) , who moreexplicitly analyzes effects of relief on investor's incentives and private riskmitigation, compares alternative transition mechanisms, and addressesinstitutional problems with government decision-making. This latter problem isalso considered by Ramseyer and Nakazato (1989) and Logue (1986) , who areboth favorable toward relief because of concerns about governmentmisbehavior. Levmore (1993) is more favorable to (uncompensated) retroactivitybecause of the possibility of an efficient capital levy. The particular problem oftransition to a consumption tax is addressed in Graetz (1979) , Kaplow (1986 , 1987 , 1995b) , Sarkar and Zodrow (1993) , and Bradford (1996) .

Another subject receiving increased attention is deregulation and otherregulatory changes. See Quinn and Trebilcock (1982) , Kolbe, Tye, and Myers(1993) , Kaplow (1993) .

D. Additional Topics

16. Additional Topics

There are a number of other dimensions along which many legal rules vary thatare sufficiently general to warrant independent study. Because these subjectsare addressed in other fields within law and economics, the present discussionwill be brief. Each of the topics considered involves the general questionwhether laws should identify certain conduct as permitted or prohibited orshould instead impose fines, provide for private liability, or use other means ofaligning private and social interests so as to induce parties to make efficientchoices on their own.

Strict Liability Versus Negligence . The choice between strict liability andnegligence has received much attention. Similar issues arise involving defensesbased upon victim behavior. And other areas of law present analogousquestions: namely, is a party responsible for all harm caused, or is the partyresponsible only if behavior was improper (and, then, only to the extent of anyadditional harm caused by the inappropriate action)? The latter type of lawrequires both defining a norm of behavior (which may be done ex ante, as a rule,or ex post, as a standard, and which may be done in a simple or precise manner),and both types of law require provisions governing the measurement of harm.This sort of choice involves many of the issues considered here, as well asconcerns for the frequency of litigation, victims' incentives, and other factors.See Brown (1973) , Shavell (1980) , and the entry on strict liability versusnegligence in tort.

Property Versus Liability Rules. The distinction between property and liabilityrules involves remedies: if one party wishes to infringe the legal rights ofanother, must the acting party first obtain permission (or else be subject toextreme sanctions and the possibility of injunction, if feasible) -- a property rule-- or may the party simply act and pay damages -- a liability rule? This schemeis first explored in detail by Calabresi and Melamed (1972) and has subsequentlybeen expanded in Kaplow and Shavell (1996b) . Other relevant work includes Ellickson (1973) , Polinsky (1979 , 1980) , Krier and Schwab (1995) . The problem ofbargaining under asymmetric information has attracted disproportionateattention, beginning with Samuelson (1985) , followed, e.g., by Farrell (1987) , Shavell (1988b) , Johnston (1995) , Ayres and Talley (1995a , 1995b) , Kaplow andShavell (1995 , 1996b) . See the entry on property versus liability rules.

Acknowledgments

I am grateful for comments from Steven Shavell and the referees, researchassistance from Enrique Arana, Judson Berkey, and Jerry Fang, and supportfrom the John M. Olin Center for Law, Economics, and Business at Harvard LawSchool.

Bibliography on General Characteristics of Rules (9000)

Arthur D. Little (1988) , Development of Methodology for Estimating the TaxpayerPaperwork Burden, Final Report to the Department of the Treasury , Washington, D.C.,Internal Revenue Service.

Baird, Douglas G. and Weisberg, Robert (1982) , 'Rules, Standards and the Battle of theForms: A Reassessment of ¤2-207', 68 Virginia Law Review , 1217-1262.

Baxter, William F. and Altree, Lillian R. (1972) , 'Legal Aspects of Airport Noise', 15 Journal of Law and Economics , 1-113.

Blume, Lawrence and Rubinfeld, Daniel L. (1984) , 'Compensation for Takings: AnEconomic Analysis', 72 California Law Review , 569-628.

Blume, Lawrence, Rubinfeld, Daniel L. and Shapiro, Perry (1984) , 'The Taking of Land.When Should Compensation be Paid?', 99 Quarterly Journal of Economics , 71-92.

