ArticlesVol. 1, No. 2

Legal Transitions and the Problem of Reliance

David M. Hasen

Professor of Law, Penn State University
DOI: 10.7916/D8KM0JP3
Citation:1 Colum. J. Tax L. 120


This Article analyzes the literature on legal transitions. The principal focus is taxation, but the analysis generalizes to other areas. I argue that the theoretical apparatus developed by scholars active in the legal transitions area suffers from significant conceptual shortcomings. These shortcomings include the unwarranted assimilation of legal to factual change, the naturalization of conventional arrangements, and the disregard of the distinction between making law and finding it. As a consequence, the recent literature offers an analysis that is unable either to explain actual transitions or to provide an adequate theory of how legal change should take place. In the end, the older view of legal transitions is more capable than the newer one of providing an adequate normative and positive framework for understanding legal transitions.


    • A. In General
    • B. The Case for Immediately Effective, Uncompensated Transitions
    • C. Nominal Retroactivity
    • A. Entitlement Rules and Transition Rules
    • B. Implications of the Entitlement Rule-Transition Rule Dichotomy
      • 1. Transitions vs. Transition Rules
      • 2. Relevant Anticipations
      • 3. Implied vs. Express Transition Rules
      • 4. Scientific Developments Are Not Legal Transitions
    • A. Analogy of Legal Transitions to Factual Developments
    • B. Problems With a Single Transition “Norm”
      • 1. Retroactive Relief—Naturalization of the Market
      • 2. Announcement Date Relief—New Interpretations As New Law
    • C. The Conventional Nature of Legal Regimes
    • A. The New View Critique of Reliance
    • B. The Positive Case for Reliance
      • 1. First Case: Change in Material Circumstances
      • 2. Second Case: Change in Legislative Judgment
        • (a) First Variation: Same Elected Body
        • (b) Second Variation: Subsequently Elected Body
    • C. The Linguistic Basis of Reliance



The academic literature on legal transitions has grown substantially over the last twenty-five years.[1] Its subject is the proper timing of legal change: When a new legal rule is announced, what rules or norms should govern the time at which the rule comes into force? Further, what relief, if any, should be made available to those who suffer an unexpected loss from the change, and what charge, if any, should be imposed upon those who enjoy an unanticipated windfall from it?[2] If Congress passed legislation tomorrow raising the top marginal tax bracket to seventy percent, should that rate apply only to income earned from tomorrow forward; only to income earned from some later date forward; or to some income already earned as well as to income yet to be earned? Further, would it be appropriate to offer relief to taxpayers who planned the timing of their income on the assumption that the old marginal brackets would persist for some period of time?

One of the notable features of the recent transitions literature is the consequentialist orientation it has adopted in answering these questions.[3] Rather than focus on the fairness or reliance aspects of various possible transition rules, scholars currently active in the area have typically adopted an anticipations-oriented approach that focuses on the efficiency consequences of various possible transition norms. From this perspective, the transition question becomes: What set of expectations about the effective date of future, as yet unknown changes to legal rules and about possible “transition relief” produces socially optimal pre-transition conduct among individuals who will be affected by such changes? The general answers upon which scholars writing in the area have settled are that norms of immediate effect (or, in some cases, earlier) and no transition relief produce optimal results.[4]

Consistent with this orientation, the recent literature has taken as broad a perspective as possible on the scope of the transition question.[5] When the focus is on expectations, the transition question arises in any setting in which norms regarding the timing of changes to legal entitlements may have an effect on pre-change expectations and, hence, conduct. These settings include not only changes to positive law, but also judicial reinterpretation of the law, administrative rule making, and even announcements of administrative policy, such as, for example, the Federal Reserve Chair’s pronouncements on the health of the economy.[6] In each, governmental action produces a change to legal entitlements (or to expectations regarding legal entitlements), and in each, expectations about the effective date of any such future change have an impact on the prechange conduct of affected parties. Thus, one commentator has defined the term legal transition as the resolution of any uncertainty regarding future legal rules to the extent that the uncertainty affects decisions made currently.[7] Another has defined it as any change to legal entitlements.[8]

In this Article, I argue that the conceptual apparatus that supports the approach just described is flawed, and that the transitions analysis that rests on it is in many ways unpersuasive as a consequence.[9] Because the literature begins by assimilating all manner of changes to entitlements to a single model of quasi-factual change, it elides distinctions that are needed to explain background assumptions that the literature tacitly makes but cannot support. These assumptions are both necessary for, but inexplicable within, a model that understands legal transitions as closely akin to market based or natural change. If the assumptions are abandoned, the market model of legal change on which so much of the transition analysis rests collapses. If the assumptions are retained, the distinctions among the various types of legal change that the recent literature tends to dismiss need to be retained, and if these are retained, many of the tenets of the older view discarded under the newer view regain their vitality.

As important as what I argue below is what I do not argue. I do not claim that the “new view” transitions literature has advanced wildly implausible normative or empirical claims, or that there is nothing to be gained from the anticipations-oriented approach. Nor, conversely, do I suggest that proponents of the “old,” reliance-based view identified the correct set of transition norms. Rather, the claims here are that the current literature is analytically deficient, to some extent schizophrenic, and largely unable to generate defensible propositions about transition norms that are different from those of, and not better understood within, the “old view” of legal transitions that it rejects. The analytical deficiencies include a failure to develop consistent definitions of such key concepts as “transition,” “legal transition,” and “transition norm”; the schizophrenia consists in the reliance of “new view” theorists on some of the same transition practices that they criticize; and the literature’s inability to generate novel and defensible claims is evidenced by its failures to account for the differing transition defaults operative in the adjudicative and legislative arenas, and to demonstrate that a reliance-based view is unable to reach correct results on the question of the proper default norms for changes to positive law and for judicial decisions that adopt new rules.

Part II sets out a brief history of the of the transitions literature and describes the currently dominant “new view” approach to legal transitions. Part III examines the assumptions on which the new view relies and reformulates that view in a way that makes explicit its conceptual underpinnings; the principal purpose of this exercise is neither to defend nor to criticize the new view, but instead to offer an account that adequately operationalizes it. Part IV offers a criticism of the new view account of legal change as so reformulated, and Part V extends the criticism to the arguments that new view scholars have made against reliance as a basis for certain forms of transition relief.


A. In General

The legal transitions literature asks the question: What norm or norms should govern the relationship between the time at which a change in law is announced and the time at which the change becomes effective? Subsidiary questions are whether and what kind of relief or penalties should apply to individuals who are affected by the change because of pre-change decisions they made in reliance on the old rule. Should individuals adversely affected receive transition relief in the form of grandfathering, phased implementation or some other method, or should they simply absorb the costs that result from unanticipated legal change? Similarly, should the unintended beneficiaries of legal change have some or all of the benefits taxed away?[10] The literature’s paradigmatic example is the repeal of the exclusion from gross income of interest earned on certain municipal bonds. [11] If Congress repealed the exclusion tomorrow with an immediate effective date, certain parties would be adversely affected b y their failure to anticipate the repeal at the time they made decisions that are affected by it. An individual who purchased a thirty-year tax-exempt bond last year would suffer a substantial transition loss, because virtually all of the benefit of the exclusion would be lost. [12] Similarly, states and municipalities would suffer transition losses to the extent they had committed resources to future projects on the assumption that they would be able to finance those projects with future borrowing at a low interest rate. Analogously, holders of taxable debt would enjoy an unanticipated windfall because the value of the debt would rise. Had any of these parties been aware of the impending repeal at the time they were making the relevant in vestment decisions, they likely would have acted differently.

If Congress thought issuers or holders of municipal bonds should be protected from the losses that arise on unanticipated repeal, then Congress could adopt an effective date of the new reule far into the future, make some sort of relief available for pre-repeal investors, or both. For example, the repeal could be implemented in steps ( such as by removing the exclusion for a fraction of the interest initially, with greater fractions removed in subsequent years), or it could be applied only to bonds issued after the enactment date. [13] Conversely, if one thought that investors’ and issuers’ anticipation of new law would produce bett er outcomes overall than does their reliance on the availability of relief, one might favor immediate or even explicitly retroactive application of a repeal with little or no transition relief.

Prior to Michael Graetz’s influential work on tax transitions, what has since become known as the “old view” of legal transitions held sway. [14] Under the old view, the norm that changes to positive law should take effect no earlier than on the date of enactment, or in certain cases the date of public announcement, [15] typically rested either on the conventional idea that fairness demanded that individuals subject to the l aw be able to know what it is before they act, or on the belief that one’s inability to rely on the law at the time one made decisions affected by it created substantial inefficiencies. [16] Further, it was widely accepted that transition relief was appropriate for certain types of legal change, such as the repeal of a tax preference or, more generally, of any legal rule th at on enactment represented an invitation to act in a certain way. By contrast, the norm of retroactivity for judicial decisions was thought to rest on the basis that adjudication clarifies what the law already is; as a consequence, no unfair change to the rules arises when a new interpretation is announced. [17] Moreover, at least the scope of uncertainty in any particular statute that might be resolved by subsequent decisional law was limited, so that the adverse incentive effects from the possibility of retroactively applied judicial “new rules” were minimal.

Graetz’s 1977 article marked a turning point in the scholarly approach. He argued that in the tax context, neither fairness nor efficiency considerations supported the widespread use of delayed effective dates or transition relief when the tax law changed. [18] Broadly stated, Graetz adopted a two-pronged attack to the transition relief norm. First, he noted that almost any legal transition is likely to have ramifications for many parties beyond the typical beneficiaries of relief. [19] For example, the rules permitting accelerated depreciation for physical capital affect not only purchasers of favorably treated capital but also manufacturers of it, who are likely to increase production in response to stimulated demand. A repeal of the accelerated depreciation rules might be grandfathered for pre-repeal purchasers, but no existing norm would suggest that manufacturers who, prior to repeal, had invested in plant and equipment to construct favorably treated capital also would enjoy transition relief.

Second, Graetz argued that even on its own terms a norm of extensive transition relief could not be justified according to any of the standard measures ordinarily used to evaluate tax rules, including efficiency, reliance, horizontal equity, vertical equity, and social contract theories of government. [20] For example, the notion that a norm of transition relief avoided inefficient precautionary behavior by investors who would not take offered tax preferences at face value made unwarranted empirical assumptions about the costs of requiring individual s rather than the government to bear transition losses. Graetz argue d that in light of both the generally accepted view that markets are efficient and the fact that legal changes are probabilistic events that affect the value of investments just as other unknowns do, it was more reasonable to assume that transition costs would be minimized if investors bore transition risk and spread it through markets than if the government bore the risk and sp read it to the populace generally. [21] Similarly, the idea that reliance interests justified or required strong norms of non-retroactivity was largely quest ion-begging. What constitutes reasonable reliance is itself a question that needs to be settled on the basis of other principles, since expectations a re not free-floating but rest on actual practices. [22] Because none of the standard fairness arguments supported a strong non-retroactivity norm, the case seemed weak on reliance grounds as well. [23]

Underlying these two arguments was Graetz’s more basic and far- reaching insight that the distinction between retro activity and prospectivity on which so much of the older literature relied was in many ways arbitrary. Once one recognized that transition gains and losses arise because rules that take effect in the future have an effect on the con sequences of past conduct, the notion that legal change that was expressly directed at past conduct (so- called nominal retroactivity) differed categorically from legal change that was only “forward looking” seemed mistaken. If a forward-looking, ungrandfathered repeal of the municipal bond interest exclusion has nearly as great an effect on a purchaser who bought a 30-y ear municipal bond in the year before repeal as would an expressly retroactive repeal, the only difference between the two types of transition would seem to be in degree. The victim of nominally retroactive repeal suffers a loss of all 30 years of exclusion of interest on the bond, while the victim of the nominally prospective repeal loses just 29 years – a difference of but three percent. Graetz observed that it is hard to see why this difference ought to matter all that much in analyzing the effective dates of new rules. [24] As a consequence, in thinking about possible transition rules, there was no reason in principle why new rules might not take effect immediately, and even in some cases retroactively, even if the rules were “legislative” rather than judicial, or articulations of “policy” rather than mere interpretations. Stated otherwise, if any change other than one that occurs so far in the future that all pre-change reliance interests are protected raises a retroactivity concern, it would appear that we are already committed to retroactivity in legal change even when the new rule is prospective only. Therefore, we should not be especially solicitous o f parties who bear transition costs because the law changes.

Subsequent commentary has generally carried forward Graetz’s observations as a basis on which to make the case against a norm of extensive transition relief. [25] Equally significantly, scholars have appropriated Graetz’s analogy of legal transitions to other market-based changes as a basis upon which to expand the analysis to all forms of legal change. For example, Louis Kaplow describes a legal transition as any unanticipated change in the expected value of legal entitlements that results from government action. [26] Under this definition, legal transitions include not only changes to positive law, but also judicial and administrative decisions, executive actions, and even statements o f policy, as long as, in each case, they were not fully anticipated before t hey became effective. [27] Further, because the definition includes resolution of any uncertainty (or the creation of any uncertainty) with respect to the future value of a present entitlement, a transition issue arises for Kaplow even if no actual change to a rule or policy has taken place. [28] Thus, the filing of a lawsuit on the basis of a novel legal theory would raise a transition is sue, even if the lawsuit ultimately were found non-meritorious (a second transition issue).

Other scholars have adopted similar approaches. Dan Shaviro poses the transition question as arising under essentially the same circumstances as Kaplow does. [29] Kyle Logue characterizes a legal transition as occurring whenever there is an unexpected change to legal entitlements. [30] Thus, as under Kaplow’s view, legal transitions a rise both when the positive law changes and when a new interpretation of existing law comes into effect. Logue also argues that legal transitions arise on scientific discoveries, if the discovery has the effect of altering legal entitlements. As an example, the discovery that asbestos causes cancer was a form of legal transition because it affected the legal entitlements of (among others) asbestos manufacturers, even though it did not involve “a change in the substantive liability rule.” [31]

The motivation for this expansive interpretation of the concept of a legal transition derives from the consequentialist orientation of the new view. When expectations are the concern, the type of change to legal entitlements becomes largely immaterial, because the prospect of any type of change affects expectations. Consequently, the distinctions among types of legal change become matters of degree, not of kind: [32] Changes to positive law are likely to be more sweeping than changes to decisional law, and actors are less likely to be able to anticipate their magnitude and direction. [33] Conversely, positive law changes tend to be forward-looking (in form), whereas judicially created legal transit ions have nominally retroactive effect. [34] These differences will have varying effects on the extent to which actors likely to be affected by the prospect of change will insure against it, but that is all. Apart from these differences, the various modes of legal change seem to differ only in formal or otherwise inessential particulars. As Kaplow puts the point, “[t]o private parties, gains are gains, losses are losses, and taxation and compensation of gains and losses have the same cost or benefit regardless of the original source of the effects being mitigated.” [35]

B. The Case for Immediately Effective, Uncompensated Transitions

On the basis of the preceding account of legal transitions, Kaplow offered a powerful and influential argument for enactment-date effectiveness of most legal change, and against almost all forms of transition relief. The argument rests on two fundamental similarities shared by all legal transitions: that they raise the question of absorbing the costs of uncertainty, and that they appear to be no different in principle from ordinary market risks that investors must bear ex ante . [36] Because it is easy to show under a wide range of assumptions that the most efficient method for absorbing market risk is private ordering, Kaplow argued that from an efficiency perspective these features of legal tran transitions made the case against delayed effect and strong forms of transition relief practically overwhelming. [37] He also argued that even enactment-date effectiveness should give way to nominal retroactivity in certain cases, such as where the prior regime turned out to have been undesirable during the period it was in effect. [38]

The efficiency argument for the superiority of private ordering over government relief in the market risk context is straightforward. Consider Kaplow’s example of the prospective builder or owner of a house on a flood plain. [39] An individual who must bear the costs of a flood in the first instance can be expected to make the efficient decision about whether and how to bear them—by absorbing the loss if the individual builds or owns and the flood occurs, by purchasing insurance (or s elf-insuring), or by building or owning elsewhere. By contrast, builder s and owners who can rely on a general policy of government relief from losses due to floods can be expected to over-invest in houses on flood plain s because they do not need to bear the cost of a flood if it occurs. In effect, government insurance creates an externality by providing a costless put option for those who invest on land in flood plains. [40] Further, privately arranged risk-shifting and risk-spreading are almost certain to be superior to government-provided relief because the latter does not distinguish among the different levels of risk tolerance of different private actors.

Turning to legal transitions, it would appear that transition risk should be borne in the first instance by private ac tors because the possibility of a legal transition, like that of a flood, is a foreseeable probabilistic event that has an effect on the market value of investments. [41] There are two aspects to the efficiency gain of requiring investors to bear transition losses. First, private insurance remains more efficient than government relief, for the same reasons that it is superior in the case of market-based events. Where investors must bear the costs of legal change in the first instance, they will spread or shift risk so that it is optimally allocated between themselves and others. Second, forcing private investors to bear the risk of legal change will lead to beneficial pre-transition anticipation of future law, as long as it is reasonable to suppose that the law is improving. While this assumption has been aptly described as “heroic,” [42] it is at least reasonable as a working assumption or a first approximation. Thus, if we think that strict product liability in tort provides a better regime than negligence for risk-bearing and -spreading, a transition norm that does not permit manufacturers to rely on transition relief will encourage them to act as though strict liability applied avant la lettre . Indeed, the prospect of retroactive application of such a regime is likely to have still greater beneficial effect. Such pre-transition changes in conduct produce welfare gains apart from the savings that private insurance provides over transition relief. [43]

C. Nominal Retroactivity

More recently, Shaviro has offered an in-depth analysis of tax transitions and retroactive relief in his book, When Rules Change . [44] Shaviro’s work is marked by subtlety, analytical rigor, and generality in its relaxation of the assumptions that law is generally improving and that legislators act benevolently. [45] However, although the book deepens and expands upon the analyses that Graetz, Kaplow and other new view theorists have developed, it does not abandon their conceptual underpinnings. [46] Rather Shaviro adopts the welfarist orientation o f the new view [47] as well as the new view’s skepticism regarding the significance of popular distinctions, such as that between nominally retroactive and nominally prospective legal change. [48]

Shaviro’s discussion of the anti-nominal retroactivity (ANR) norm in the tax context offers a particularly clear explication of the grounds for this skepticism. [49] The ANR norm is simply the prevailing idea that t ax legislation ought not apply with expressly retroactive effect. Shaviro’s discussion purports to show that the ANR norm does not rest on any categorical distinction between nominally retroactive and nominally prospective transition rules, but instead “is strongly rooted in popular sentiment, legislative practice, and perhaps even the Constitution as the courts are likely to interpret it.” [50]

Shaviro begins by acknowledging the appropriateness of the ANR norm for what he calls “rules of the road,” or matters of pure convention. [51] These are cases in which the problem that a legal rule solves is primarily one of coordinating conduct among persons, rather than optimally conforming their conduct to an externally-given constraint. It does not much matter whether everyone drives on the left or the right; what matters is that everyone drives on the same side. Similarly, it does not matter whether the default taxable year begins January 1 or July 1; what matters is that the default applies to everyone in a given class.

It might seem that to acknowledge a set of cases in which the ANR norm would be inappropriate concedes the basic point, for if the ANR norm were entirely arbitrary, then it should have no pro per application to any category of legal transitions other than by chance. But Shaviro’s point seems to be simply that rules of the road should no t have retroactive effect at all, be it through nominal retroactivity or through nominal prospectivity that does not provide relief. Accordingly, when a rule of the road is adopted or altered, there is no particular reason either for the new rule to apply with nominally retroactive effect, or for transition relief not to be provided to those who relied on the old rule. [52] An arbitrary rule whose content is that it functions as a coordinating device typically provides no particular clue to individuals subject to it that t hey should anticipate its change. More importantly, no efficiency gain from eliminating retroactivity is likely to arise by dispensing with the ANR norm or transition relief for such cases, because the gains derive from the coordination of parties with each other, not from conforming the conduct of the parties to an independent state of affairs to which the rule responds.