Blumenthal, Marsha and Slemrod, Joel B. (1992) , 'The Compliance Cost of the U.S.Individual Income Tax System: A Second Look After Tax Reform', 45 National TaxJournal , 185-202.

Bouckaert, Boudewijn and Schäfer, Hans-Bernd (1995) , 'Mistake of Law and the Economicsof Legal Information', in Bouckaert, Boudewijn and De Geest, Gerrit (eds.), Essays inLaw and Economics II: Contract Law, Regulation, and Reflections on Law &Economics , Antwerpen, Maklu, 217-245.

Bradford, David F. (1996) , 'Consumption Taxes: Some Fundamental Transition Issues', inBoskin, Michael J. (ed.), Frontiers of Tax Reform , Stanford, Hoover Institution Press.

Calfee, John E. and Craswell, Richard (1984) , 'Some Effects of Uncertainty on Compliancewith Legal Standards', 70 Virginia Law Review , 965-1003.

Cordes, Joseph J. and Weisbrod, Burton A. (1979) , 'Governmental Behavior in Response toCompensation Requirements', 11 Journal of Public Economics , 47-58.

Coulange, Pierre (1990) , 'La Production de Droit (The Production of Legal Rules)', 3 Journal des Economistes et des Etudes Humaines , 277-302.

Coulange, Pierre (1990), Analyse Économique de la Production de Droit (EconomicAnalysis of the Production of Legal Rules], Ph.D., Aix-en-Provence, Universitéd'Aix-Marseille III, Faculté d'Economie Appliquée).

Craswell, Richard and Calfee, John E. (1986) , 'Deterrence and Uncertain Legal Standards', 2 Journal of Law, Economics, & Organization , 279-303.

Diver, Colin S. (1983) , 'The Optimal Precision of Administrative Rules', 93 Yale LawJournal , 65-109.

Doernberg, Richard L. and McChesney, Fred S. (1987) , 'On the Accelerating Rate andDecreasing Durability of Tax Reform', 71 Minnesota Law Review , 913-962.

Ehrlich, Isaac and Posner, Richard A. (1974) , 'An Economic Analysis of Legal Rulemaking', 3 Journal of Legal Studies , 257-286.

Feldstein, Martin (1976a) , 'On the Theory of Tax Reform', 6 Journal of Public Economics ,77-104.

Feldstein, Martin (1976b) , 'Compensation in Tax Reform', 29 National Tax Journal ,123-130.

Fisch, Jill E. (1997) , 'Retroactivity and Legal Change: An Equilibrium Approach', 110 Harvard Law Review , 1055-1123.

Fishel, William A. and Shapiro, Perry (1989) , 'A Constitutional Choice Model ofCompensation for Takings', 9 International Review of Law and Economics , 115-128.

Graetz, Michael J. (1977) , 'Legal Transitions: The Case of Retroactivity in Income TaxRevision', 126 University of Pennsylvania Law Review , 47-87.

Graetz, Michael J. (1979) , 'Implementing a Progressive Consumption Tax', 92 Harvard LawReview , 1575-1661.

Graetz, Michael J. (1985) , 'Retroactivity Revisited', 98 Harvard Law Review , 1820-1841.

Hale, Robert L. (1927) , 'Value and Vested Rights', 27 Columbia Law Review , 523-529.

Hermalin, Benjamin E. (1995) , 'An Economic Analysis of Takings', 11 Journal of Law,Economics, & Organization , 64-86.

Hutter, Michael (1989) , Die Produktion von Recht. Eine selb streferentielle Theorie derWirtschaft, angewandt auf den Fall des Arzneimittelpatentrechts (The Production ofLaw. A Selfreferential Theory of the Economy, applied to the Case of PharmaceuticalPatent Law), Tübingen, J.C.B. Mohr, p. 213.

Kahan, Marcel (1989) , 'Causation and Incentives to Take Care Under the Negligence Rule', 18 Journal of Legal Studies , 427-447.

Kaplow, Louis (1986) , 'An Economic Analysis of Legal Transitions', 99 Harvard LawReview , 509-617.