Shaviro contrasts rules of the road with tax rules whose significance consists at least in part in their furtherance of some independent policy. [53] Despite the fact that the ANR norm extends to mos t such rules, the difference in real terms between ex press retroactivity and express prospectivity is, in his view, essentially arbitrary for the now- familiar reasons first identified by Graetz and developed by Kaplow. [54] As Shaviro puts it, neither the apparent effect nor the apparent command of the ANR norm bars retroactive application of new legal rules. The apparent effect of the norm is to cause each period’s events to be taxed under the rules in force at the time rather than at a later time; its apparent demand is that a prior period’s events not be revisited by applying a later period’s rules to them.

Shaviro notes two objections to the norm. The first is that the tax law often takes account of economic events many yea rs after they occur. [55] A simple example is the taxation of accrued capital gain or loss on an asset in the year of disposition of the asset rather than in the years in which, and at the rates at which, the gains or losses economic ally accrued. The second is that “completed” transactions from prior years m ay have been undertaken because of future tax benefits that could be repeal ed on a nominally prospective basis, before the taxpayer has a chance to reap the benefits of the pre-repeal decision. [56] This is the example of a prospective and un- grandfathered repeal of the interest exclusion on m unicipal bonds. Shaviro argues that because the tax rules under such subsequent regimes have an effect on the tax consequences of transactions comp leted in prior periods, “mere accounting considerations govern the definition of nominal retroactivity.” [57] That is, nominal retroactivity differs from nomin al prospectivity only in arbitrary accounting details, because both apply to alter the consequences of past conduct. Ultimately , Shaviro concludes that “[n]ominal retroactivity is inherently a formal, rather than a substantive, economic category.” [58]

Shaviro’s view can perhaps be made clearer upon consideration of the argument that is sometimes made that nominally retroactive statutes differ from prospective ones in that the former apply to completed conduct and in this regard cannot function as guides to con duct, [59] whereas nominally prospective changes can so function. Consider the significance of this distinction with regard to the consequences of a pre-transition act. As previously discussed, if B purchased tax-exempt bonds prior to the repeal of the interest exclusion, the fact that the new rule is prospective only does not mean that it does not apply to consequence s of the pre-transition conduct, just as a nominally retroactive rule would . There appear to be only two differences. First, the nominally retroactive statute applies to conduct, not just to its consequences, and second, its extent will be greater because it will apply to all the consequences of pr e-enactment conduct (back to the effective date of the new rule); the nominally prospective enactment applies only to current and future consequences. The second point is conceded by all, but if the issue is that there is a categorical distinction between nominal and merely “effective” retroactivity, the question of degree does not matter. The first point similarly loses its significance if one assumes that the purpose of the conduct—for example, the purchase of tax-exempt bonds—is to realize the consequences. Where the only concern is with the value yielded by the investment, there appears to be no reason to place independent significance o n preserving the rules applicable to the conduct itself. [60] Likewise, if a nominally retroactive enactment reaches back into the past, it appears that a nominally prospective enactment also reaches back and does so in the same way: the nominally prospective enactment reaches back into t he past as to its consequences, which is what actually matters about that conduct subject to the enactment.


Based on the preceding discussion, one might summarize the main points of the new view as follows. Legal change is just like any other random event that individuals need to take into account in arranging their affairs. As a consequence, all unexpected changes to legal entitlements raise the same basic issue, which is that parties w ho made investments on the assumption that prior entitlement rules would continue to hold suffer gains or losses because the assumption was false. Because this characteristic of legal change is not meaningfully different from market change, and because the optimally efficient rule for managing market risks is private ordering, or self-help, private ordering should apply to managing legal transition risk as well. Further, under the assumption that law is generally positive in direction, a norm of self-help encourages efficient anticipatory modifications to behavior that are absent under a norm of government relief. If it is a bad idea to provide a tax exclusion for interest on municipal bonds, then investors facing the possibility of repeal of the exclusion without transition relief will have a greater incentive not to invest in the bonds solely for tax reasons than will investors expecting transition relief on any subsequent repeal. The incentive will be greater still if the prospect of a nominally retroactive repeal is in pl ay. Conversely, while transition relief may confer benefits such as providing welfare-enhancing redistribution, these benefits can be more efficiently provided through other mechanisms, such as transfer payments or progressive taxation. Finally, although problems of public choice raise additional concerns, it is not at all clear that they point in favor of a norm of transit ion relief; in fact they may suggest that absence of relief is closer to the optimal norm. [61]

This part focuses more closely on the distinctions that the recent transitions literature has drawn in reaching these conclusions. The objects are, first, to formulate more precisely the conceptual framework that is implicit in the new view and, second, to draw out some of the implications of the view that may not have been recognized by its proponents.

A. Entitlement Rules and Transition Rules

The new view understands legal transitions as occur ring whenever there is a change to legal entitlements. Implicit in the account is a distinction between legal entitlements and rules th at specify when distinct entitlements hold. Thus, the rule that interest ea rned on municipal bonds is excluded from gross income specifies an entitlement , as does its opposite. Similarly, a rule that, for example, loans for medi cal school will be forgiven for doctors working in low-pay practices constitute s an entitlement rule, as does the rule that such loans must be repaid accord ing to their nominal terms. If the bond interest exclusion or the favor able rule on medical school loans were repealed, a rule likely would app ly that specifies when the respective entitlement rules hold. That specification could be express, or it could be implicit through the operation of so me default rule, such as one stating that changes to positive law are understood to take effect on the date of enactment in the absence of explicit legislation to the contrary. Whether express or tacit, such a rule is a transition rule. [62] Stated more generally, transition rules are meta-rules that specify the temporal relationship between or among distinct entitlement rules. [63]

Under this model, one could reformulate a repeal of the bond interest exclusion taking effect one year after enactment as follows:

Entitlement Rule 1: Interest on municipal bonds is excluded from gross income.

Entitlement Rule 2: Interest on municipal bonds is not excluded from gross income.

Transition Rule: Entitlement Rule 1 shall be effective until one year from the date hereof, and Entitlement Rule 2 s hall be effective for all dates on or after one year from the date hereof.

Framed in terms of the distinction between entitlement rules and transition rules, the transition question is a third-order meta-question about a second-order meta-rule: What (third-order) policy or norm should govern the (second-order) transition rules that determine when certain (first-order) entitlement rules apply? [64] A number of answers, of course, are possible. One is that the date the new rule is announced (the “announcement date”) and the date it becomes effective (the “effective d ate”) should be the same; another is that the announcement date should precede the effective date by some interval (delayed effective date); a third would be the opposite: that the effective date should precede the announcement date by some amount of time (nominal retroactivity). Other possibilities include having distinct transition rules for distinct groups (i.e., what is commonly termed “transition relief”): delayed effective dates to t he extent that the investor holds investments that were subject to the old rule on the announcement date (grandfathering), and, perhaps, announcement d ate effectiveness for all others.

B. Implications of the Entitlement Rule Transition Rule Dichotomy

Several observations about the method of conceptualizing legal transitions as changes to entitlement rules are in order. For the most part, these observations do not call into question the new view approach to legal transitions, but they do illustrate the ways in which the new view literature has elided important questions that it has raised. They also make possible a more rigorous analysis of the approach the new view has developed.

1. Transitions vs. Transition Rules

The first observation is that the existence of a legal transition does not imply the existence of a transition rule. This conclusion follows from the definition of a transition rule as the rule specifying the time at which two distinct entitlement rules apply to a type of arrangement. [65] It is logically possible for there to be no rule, express or tacit, that specifies when two inconsistent entitlement rules apply. In such a case, the new entitlement simply supplants the old one and applies at all times on the separate basis that the actions of currently empowered legal actors supersede the actions of legal actors no longer in power. [66] While the absence of a transition rule does not ordinarily obtain in practice, it is important to keep in mind the analytical distinction so that it is possible to formulate the transitions problem accurately, as conceived by the new view—after all, it might be desirable in some circumstances to do away with transition rules entirely. (Recall that it was the dissatisfaction with the conventional understanding of legal transitions as changes to positive law that gave rise to the need for a definition that is neutral with regard to the time and manner of any changes to legal entitlement s.) One example that illustrates the idea of a transition without a transition rule is the analogous case of a change to scientific understanding—the shift, perhaps, from the Ptolemaic to the Copernican view of the solar system. When the Copernican view supplanted the Ptolemaic, there was of course no notion that the Ptolemaic had been correct at some time in the past; rather the thought was that the Copernican view had always bee n correct; it merely had not been recognized as such. The notion that a transition rule would apply in such a circumstance is meaningless.

A second, more familiar, example is that of ordinary judicial interpretation of positive law. If a court announces a new interpretation of a statute, the new interpretation may purport to explicate the true meaning of the statute—at least under a conventional view of adjudication, sometimes referred to as the declaratory theory. [67] Because the court’s interpretation states the meaning at the time of the conduct to which it applies, the interpretation typically applies with at least some retroactive force. Of course it does not literally apply with full retroactivity, but the reasons typically are grounded in separate policy considerations such as finality, not in the notion that the court’s interpretation says what the law means now but not at other times.[68]

2. Relevant Anticipations

The second observation about the new view’s way of formulating the concept of a legal transition is that if no transition rule were the norm, the relevant anticipation (setting aside the fact t hat investors are mortal) would not be of the next legal change, but of the ultimate change to the “best” legal regime, or at least to the final one. The reason is that in the absence of a transition rule, the full retroactivity of every legal rule would nullify the prophylactic measures that investors would take with respect to all intermediate legal changes. This follows, again, from the distinction between entitlement rules and transition rules that is presupposed in the new legal transitions literature. If only entitlement rules change when legal change occurs, then no temporal limitation on the new entitlement rule applies; the rule has full retroactive effect and t he detriments and/or windfalls from the new regime apply to all time periods, past and future. For example, suppose it turns out that the “best” t ax regime is some form of consumption tax, and it is assumed that legal change is positive in direction. If no transition rules operated in legal change, then investors who are far- sighted enough to recognize the ultimate direction of the law should not react to an intermediate repeal, under the current income tax, of the preference on municipal bonds by investing in other assets until the market price of munibonds drops to reflect the anticipate d repeal, even though investors will be subject to tax on all interest they have ever received on such bonds when that repeal occurs. Although the p reference repeal would cause investors to pay tax on the preference and in deed with full retroactivity, that change would itself be nullified when the final shift to a consumption tax occurs, again with full retroactivity.

3. Implied vs. Express Transition Rules

Third, a transition rule need not be express to be in effect. Returning to the municipal bond interest example, repeal of the exclusion for all interest accrued from the date of enactment forward would constitute adoption of a new entitlement rule and a transition rule, whether the effective date were spelled out in the statute or s imply assumed under a default rule. The repeal would provide both a new rule about bond interest (that it is included in gross income) and a rule ab out the extent to which the old rule continued to apply to bond interest (that it remained valid for pre- enactment dates). [69] Thus, the legislative norm that in the absence of an express transition rule or effective date, new legislation takes effect on the date of enactment constitutes a transition rule, assuming that a legal transition is defined as the literature has defined it. [70] Similarly, the Constitutional ban on ex post facto criminal legislation [71] creates a mandatory transition rule for federal criminal statutes, in the sense of a limitation on retroactive application. Conversely, the default norm for judicial decisions is that no transition rule operates (apart from a separately provided statute of limitations and a separate rule for already decided cases), so that the absence of an express transition rule gives rise to full retroactivity for all cases not already settled or beyond the applicable statutes of limitations.

4. Scientific Developments Are Not Legal Transition s.

Finally, note that even the expansive definition of a legal transition offered here does not appear to embrace factual or scientific developments that have a bearing on legal entitlements, contrary to the claims of some new view theorists such as Logue. [72] To illustrate the idea that the category of legal transitions includes such developments, Logue uses the example of newly discovered scientific evidence demonstrating that a product causes a certain type of harm. 73 In the example, he assumes that an underlying strict liability regime is in place and does not change. [74] He analyzes a court’s decision to accept the new causation theory for the first time as a legal transition because the decision changes legal entitlements and the change was the result of a governmental action to accept t he new theory. [75]

The difficulty with this analysis is that the change to legal entitlements resulting from a scientific development appears to be no different from the change that such a development b rings to the market value of an investment; only the governmental actor responsible for enforcing it differs. Suppose A owns a copper mine and it is discovered that copper is particularly useful in a new compute r chip manufacturing process. The value of A ’s entitlement increases dramatically because the price of copper has risen. But this value will be accorded to A only on the assumption that the legal system will continue to  protect A ’s property rights in the mine. The difference between this case and the products liability case is only the operative legal regime—property rights in one, and tort in the other. In each case, the legal regime continue s to do what it always did, which is to assign legal entitlements and obligations according to an unchanged set of rules. Unless one is prepared to say that the discovery of the new use for copper creates a new entitlement rule, it appears one cannot say that the new theory of causation in tort does either.


With a more precise definition in hand of the concept of a legal transition as it is understood by scholars under the new view, it becomes possible to evaluate the view more directly. In this part I examine two of its aspects: first, the claim that legal transitions a re closely akin to factual or market changes, and second, the tacit assumption that legal transition norms can be evaluated apart from certain questions of political legitimacy and authority. The arguments made in this part give rise to a third set of questions that I take up in Part V: whether the new view in fact has much to say about legal transitions that has not already been said before. The sequence of the discussion is intended to be progressive. The second argument developed in this part builds on the first , while the analysis of Part V follows naturally from the considerations discussed in this part. Once it becomes clear that legal transitions are not closely akin to factual changes, it becomes incumbent on new view theorists to offer a defense of the new view that does not depend on the analogy. When that defense fails, the doubt of the extent to which the new literature has made significant contributions to our understanding of legal transit ions becomes salient.

A. Analogy of Legal Transitions to Factual Developments

I have previously noted that scholars in the legal transitions area have relied on the analogy between legal and factual change to support a norm of enactment date effectiveness. Kaplow offer s perhaps the clearest statement of the analogy:

[W]here the focus is on private actors and government substantive policy is taken to be optimal, government- created risk and risk due to market or natural forces, as well as government mitigation of such risk, are almost precisely analogous. To private parties, gains are gains, losses are losses, and taxation and compensation of gains and losses have the same cost or benefit regardless of the original source of the effects being mitigated. Likewise, market relief for risk—whether through explicit insurance arrangements or implicitly, such as through diversified financial ownership—is in principle equally effective regardless the source of the risk. [76]

As Kaplow’s statement makes clear, the analogy is meant to function as more than a casual observation about the parallels between legal and market transition risk. Rather, it amounts to a restatement of his “demonstration that uncertainty concerning government policy is analytically equivalent to general market uncertainty.” [77]

The distinction between entitlement and liability rules developed in Part III provides useful purchase on the question o f whether the analogy is accurate or persuasive. As explained there, entitlement rules specify underlying legal rights and obligations, and transition rules specify the temporal conditions under which the former apply. This highly generic definition of legal transitions follows from the requirement that all forms of legal change or development be grouped together without regard to the nature or authority of the decision-maker initiating the transition or to what gives rise to the decision-maker’s decision. Although a conventional understanding of legal change would not need to insist on abstracting entitlements from meta-rules regarding their temporal application, that course is unavailable under the new view because of the expansive concept of transitions it adopts as a result of the focus on anticipations.

Once the distinction between entitlement rules and transition rules is accepted, however, it becomes apparent that a legal change has a different kind of effect from a factual or market-based development, because factual and market changes do not reach back into the past (or, for that matter, stretch infinitely into the future) in the way that a mere change in entitlement rules does. Unlike an uncompensated legal change regarding the availability of government relief for floods, t he fact of the flood itself does not mean that flooded land was always flooded. Nor, for that matter, does it mean that the land became flooded sometime “before” the flood (in analogy to a partially retroactive legal change) or sometime after it (in analogy to a delayed effective date for a new legal rule). The “announcement” date of the change to entitlements, in the case of a factual change, is typically the same as the change itself. When it is not (as when there is warning that a change will come), it still occurs on some particular date, prior to which it had not occurred.

From the proposition that a legal transition implies only a change to entitlement rules, it follows that the most one can say about the similarity of legal change to factual or market change is that a legal transition plus a transition rule that takes effect on the announcement date is “almost precisely analogous” to market change. [78] Only that combination of change to entitlement rule and transition rule mimics a factual change, because only that combination preserves the essential feature of factual change that need not be present in legal change: the fact that the change is its own announcement. The problem, of course, is that this formulation demonstrates that the analogy is entirely question- begging to the extent it purports to support a presumption of enactment date effect without transition relief, since the issues in the transitions literature are precisely whether the new rule should take effect on the announcement date or on some other date and whether transition relief should be available. In other words, the analogical argument assumes its conclusion. If one presupposes that a legal change is just like a factual change, then a legal change by definition occurs on its date of announcement, and it should not be surprising that the most efficient way to deal with it is the same as the most efficient way to deal with natural or market-based changes. But the question, of course, is whether the announcement date should be the same as the effective date.

Indeed, it is precisely the fact that the announcement date of a new legal rule can differ from its effective date that there is even a transitions question in the first place. Consequently, to say that a transition norm of having identical effective and announcement dates i s like a factual change is merely to state the consequence of one possible transition norm: announcement date effectiveness. It does nothing t o justify the transition norm itself. One can state the point in the converse: if it were possible to have unexpected factual developments take place on a date different from their “occurrence,” would one assume that the optimal date would be the occurrence date?

One might accept these observations but counter that they reflect a merely semantic point: if one defines transition rules as meta-rules that determine the temporal extension of entitlement rules, then of course any legal transition that is not fully retroactive will be accompanied by a transition rule and in that sense by transition “re lief.” Is it not possible to resolve the problem by saying that a legal transition consists of a change to an entitlement rule plus a transition rule, which m ay or may not be explicit? The argument would then be that legal changes, so defined, are just like factual developments. They happen unannounced, taking effect from some date forward, and we expect investors to deal with them through market mechanisms.

The problem with this claim is that it is false. Legal changes differ from factual developments only in that they can apply retroactively or prospectively rather than when they are announced; as stated above, factual developments are neither pre- nor post-announced. To say that investors should bear the costs of legal transitions on analogy with natural or market- based uncertainty disregards both the flexibility inherent in legal change, which is that there is in principle much control ov er the effective-date timing of the new entitlement rule, and the fact that it is just this flexibility that gives rise to the “transition question” in the first place. If the flexibility were lacking, there would be no transition debate, because all changes to legal entitlements would (necessarily) occur on the date of announcement, or perhaps on some other date, but in any case on a date over which one had no control. [79] In short, the validity of the analogy presupposes the decision that the best transition rule is the default rule f or changes to positive law, yet the analogy is used to justify that decision.

B. Problems With a Single Transition “Norm”

Ultimately, the difficulties with the analogy of legal change to factual or market-based change point to the crux of the problem with the anticipations-based literature. As argued below, t he failure to appreciate fully either the flexibility or the limitations inherent in the power to adopt a transition rule means that the new view literature fails to account for what is distinctive about legal transitions. On the limitations side, it fails to recognize that the markets that provide efficient methods for managing risks and internalizing costs, and that serve as models for a legal transition norm of announcement-date effectiveness, themselves depend upon the existence of a stable legal regime that presupposes limits on possible transition norms. Some of the norms advocated in the recent transitions literature do not take cognizance of these limits. Put differently, certain transition norms, including a universal norm of enactment-date effect with no transition relief, would undermine regime stability and, hence, would vitiate or destroy the markets that are supposed to function as models for managing risk in the first place, including “transition risk.”

On the flexibility side, the new view fails to appreciate the distinctive benefits that different transition “nor ms” can offer depending upon the different type of, and justification for, the adoption of a new entitlement rule. For example, there may be distinctive and systematic reasons favoring greater retroactivity for judicial action than for legislative action, and these reasons may relate to the type of legal change that the respective legal actors initiate, rather than, for example, to the capacity of the respective institutions to adopt “good” or “bad ” law or to be influenced by rent-seeking or other undesirable behavior. [80] If one insists on treating all forms of government action as essentially the same, these distinctive reasons become invisible.