Kaplow, Louis (1987) , Optimal Transition Policy: Replacing Horizontal Equity With an ExAnte Incentives Perspective , dissertation accepted by Harvard University Departmentof Economics, Cambridge, MA.

Kaplow, Louis (1990) , 'Optimal Deterrence, Uninformed Individuals, and AcquiringInformation About Whether Acts are Subject to Sanctions', 6 Journal of Law,Economics and Organization , 93-128.

Kaplow, Louis (1992a) , 'Rules versus Standards: An Economic Analysis', 42 Duke LawJournal , 557-629.

Kaplow, Louis (1992b) , 'Government Relief for Risk Associated with Government Action', 94 Scandinavian Journal of Economics , 525-541.

Kaplow, Louis (1993) , 'An Ex Ante Perspective on Deregulation, Viewed Ex Post', 15 Resource and Energy Economics , 153-173.

Kaplow, Louis (1994a) , 'The Value of Accuracy in Adjudication: An Economic Analysis', 23 Journal of Legal Studies , 307-401.

Kaplow, Louis (1994b) , 'The Standard Deduction and Floors in the Income Tax', 50 Tax LawReview , 1-31.

Kaplow, Louis (1994c) , "Accuracy, Complexity, and the Income Tax", Working Paper No.4631, Cambridge, MA, National Bureau of Economic Research.

Kaplow, Louis (1995a) , 'A Model of the Optimal Complexity of Legal Rules', 11 Journalof Law, Economics, & Organization , 150-163.

Kaplow, Louis (1995b) , 'Recovery of Pre-Enactment Basis Under a Consumption Tax: TheUSA Tax System', 68 Tax Notes , 1109-1118.

Kaplow, Louis (1996) , 'How Tax Complexity and Enforcement Affect the Equity andEfficiency of the Income Tax', 49 National Tax Journal , 135-150.

Kaplow, Louis and Shavell, Steven (1992) , 'Private Versus Socially Optimal Provision of ExAnte Legal Advice', 8 Journal of Law, Economics, & Organization , 306-320.

Kaplow, Louis and Shavell, Steven (1994) , 'Accuracy in the Determination of Liability', 37 Journal of Law and Economics , 1-15.

Kaplow, Louis and Shavell, Steven (1996a) , 'Accuracy in the Assessment of Damages', 39 Journal of Law and Economics , 191-210.

Kessler, Daniel and McClellan, Mark (1996) , 'Do Doctors Practice Defensive Medicine?', 111 Quarterly Journal of Economics , 352-390.

Kitch, Edmund W. (1977) , 'Can We Buy Our Way Out of Harmful Regulation?', in Martin,Donald L. and Schwartz, Warren F. (eds.), Deregulating American Industry: Legal andEconomic Problems , Lexington, MA, Lexington Books.

Kolbe, A. Lawrence, Tye, William B. and Myers, Stewart C. (1993) , Regulatory Risk:Economic Principles and Applications to Natural Gas Pipelines and Other Industries ,Boston, Kluwer Academic Press.

Landes, William M. and Posner, Richard A. (1976) , 'Legal Precedent: A Theoretical andEmpirical Analysis', 19 Journal of Law and Economics , 249-307.

Landes, William M. and Posner, Richard A. (1994) , 'The Economics of AnticipatoryAdjudication', 23 Journal of Legal Studies , 683-719.

Levmore, Saul (1993) , 'The Case for Retroactive Taxation', 22 Journal of Legal Studies ,265-307.

Logue, Kyle D. (1996) , 'Tax Treatments, Opportunistic Retroactivity, and the Benefits ofGovernment Precommitment', 94 Michigan Law Review , 1129-1196.

Long, Susan B. and Swingen, Judyth A. (1987) , 'An Approach to the Measurement of TaxLaw Complexity', 8 Journal of the American Tax Association , 22-36.

MacKaay, Ejan (1979) , 'Les notions floues ou l'économie de l'imprécision (Fuzzy Conceptsor the Economics of Imprecision)', 12 Langages , 33-50.

MacKaay, Ejan (1980) , 'Le Nozione "Fluide" Ovvero l'Economia Dell'Imprecisione (FuzzyConcepts or the Economics of Imprecision)', Informatica e Diritto , 253-274.

McKenzie, Richard B. (1986) , 'Tax/Compensation Schemes: Misleading Advice in aRent-Seeking Society', 48 Public Choice , 189-194.

Michelman, Frank I. (1967) , 'Property, Utility and Fairness: Comments on the EthicalFoundations of "Just Compensation" Law', 80 Harvard Law Review , 1165-1258.

Munzer, Stephen R. (1977) , 'Retroactive Law', 6 Journal of Legal Studies , 373-397.

Ogus, Anthony I. (1981) , 'Quantitative Rules and Judicial Decision Making', in Burrows, Pauland Veljanovski, Cento G. (eds.), The Economic Approach to Law , London,Butterworths, 210-225.

Ogus, Anthony I. (1992) , 'Information, Error Costs and Regulation', 12 International Reviewof Law and Economics , 411-421.

Posner, Richard A. (1992) , Economic Analysis of Law (4th ed.) , Boston, Little, Brown andCompany.

Quinn, John and Trebilcock, Michael J. (1982) , 'Compensation, Transition Costs, andRegulatory Change', 32 University of Toronto Law Journal , 117-175.

Ramseyer, J. Mark and Nakazato, Minoru (1989) , 'Tax Transitions and the ProtectionRacket: A Reply to Professors Graetz and Kaplow', 75 Virginia Law Review , 1155-1175.

Russell, A.M. and Rickard, J.A. (1987) , 'A Model of Tax Evasion Incorporating IncomeVariation and Retroactive Penalities', 26 Australian Economic Papers , 254-264.

Samuels, Warren J. (1974) , 'Commentary: An Economic Perspective on the CompensationProblem', 21 Wayne Law Review , 113-134.

Sarkar, Shounak and Zodrow, George R. (1993) , 'Transitional Issues in Moving to a DirectConsumption Tax', 46 National Tax Journal , 359-376.

Schauer, Frederick (1991) , Playing by the Rules: A Philosophical Examination ofRule-Based Decisionmaking in Law and in Life , New York, Oxford University Press.

Schwartz, Gary T. (1983) , 'New Products, Old Products, Evolving Law, Retroactive Law', 58 New York University Law Review , 796-852.

Shavell, Steven (1988a) , 'Legal Advice About Contemplated Acts: The Decision to ObtainAdvice, Its Social Desirability, and the Protection of Confidentiality', 17 Journal ofLegal Studies , 123-150.

Shavell, Steven (1988b) , 'Property Rights and the Rule of Liability in a Simple BargainingModel', unpublished manuscript, on file with the Harvard Law School Library,Cambridge, MA.

Skogh, Göran (1985) , 'Rättssäkerhet I Marknadsekonomin (Legal Certainty in the MarketEconomy)', in Rättssäkerhet och Demokrati , Ratios förlag.

Slemrod, Joel B. (1989) , 'Complexity, Compliance Costs, and Tax Evasion', in Roth, JeffreyA. and Scholz, John T. (eds.), Taxpayer Compliance, Volume 2: Social SciencePerspectives , Philadelphia, University of Pennsylvania Press.

Slemrod, Joel B. and Yitzhaki, Shlomo (1994) , 'Analyzing the Standard Deduction as aPresumptive Tax', 1 International Tax and Public Finance , 25-34.

Slot, P.J. (1983) , Regelen en ontregelen. Oratie Leiden (Regulating and De-regulating),Kluwer, Deventer.

Spier, Kathryn E. (1994) , 'Settlement Bargaining and the Design of Damage Awards', 10 Journal of Law, Economics, & Organization , 84-95.

Tullock, Gordon (1978) , 'Achieving Deregulation--A Public Choice Perspective', 2 Regulation , 50-54.

Usher, Dan (1995) , 'Victimization, Rent-Seeking and Just Compensation', 83 Public Choice ,1-20.