In the following two subparts, I illustrate the limitations point by considering the consequences of extending the default rules that apply to judicial decisions to positive law, and the flexibility point by considering the reverse extension. These potential extensions (or some other uniform norm) should seem natural under a conception of leg al change that insists on effacing the distinctions among judicial, executive and legislative decisions, because, under that conception, whatever the correct norm or norms should be, they should not depend upon which government actor initiates the change, but upon which rule provides the proper incentives to legal actors. [81]

1. Retroactive Relief—Naturalization of the Market

Consider first the consequences of extending the judicial rule of nominal retroactivity to positive law. New view scholars have noted that such a norm would have a chilling effect on investors, for the obvious reason that investors would lack confidence that the rules on which they relied in arranging their affairs would in fact be the rules applicable to their affairs. [82] Unless investors could be sufficiently certain either that current law would not be changed unfavorably, or that if it did change, the retroactive effect would not extend far enough back to reach their earlier conduct, they would be unlikely to commit their resources in ways that would otherwise be efficient. For example, taxpayers under a steady-state income tax can be expected to allocate their resources in ways that favor current consumption to a greater extent than do tax payers under a steady- state consumption tax. [83] Unless taxpayers believe they can make an accurate prediction of the likelihood that a future transition to a consumption tax will occur and will apply retroactively to pre-change decisions, they are apt to adopt a wait-and-see attitude and to refrain from committing resources either way.

This analysis, while technically correct, is importantly incomplete. First, it is hard to square with new view theorists ’ general skepticism about the significance of the distinction between nominal ly retroactive legal change and nominally prospective change that has retroactive effects. Moreover, it has not been universally accepted. Fo r example, Kaplow argues in favor of a norm of nominal retroactivity when prior law was bad law for the period it was in effect. [84] More to the point, however, a nominal retroactivity norm would do more than make investor s hesitant to make new investments for fear of expropriation of gains on repeal; it would create enough uncertainty in the stability of established law to make reliance on markets as a means of managing risk unreasonable altogether. Indeed, it would make investors question the validity of the legal regime itself, and, moreover, in ways that a norm of enactment-date effectiveness would not, even if the impact on investments in the latter case were systematically comparable to those in the former.

The difficulty stems from the fact that a norm of retroactivity of positive law is inconsistent with the idea that law -making institutions derive their authority from consent. [85] Consider that a genuinely retroactive pronouncement, such as a new rule for taxation of a prior year’s income, does not merely call into question the firmness of any rule that one might follow presently. It also implies that the current ly operative legislature has departed from the rules that determine who is the legislator; in particular, it has usurped the role of the body that was authorize d to act as legislator during the period to which the nominally retroactive enactment applies. In effect, individuals under the prior regime lived under no law. While that period has of course passed, individuals in the cur rent regime, anticipating the possibility of future nominal retroactivity, find themselves under no law at all. Further, they also find that they were actually not under law to the extent they also were under prior law.

The point may be illustrated through the following example. Suppose a neighbor appears at your doorstep with th e pronouncement that he has just proclaimed himself the law on the basis of his decision to amend the Constitution, and that among other new rules ta king immediate effect is one that transfers all of your property to him. Yo u might counter that his proclamation has no legal effect because he hasn’t followed the rules set forth in the Constitution that provide for its amendment to make him the new legislator, but he replies that he has amended those as well.

Though fanciful, the example highlights what is problematic about the case of nominally retroactive legislation. It, like the example, has a boot-strapping quality. A prior legislature, elect ed to govern during a prior term, followed procedures set forth in the relevant legal document to establish the law that would apply during that period. Later representatives (chosen by others) have now determined that that la w did not apply during that period. If that is true, then the substantive law operative during the putative tenure of the prior legislature was not established by its electorate; it was established by a subsequent electorate. The difficulty is that if law derives from consent, then law that is not established with the consent of the electorate is not really law for the electorate , any more than the law of Latvia is the law of England. The enactment of such retroactively effective legislation can only have been through a valid exercise of authority (rather than through force) if either the subsequent legislature validly amended the rules that determine who was authorized to promulgate substantive law during the prior period, or the legislature in office during that period authorized the subsequent legislature to override i ts enactments. In other words, nominal retroactivity cannot go “all the way down” but must instead be grounded in some procedure that is not itself op en to retroactive revision.

By hypothesis, the subsequent legislature did not follow the procedures set forth in the relevant document that would permit retroactivity of the change to a legal entitlement. Therefore, in order for the retroactive legislation to have been valid, the principle that later legislatures may retroactively revoke prior legislation must have been implicit in the earlier enactment. If so, the case would be distinguished from that of the neighbor, for whom no consent to adopt new rules is implied. The problem with this argument is that once the idea is accepted that a subsequent legislature can read tacit terms into prior legislation, there is no way to limit this principle to the substantive statutory law; it could apply equally to any subsequent statement that purports to set forth who is entitled to say what the law is, including the neighbor’s claim about th e earlier constitution. [86]

The basic point may be stated as follows. From a conceptual standpoint, at the top of the chain of authority must lie express, not merely tacit, agreement by everyone who would be bound. [87] That agreement cannot be amended in ways inconsistent with its express terms other than by unanimous consent; otherwise, any subsequent change lacks authority for those who do not consent. This principle does not mean that even the highest- level document cannot have procedures for amendment that call for less than unanimity, but it does mean that a depart ure from the rules for amending the highest-level document that are set fo rth in that document would be invalid for anyone who opposed it.[88]

A proponent of the new view might respond to this a rgument by raising the claim discussed previously: the signif icance to private actors of their conduct is the results that it produces and n ot the means by which they are produced. [89] From this perspective, whether a new rule applies with nominal retroactivity or merely affects the consequ ences of completed pre- transition conduct going forward does not particula rly matter. Since the retroactive effect of nominally prospective changes is not inherently problematic, it is not clear why we care about nomi nal retroactivity. In other words, it does not matter whether the new rul e affects new conduct or the consequences of conduct that were its raison d’ être. This argument, however, simply assumes the conclusion that the dis tinction between pre- announcement-date conduct and its future consequenc es is not significant, because it assumes that legal rules do not depend u pon consent to function. If a transition norm vitiates that consent, there i s literally no legal order; it becomes impossible to act in any way other than as one in something like a state of nature.

Another possible objection is that only a transitio n norm of nominal retroactivity to positive law would call government legitimacy into question, and that kind of nominal retroactivity is a relative rarity. But, again, the new view is predicated on the idea that the difference between nominally retroactive and nominally prospective leg al change is largely illusory. [90] The distinction is said to be material only in th e sense that nominally retroactive changes are likely to have a greater effect on the magnitude of the legal change than are nominally pr ospective changes. [91] Either of two possible conclusions follows. One is that there is, in fact, a significant distinction between nominal prospectivi ty and nominal retroactivity that is not captured in the differenc e between the size of the effect of either on the market value of investments . The other would be that even a norm of nominal prospectivity raises the sta bility question, since certain nominally prospective changes can greatly a lter the value of pre- change investments. The former would demand a reco nsideration of the decision to differentiate nominal prospectivity fro m nominal retroactivity on purely quantitative grounds, while the latter se ems to be false: In many cases, nominal prospectivity does not raise the sam e difficulties that nominal retroactivity does, as evidenced by the fac t that the strength of the ANR norm is not matched by any comparable norm for purely prospective legal change.[92]

These points may be summarized as follows. From th e perspective of the private investor, the significance of the AN R norm lies not simply in the fact that it prevents application of new rules to investments already made. It is that it gives expression in a particul arly vivid manner to the idea that, as a general matter, law under organized gove rnment derives from consent. [93] If law were the means by which an independently a uthorized or naturally existing sovereign implemented its comman ds (such as a philosopher king, for example), then neither instan taneously promulgated nor retroactively extended legislation would raise a transition issue, apart from the question of incentive effects amply discus sed in the literature. [94]

But where law is a means of government by consent, the strict separation between the law-giver and the persons subject to la w does not hold, and the idea that laws need not be guides to conduct, which , in a certain sense, investors apply to themselves, gives way. If law i s law because it is consented to, then a transition norm that vitiates consent vitiates law.

2. Announcement Date Relief—New Interpretations As New Law

If the hypothetical extension of the rule for judic ial interpretations to positive enactments illustrates the failure of t he new view to appreciate the limitations that apply to the choice of transit ion norms, the inflexibility of the new view emerges upon consideration of the c onverse case: extension of the default norm of nominal prospectiv ity to judicial decisions that announce new interpretations of existing rules . [95] The difficulty here is that it becomes impossible under the new view to ma ke out a reason for differing transition norms corresponding to differi ng methods of propounding new rules, because, as previously discu ssed, the method by which the government adopts a new rule is considere d to be a formal or arbitrary feature of the rule.

In conventional or “old view” terms, the basis for retroactivity of judicial relief is, of course, that the pronounceme nt of a new interpretation merely states what the law already was, or perhaps more realistically states the “best” or “appropriate” interpretation among a small number of candidates. [96] To be sure, in some circumstances that notion is a fiction, [97] but the norm, and certainly the justification commo nly offered for the norm, conforms to the idea. [98] Under the new view, the rule for judicial “new rules” therefore presents a puzzle: why have retro activity for these rules if a norm of nominal prospectivity should operate for positive law? If the same general considerations apply regardless of the form in which a legal transition occurs, it becomes difficult to explain different defaults for different types of government action. While some n ew view scholars have noted that different norms for different types of c ases may be appropriate, they have generally defended the idea with referenc e to the substance of the underlying rule, not the form in which the new rule is promulgated. For example, as previously mentioned, Kaplow supports a norm of nominal prospectivity without transition relief for transit ions that respond to changed circumstances, but he argues for a norm of retroactive application of new rules “when the justification for a reform s uggests that the prior activity was undesirable.” [99] Needless to say, both a court and a legislature (as well as an administrative agency or any other p olicy-making entity) can be confronted with either type of case.

The case for retroactivity of judicial interpretati on, however, has nothing to do with whether the prior activity was u ndesirable. It has to do with the fact that judicial action purports to be a statement of what the law is, even though on occasion the statement may refle ct a court’s view of what it ought to be. [100] At all events, the uncertainty that adjudication resolves is very different from legislative uncerta inty. Uncertainty about the content of law is a feature of any legal system for the simple reason that no law can contain a full specification of all case s to which it applies. [101] That is why, as contrasted with legislation, adjudi cation typically resolves a legal question in favor of one among a very small n umber of possible interpretations. Each of these interpretations is typically known to the litigants not only before trial but, in the ordinar y case, at the time at which they engage in the relevant conduct. Hence it is i n this setting, much more than in the legislative setting, that the case for retroactivity is strongest. Where the legal risk is quantifiable beforehand, it is much easier to insure against it. The fact of limited possible interpret ive outcomes also places much less strain on concerns about rule of law than would the possibility of open-ended retroactive application of new law, whic h can in principle have any content. Finally, unlike positive law, in whic h it may make sense for the party controlling its content to take account o f the consequences of changes to the content, new judicial interpretation s are an inherent part of the content, in the sense that any statement of a r ule has possible ambiguity. In short, because the new transitions literature di sregards the different nature of the uncertainty that legislation and adju dication resolve, the distinctive justification for retroactive applicati on of judicial interpretation becomes unintelligible for the new view. That just ification is ultimately the same one that applies to rules more generally, whic h is that parties nominally subject to them should, in fact, be subje ct to them, even if there is by necessity some uncertainty in their content i n borderline cases. [102] In this regard, the “new” legal rule that a tribunal m ay announce does not raise the same concerns that nominally retroactive legisl ation does, because it is grounded in the nature of law generally. Indeed, b oth the ANR norm and the norm of nominal retroactivity for judicial deci sions seem to derive from the nature of law, or at least the nature of law as a means of rule by consent. Where nominal retroactivity as a norm for legislati on is inconsistent with rule through law by consent, nominal retroactivity as a norm for adjudication arises from the basic incompleteness o f any law.

C. The Conventional ‘ature of Legal Regimes

Both of the problems of the new view literature des cribed in the preceding subparts point to a fundamental failure o f proponents of the new view to appreciate the conventional nature of law a nd therefore the institutional context in which the transitions ques tion arises at all. On one hand, as artificial institutions, legal regimes are instantiations of convention or agreement, and agreement is not subject to natur al constraints. The absence of natural constraint means that legal tran sitions can occur at times different from their announcement dates and indeed at times that depart from the announcement dates in ways that vary depen ding upon the types of conventional settings in which the transition arise s. While scholars writing from a new-view perspective have of course entertai ned the possibility of a norm for legal transitions that incorporates some a mount of relief and some amount of retroactivity, the literature has general ly settled on a norm of enactment-date effectiveness because it cannot iden tify the institutional reasons that would support a different date—namely, either prospective effectiveness (or some transition relief variant th ereof) for reliance-based reasons when they might apply, or retroactive effec tiveness for nominal transitions that in fact further the principle or p urpose of an already-enacted rule. In short, the literature is unable to compre hend the variation that arises from different institutional settings, rathe r than from either different substantive settings, such as whether the underlyin g rule was good or bad, or from the varying propensities of different actor s to be subject to non- socially optimizing conduct because of capture, pri ncipal-agent problems, and the like. [103]

On the other hand, as artificial institutions of a certain sort, legal regimes cannot be instantiations of just any conven tion. They must be instantiations of conventions that are consistent w ith the terms of the implicit bargain upon which the regime rests. By n aturalizing the circumstances within which legal transitions occur— for example, by assuming the existence of well functioning markets no matter what the legal regime—the transitions literature fails to apprecia te the relationship between transition norms and the set of legal insti tutions and relationships that depend, in part, on those norms and that provi de the framework within which conventional arrangements, including markets generally and insurance markets in particular, appear natural and are able to function. As a result, the literature fails to appreciate an imp ortant set of considerations at stake in adopting one set of norms rather than a nother in different transition settings; it is able to comprehend the d ifferences among possible norms only in terms of the impact on private expect ations under the assumption that private expectations will not diffe r, for systematic reasons, depending upon which governmental actor initiates t he legal transition It is not able to comprehend the varying impact of possib le norms on the persistence of the conventions that must persist fo r the legal regime to exist at all.

These problems may be summarized with the observati ons that the transitions question arises from the conventional n ature of law and that any attempt to answer that question without due regard for the distinctive nature of convention is apt to fail. One could justifiabl y say, perhaps with some irony, that the “nature” to which the new-view anal ysis appeals in analyzing the legal transitions question (namely, markets) wo uld not exist if a set of conventions that themselves presuppose certain cons traints on transition norms were not also in place. These conventions ar e both more fragile and more flexible than that nature: more fragile in th at they place limits on transition “norms” that might otherwise appear more efficient, and more flexible in that, unlike the natural events to whic h they are inappropriately analogized, they may occur at times different from their “announcement.” The conventional source of the transition problem i s also the reason why the appeal to the rules for market-based change as a basis for legal transition policy is fundamentally unavailing. If what must be taken as given is the possibility that the two dates differ (and it must if the transition issue is to arise at all), then the appeal to the s olution of having the dates be the same where that solution presupposes that they are the same offers no help whatever in analyzing the transition problem.


None of the observations made in the preceding part imply that the actual policy recommendations of the recent transit ions literature are incorrect. Nor, conversely, does the fact that the law occasionally has been applied in ways inconsistent with principles that s upport political stability refute the observations made there, because the que stion is what an appropriate transition norm should be, not what con sequences follow from the fact that the government occasionally has devia ted from one norm or another. What the observations do imply, however, is that any particular transition norm must be evaluated with a view to th e impact it will have both on “investor” incentives and on the conditions that permit investors to react to incentives effectively under law. When th e new-view literature is evaluated according to these criteria, it turns out that its recommendations tend to be either uncontroversial—that is, not part icularly different from those of the “old view”—or questionable. In the en d, the recent transitions literature has little to say that is both new and t rue. [104]

In this part, I develop these claims through an exa mination of the case for transition relief under nominally prospect ive legal change. Prospective change is the focus here because I have already argued that the transition literature’s arguments for nominally ret roactive rules, or more accurately its arguments for what is actually wrong with nominal retroactivity of positive law, are unpersuasive. [105] The first subpart reviews the charge that reliance-based arguments for transi tion relief are circular. Commentators have tended to make the conventional a rgument for reliance against which this charge has been raised in two di stinct settings, a fact typically unnoticed in the recent transitions liter ature: to oppose the nominal retroactivity of positive law, and to advoc ate transition relief for a limited class of nominally prospective changes to t he law. [106] I argue that although different justifications for the reliance argument apply in the two settings, the blanket criticism offered under the n ew view has tended to collapse the cases because new view scholars reject the idea that there is a meaningful distinction between nominal retroactivit y and nominal prospectivity that has retroactive effect. [107] The discussion in the previous part has indicated why the distinction is meaningfu l and why reliance on the notion that legal transitions will not generally be nominally retroactive is reasonable.

The second subpart demonstrates that the case for r eliance as a basis for certain forms of relief from nominally pr ospective legal change is perfectly sensible as well. As a general matter, m any types of legal rules can be analogized to conditional quasi-promises of one sort or another. Where the original rule has this character, the que stion is whether the conditions that would justify an uncompensated depa rture from the continued operation of the rule obtain. If they do not, then relief—what is called “transition relief”—is generally appropriate . If they do, all bets are off and no transition relief would seem to be requi red—at least by the terms of the original promise, though other reasons for r elief might exist. Underlying all of this is the view that reliance as a norm that ought to play some role in the analysis is not entirely arbitrary . The fact that reliance cannot justify itself does not mean that it is arbi trary. Rather, as argued below, reliance derives its role from the nature of language. If words are to mean something, then they must be able to be relied upon under certain circumstances. The question is what those circumst ances are.

A. The New View Critique of Reliance

As previously discussed, under the new view, “trans ition relief” should generally be denied because it creates an un desirable government externality. [108] The asserted parallel to market-based uncertainty is said to justify this principle, while reliance and reasonab le-expectations arguments are said not to weigh in the other direction, becau se reliance and reasonable expectations are second-order principles that depen d upon underlying practices. [109] If absence of relief were the expectation, then r easonable reliance would not supply a basis for relief. I ha ve argued that the positive part of the argument—the analogy of legal change to market-based changes—fails because various aspects of the legal regime on which markets depend cannot be assumed to persist under c ertain transition norms, such as a norm of extensive retroactivity to positive law. [110] More generally, I have suggested that the positive case for a norm of no relief, to the extent it rests on the analogy, is conceptually flawed because it presupposes the transition rule it purports to justify. [111]

In this subpart, I focus on the charge of circulari ty as applied to the reliance norm. As contrasted with the positive cas e that, for example, Kaplow and Graetz make for no transition relief, [112] this negative part of the argument is perfectly valid as far as it goes. One cannot justify transition relief on the basis of reasonable reliance or ratio nal expectations without justifying the practice upon which the expectations are based, for if the practice were different, then rational expectations and reasonable reliance would differ too. [113] Therefore one would need to justify the underlyin g practice in order to defend the reliance principle.

The important point about this criticism is that, i n principle, it is not inconsistent with the old view. Stated somewhat di fferently, a proponent of the old view is not committed to extensive relief f or nominally prospective changes to positive law as a general matter, or eve n to the idea that transition relief should operate as a default for s uch transitions. These conclusions should not be surprising, since the val idity of the circularity charge does not depend upon anything about either t he consequentialist view of legal transitions or the anticipations-orie nted approach. For the same reason, the suggestion that the old view uncri tically favors transition relief on the basis of reliance represents somethin g of an overstatement. [114] Old-view scholars have not typically offered free-f loating reliance arguments in favor of transition relief for nominal ly prospective transitions. Although not always clearly articulated, the argume nts for such relief (as contrasted with relief from nominally retroactive e nactments) generally have applied to a relatively narrow range of enactm ents.