Walz, W. Rainer (1988) , 'Rechtssicherheit und Risikozuweisung bei Steuerrechtsänderungen- Verbotene Rückwirkung, gebotene Übergangsregelung, richterliche Vertragsanpassung'[Legal Certainty and Risk Allocation for Changes in Tax Law - Illegal Retroaction,Transitional Rules], Recht und Risiko , 252-286.

White, Michelle J. (1992) , 'Legal Complexity and Lawyers' Benefit from Litigation', 12 International Review of Law and Economics , 381-395.

Yitzhaki, Shlomo (1979) , 'A Note on Optimal Taxation and Administrative Costs', 69 American Economic Review , 475-480.

Other References

Arnott, Richard and Stiglitz, Joseph E. (1991) , 'Moral Hazard and Nonmarket Institutions:Dysfunctional Crowding Out or Peer Monitoring?', 81 American Economic Review ,179-190.

Ayres, Ian and Talley, Eric (1995a) , 'Solomonic Bargaining: Dividing a Legal Entitlementto Facilitate a Coasean Trade', 104 Yale Law Journal , 1027-1117.

Ayres, Ian and Talley, Eric (1995b) , 'Distinguishing Between Consensual and NonconsensualAdvantages of Liability Rules', 105 Yale Law Journal , 235-253.

Becker, Gary S. (1968) , 'Crime and Punishment: An Economic Approach', 76 Journal ofPolitical Economy , 169-217.

Brown, John (1973) , 'Toward an Economic Theory of Liability', 2 Journal of LegalStudies , 323-350.

Calabresi, Guido and Melamed, A. Douglas (1972) , 'Property Rules, Liability Rules, andInalienability: One View of the Cathedral', 85 Harvard Law Review , 1089-1128.

Downs, Anthony (1957) , An Economic Theory of Democracy , New York, Harper.

Ellickson, Robert C. (1973) , 'Alternatives to Zoning: Covenants, Nuisance Rules, and Finesas Land Use Controls', 40 University of Chicago Law Review , 681-781.

Farrell, Joseph (1987) , 'Information and the Coase Theorem', 1 Journal of EconomicPerspectives , 113-129.

Johnston, Jason Scott (1995) , 'Bargaining Under Rules Versus Standards', 11 Journal of Law,Economics, & Organization , 256-281.

Kaplow, Louis (1991) , 'Incentives and Government Relief for Risk', 4 Journal of Risk andUncertainty , 167-175.

Kaplow, Louis and Shavell, Steven (1995) , 'Do Liability Rules Facilitate Bargaining? AReply to Ayres and Talley', 105 Yale Law Journal , 221-233.

Kaplow, Louis and Shavell, Steven (1996b) , 'Property Rules Versus Liability Rules: AnEconomic Analysis', 109 Harvard Law Review , 713-790.

Krier, James E. and Schwab, Stewart J. (1995) , 'Property Rules and Liability Rules: TheCathedral in Another Light', 70 New York University Law Review , 440-483.

Niskanen, William A. (1971) , Bureaucracy and Representative Government , Chicago,Aldine, Atherton.

Olson, Mancur (1965) , The Logic of Collective Action; Public Goods and the Theory ofGroups , Cambridge, MA, Harvard University Press.

Polinsky, A. Mitchell (1979) , 'Controlling Externalities and Protecting Entitlements:Property Right, Liability Rule, and Tax-Subsidy Approaches', 8 Journal of LegalStudies , 1-48.

Polinsky, A. Mitchell (1980) , 'Resolving Nuisance Disputes: The Simple Economics ofInjunctive and Damage Remedies', 32 Stanford Law Review , 1075-1112.

Samuelson, William (1985) , 'A Comment on the Coase Theorem', in Roth, Alvin E. (ed.), Game-Theoretic Models of Bargaining , New York, Cambridge University Press.

Shavell, Steven (1980) , 'Strict Liability Versus Negligence', 9 Journal of Legal Studies ,1-25.

Shavell, Steven (1987) , Economic Analysis of Accident Law , Cambridge, MA, HarvardUniversity Press.