Graetz’s discussion of reliance arguments in the pr ior literature is illustrative of the tendency to group together reli ance arguments made against nominal retroactivity with reliance argumen ts made against the failure to provide transition relief for nominally prospective transitions. Graetz quotes the following statement by Edwin S. C ohen, Assistant Secretary of the Treasury for Tax Policy during the first Nixon Administration, in support of his claim about the p redominance of the old- view way of thinking:

[P]rovisions have been deliberately kept in the tax law over many years, and they constitute standing invitation s for taxpayers to erect new buildings, drill for oil, or embark on programs of charitable contributions. Even if we s hould conclude that it would be unwise to continue some o f these benefits or if we should alter some of them, it wou ld not be appropriate to remove the preference precipitously after taxpayers have embarked on programs which they migh t not have adopted except for these provisions.[115]

Graetz goes on to observe that “[a]n argument in th is form, without more, ends analysis.” [116] But the observation, while true, is of limited relevance. Rarely have arguments for reasonable re liance been made simply on the basis that reliance alone justifies r elief.

Indeed Cohen’s statement represents a quite limited call for reliance-based relief. As the latter part of the q uotation intimates, Cohen is discussing the repeal or limitation of specific ite ms of tax preference, not all proposed changes to the tax law, a point confirmed by other parts of his text. [117] In fact, in the same statement to Congress, Cohen advocates some changes to existing law that would be effective ret roactive to the date he made the statement. [118] The reason for this limited retroactivity is that the provisions to which it applies are intended to corr ect unintended errors contained in, or opportunities for abuse offered un der, prior law. [119] Thus, if Cohen is guilty of anything regarding the reliance argument, it is his failure to identify which of the conventional justification s for protection of reliance interests apply when Congress repeals tax preferenc e items on the basis of new policy judgments.

A similar example is found in Kaplow’s work. He states:

One of the most commonly noted arguments against allowing private actors to bear losses resulting fr om changes in government policy is that they reasonabl y relied on preexisting law in making investment decisions. Many legal theorists, however, have long recognized the circularity of such arguments, which implicitly ass ume that it is reasonable to expect laws never to change . . . . [120]

Although Kaplow notes that whether reliance is in f act appropriate depends on the circumstances of each case, he does not explore what factors are or should be relevant in the various ca ses. Rather he tends to discount the reasonableness of reliance altogether. [121]

Perhaps ironically, one reason that new view schola rs may have tended to overstate the role reliance arguments pla y under the old view is their insistence on regarding nominally retroactive enactments as essentially similar to nominally prospective enactments that ha ve retroactive effects. Once one has effaced the distinction between nomina lly prospective and nominally retroactive legal change, there is not mu ch point in distinguishing between old-view discussions in whic h reliance is featured as a defense to nominal retroactivity from those in which it is raised as a defense to transition relief understood in the conventional, that is, forward- looking, sense.

Indeed, many of the discussions to which the new-vi ew theorists point as examples of excessive deference to relianc e arguments concern nominally retroactive statutes, not transition reli ef offered in the context of nominally prospective legal change. [122] But the old view does not efface the distinction between nominally retroactive legal cha nge and the purportedly retroactive effects of nominally prospective legal change, for the reasons discussed previously. As a consequence, the old vi ew’s near-blanket rejection of nominal retroactivity on reliance grou nds is perfectly compatible with a much more skeptical view of relia nce as a basis for transition relief. [123] The distinction seems simply to be lost on new vi ew scholars. [124]

Correlatively, and not surprisingly, the idea that reliance interests support transition relief for all nominally prospec tive changes to the tax law does not appear ever to have been seriously enterta ined by anyone, a point well recognized in the transitions literature itsel f. For example, changes in tax rates, [125] including rates on capital income, never have been accompanied by relief, and serious arguments for re lief for such changes have not been made. Similarly, provisions designed to address arrangements that comply with the law but do not pr oduce tax results considered appropriate often have been enacted with at least some retroactive effect. [126] Outside of the tax area, various changes to the l egal landscape routinely have substantial adverse impact s on individuals who have relied on prior law, but the idea that Congres s should make extensive grandfathering relief available is not entertained. In short, the types of legal change for which reliance-based transition relief ( as opposed to reliance- based relief against nominally retroactive legislat ion) has been advocated have generally been considered changes to provision s that invited reliance for reasons other than the mere fact that they were law.

B. The Positive Case for Reliance

In this subpart I argue that the same consideration s that support distinguishing between market-based transitions and legal transitions support the introduction of a concern with reliance . In the end, the concern rests on an acknowledgment that legal rules, and he nce changes to legal rules, derive from authority that is based in large measure on consent. This feature of legal rules is to be contrasted with una nticipated market-based changes, which simply happen. Consent implies that whoever has authority to effect a legal change of one type or another has the authority on the basis of some sort of agreement, however attenuated, betw een those who promulgate the new rule and those who are subject t o it. To the extent that the presence or absence of transition relief bears on reasonable assumptions about the government actor’s conformity with that a greement, reliance may or may not be appropriate.

Underlying this view is the assumption that some po sitive case for reliance exists, for the circularity critique of re liance demonstrates that the fact of reliance cannot, by itself, establish the r easonableness of reliance. At its core, this positive case is the same in the transitions setting as in others. It is simply that if an announced legal ru le is to have an effect on the conduct of parties to whom it purports to apply , they must have some basis to believe in a correspondence between that s tatement of the rule and the rule that turns out to apply in fact. [127] In other words, they must believe it is the actual rule. If the expectation of corre spondence is lacking, the statement of the rule would be pointless. This exp ectation is a feature of language and represents a second difference between natural or market- based change and legal change. Whereas the fact of natural change simply presents itself to investors who must deal with it, the “fact” of legal change depends for its facticity upon the acceptance of th ose to whom the statement of the legal rule applies. A statement o f the rule must have some basis to be trusted if it is to be accepted as the actual rule. That, of course, does not mean the statement is to be taken as true no matter what. There may be explicit or implicit caveats on the circumst ances in which it will continue to be the rule, but if there are no norms whatever that limit the authority of the lawmaker to withdraw or change the rule, then there is no basis to consider the statement of the rule to be t rue in the first place. [128]

These observations indicate that the question is ho w the fact that a legal change concerns prospectively rather than ret roactively effective rules alters the case against protection of reliance inte rests. What is it about the fact that a new rule is made to operate solely with nominally prospective effect, rather than with expressly retroactive effe ct as well, that justifies less solicitude for expectations regarding the fixity of legal rules? Why, in short, is changing the future different from changi ng the past? The following sections address these questions by focus ing on the possible causes of legal change. Notably, one of these sour ces—a change in circumstances—can occur solely going forward, where as the other, a change in judgment about the advisability of a rule can occur both going forward and going back.

In order to illustrate the basis for limited claims of reliance, in the following subparts I explore its role in two settin gs, with the second setting, in turn, having two variations of its own. Each of the two broader settings involves the repeal of the municipal bond interest exclusion. Throughout, I assume that Congress initially adopted the exclusio n because Congress had concluded that state activities should be subsidize d by the federal government on a steady-state basis and that the exc lusion was an appropriate method for doing so. The question is w hether transition relief should accompany a repeal of the exclusion.

1. First Case: Change in Material Circumstances

Under the old view, it would seem that whether repe al of an incentive provision such as the interest exclusion for municipal bonds merits some sort of transition relief depends upon the reasons for the repeal. Suppose in the first case that Congress repeals the exclusion in response to an unforeseen change of material circumstances, suc h as a need to raise revenue. Under these circumstances it does not see m that transition relief should be expected, though one could imagine reason s why in any particular case it might be made available. In thi s setting, one could well say—and it seems for reasons well documented in the literature, it would be preferable to say—that no transition relief should be in the offing absent significant distributional or other considerations that would make relief a good idea, because it remains unclear why relief sh ould presumptively apply when the conditions under which it makes sens e to offer the preference no longer obtain. There is no obvious r eason why investors ought not to be charged with assuming the risk that if circumstances make the law a bad idea, they ought not be entitled to e njoy the benefits of the law. [129] After all, the decision to offer the preference w as made, implicitly or not, on the basis that the circumstances made th e preference a good idea. There is never certainty about the future, and it i s difficult to see why the government, rather than investors, presumptively ou ght to absorb the costs of unexpected developments that make existing law n on-optimal.

2. Second Case: Change in Legislative Judgment

Contrast the preceding situation to this second cas e, in which repeal rests upon a new policy evaluation of essentially t he same basic circumstances. Here the material conditions that w ere thought to support the preference have not changed; rather Congress ha s reconsidered the wisdom of the initial decision to offer the prefere nce under those circumstances. This case presents, in turn, two va riations. The first is where the same legislative body is in place at all relevant times, while in the second a subsequently elected legislature initiates the transition.

(a) First Variation: Same Elected Body

In the first variation, the same elected body both enacts and repeals the tax preference—again, with no change in materia l conditions. The critical issue in this setting is the fact that the passage of time really should play no normative role. As a first step in the ana lysis, consider the difference between a preference that provides benef its on an immediate basis, such as the deduction for certain research a nd experimentation expenditures, [130] and a preference the benefits of which materialize over time, such as the bond interest exclusion. If the research and experimentation deduction were enacted in Year 1 an d repealed prospectively in Year 2, all of the benefits expect ed from expenditures already made by the time of the repeal would have b een realized. By contrast, a prospective, non-grandfathered repeal o f the bond interest exclusion would preclude investors from reaping any of the benefits expected to be realized in the future. [131] Despite these different timing results, it is by now a familiar truth that on a st eady-state basis the two methods of providing a benefit differ only in forma l terms. [132] Assuming no change in law, one could offer the same benefit tha t the bond interest exclusion provides by permitting a deduction for th e purchase price of the bond in the year of purchase and taxing all subsequ ent returns (principal and interest) as they are received. [133] Under either model, the relevant investor decision occurs in the pre-repeal period, but under the deduction- inclusion model, a prospective, non-grandfathered r epeal would not preclude realization of any benefits that investors would be awaiting under the interest exclusion model. In this context, the material equivalence of the two methods under steady states, coupled with t he previously made argument against the propriety of nominal retroacti vity, seems to indicate that grandfathering is appropriate on repeal where Congress chooses to make the form of the preference require its persist ence into future periods in order to be fully realized.

The obvious response to this argument is that it se ems to prove too much, for the same could be said of the case previo usly described, in which repeal takes place under changed circumstances. In that setting it was suggested that Congress’s decision to implement the tax benefit in the form of an interest exclusion rather than as an up-front deduction signaled that investors are at risk of no relief in the event of repeal. [134] The argument just offered, however, suggests that the equivalence und er certain background assumptions of an up-front deduction with an exclus ion over time means the same total benefits should be made available on repeal in either case. It therefore might seem that both claims cannot be cor rect, but this suggestion rests upon a failure to account for the decisive di fference that the different background conditions make for the two cases. It i s precisely the fact that a forward-looking preference presupposes appropriate material conditions that makes appropriate a presumption of no relief w hen these conditions change. In other words, a forward-looking preferen ce signals that Congress intends the preference to be available only as long as the conditions that make it appropriate persist. As long as there is n o change in material circumstances, the forward-looking preference and t he immediate preference remain indistinguishable, for the risk a ssumed under the forward-looking preference has not materialized, an d it was that risk that justified the presumption against transition relief .

One might accept that a difference between changed material circumstances and changed legislative judgments has some significance for transition policy, but query why investors might no t also assume the risk of no relief in the changed-judgment case when a prefe rence is adopted on a forward-looking rather than an immediate basis. Th e difficulty is that if such a presumption obtained, then the enactment of a tax preference whose benefits extended to future periods would rationall y be viewed as equivalent, at the front end, to Congress’s saying that it does not want to provide the preference on a steady-state basis unde r the existing facts, for if it did it would have made the preference available as an immediate deduction. It is entirely unclear why such a signa ling presumption should operate. It needlessly limits Congress’s ability t o offer tax benefits in the form it prefers, when Congress can limit the benefi ts on enactment directly, simply by sunsetting the preference or by making it otherwise conditional. For although a deduction and an exclusion may be th e same on a steady- state basis, they have different properties with re spect to transition periods, and Congress might well prefer to plan for the poss ibility of changed material circumstances in designing the tax prefere nce; removing the presumption that any forward-looking preference com es with a risk of uncompensated change even where material conditions do not change leaves that flexibility available. Consider that w here a change in circumstances makes the preference a bad policy ide a, having offered the exclusion is cheaper than having offered the deduct ion precisely because the exclusion’s benefits materialize later than the deduction’s; the forgone revenue under the exclusion is always less than und er the deduction as long as any bonds are outstanding on the repeal date. B ecause a rejection of the presumption of no relief when relevant circumstance s have not changed does not entail rejecting the presumption when they have changed, there is no benefit to be had from adopting the presumption.

Note, further, that the availability of means by wh ich to make relief tentative other than by having a presumption of no transition relief when the facts have not changed makes the decision not to of fer transition relief in this context irrational. There is literally no rel evant difference on a steady- state basis between the deduction method and the ex clusion method, and given that the case under consideration involves no material change to the facts, the potential penalty of non-grandfathered r epeal would indeed depend upon “mere accounting considerations,” [135] because what obtains is, in effect, a steady state. And this equivalence in dicates, in turn, that failure to provide grandfathering relief in the no-change-o f-circumstances case is analytically equivalent to nominally retroactive ta xation. In other words, the relevant difference between prospective repeal without relief and retroactive taxation is that one can rationally con sider investors to have assumed the risk of changed material circumstances that may make current law sub-optimal, but not that they have assumed the risk of a mere change of mind. And the reason they have not assumed it i llustrates the relevance of reliance. If words (statutes, rules, etc.) are to have meaning, then in the absence of a change to the circumstances in which t hey are set forth, one must be able to rely on them. The assumption of th e former risk is what justifies a presumption of no relief on repeal when circumstances do change. The new transitions literature has made th is point, but it has not adequately distinguished between that case and the change-of-mind case. [136] Where there are no changed circumstances, the diffe rence between changes that illicitly reach back into the past and those t hat account for new present and future conditions disappears, because the past and the future are relevantly the same. The implication of the equiva lence, however, runs in the opposite direction from that assumed by the new transitions literature: where conditions do not change, forward-looking un- grandfathered repeal, at least of incentive provisions, is analytically e quivalent to illicit nominal retroactive taxation.

(b) Second Variation: Subsequently Elected Body

Under the second variation of the case in which mat erial circumstances have not changed, the question is whe ther the analysis runs differently if the repealing body differs from the enacting body. More precisely, the question is whether this difference changes the reasonableness of a presumption that investors do n ot run the risk of an uncompensated repeal of a preference where the mate rial conditions supporting its prior enactment have not changed. T he analysis under first variation indicated that failure to offer relief wa s not rational, because there is no advantage to be gained from the presumption t hat relief would not be forthcoming. A significant difference does exist, however, between that case and the case in which a new legislature repeal s the preference. It is to be found in the parallelism between the latter situ ation and nominally retroactive positive law. The principal objection to such law, it will be recalled, [137] was that it supplanted the elected legislature wit h a different, and unauthorized, legislative body. The converse p roblem arises in the present case. Where the previously enacted tax pre ference materializes during future legislative periods, a requirement of relief on repeal (where material circumstances have not changed) threatens to supplant the current legislature with the past one. In the extreme, the present legislature would be precluded from doing anything at all in the abse nce of a change in material circumstances, if offering relief under pr ior law required its extension sufficiently far into the future. [138]

The same considerations also indicate that the rati onality objection to the presumption of no relief in the same-legisla ture case does not apply here. The rationality objection rests on the assum ption that a rational legislature, though it may change its mind, makes p rovision for doing so ex ante , when it enacts a provision. Under that assumptio n, a legislature rationally signals a decision to make preferences t entatively available under steady-state conditions expressly, by enacting a pr ovision for a limited period of time, rather than by entertaining a presu mption of no relief on repeal. Once the assumption is abandoned that the same legislature that enacts the provision repeals it, no inference about the course of future legislative conduct follows from the enactment of a relief provision in one form rather than another. Accordingly, no inconsis tency or irrationality automatically results from a failure to provide rel ief on repeal of a forward- looking tax preference where material conditions ha ve not changed.

By the same token, the problem of potentially bindi ng future legislatures does not point to an automatic presump tion of no relief where the repealing legislature differs from the enacting one, but rather to the need for a balancing of considerations. On one hand, pr ior legislatures ought not be considered to bind future legislatures on matter s of policy judgment. On the other hand, the same considerations of flexibil ity operate in this setting as in the same-legislature setting. Since investor s know that legislatures change periodically, the same questionable incentiv es would be created if all bets were automatically off with every election . At the other extreme, an obvious effort to lock in changes in law by enac ting forward-looking reforms that necessarily materialize over long peri ods of time ought not be respected by a future legislature that wishes to pu rsue different priorities or takes a different view of the effectiveness of prio r policies. Hence one would expect to find a variety of outcomes in this setting, depending upon the particular circumstances under which the prefer ence was enacted.

A final point is in order here, and it illuminates one reason that no clear transition norms have in fact emerged, and wh y they are unlikely to. In this subpart I have assumed that one can divide cases according to whether or not relevant material circumstances have changed. In practice, the classification of actual cases is likely to be contested because circumstances are constantly changing, and it will frequently be an open question whether the circumstances that have change d are material to a change in law. Wholly apart from concerns of publi c choice—paying off constituencies, log-rolling and the like—the questi on of relief is contestable simply because of the difficulties in classifying c ases. Hence one would expect to see what one does see in the transition c ontext: a variety of transition rules for seemingly similar cases.

C. The Linguistic Basis of Reliance

I have argued that some basis for reliance is impli cit in any statement of a legal rule, because it is a requirem ent of a statement of what the rule is that it be able to be relied upon under at least some circumstances. If language did not have this chara cter, it would be meaningless. Accordingly, the question for relianc e analysis becomes what those circumstances are. The fact that some basis for reliance is presupposed in any statement of the rule does not m ean that reliance is always or even mostly appropriate. It simply means that one must distinguish the cases in which reliance is or shoul d be implied from those in which it need not.

The preceding discussion has suggested what the out er limits of a rational norm of no reliance would be. It is a pos sible rational norm that all bets are off when the circumstances under which a p rior rule was adopted have changed. That is, there is nothing incoherent about an implicit norm that legal rules may be relied upon only as long as the conditions that gave rise to their adoption obtain. But I also have not ed that a change in material circumstances is not the only occasion for a change in legal rules. Sometimes rules change simply because the legislato r has changed its mind about the advisability of the rule even though the circumstances that gave rise to the rule have not changed. In that case, a refusal to protect reliance interests when the rule changes, at least where it is the same legislator that adopted the original rule, becomes inconsistent wit h the idea that the legislator rationally states the legal rule in the first place. It is not really distinguishable from someone’s pronouncing that p , and then immediately pronouncing that not7p . That is, there is something incoherent in an implicit norm that says legal rules may never be re lied upon, or may not be relied upon when they are reversed by the same legi slator that adopted them when there has been no relevant change to the condi tions under which they were adopted. If that were not the case, then one could not rely, for example, on the statements made by anyone, includin g in a contract or, for that matter, in any other setting.

The new view theorists, by failing to distinguish b etween these different circumstances of legal change, reach the conclusion that a norm of no reliance could be perfectly coherent and that fo r other reasons, it may well be. There is some irony in the fact that that view is to some extent undermined by their own occasional acknowledgment t hat a norm of nominal retroactivity would frustrate investor expe ctations: [139] If there is no significant difference, other than in degree, betwe en nominal retroactivity and nominal prospectivity, then there ought not be a concern about adopting a norm of nominally retroactive legislation if reli ance interests are circular and not independently justifiable. Conversely, if that concern is valid, then it ought to apply with nearly equal force to uncomp ensated, nominally prospective change, the only difference being the a forementioned one of degree. [140] In that case reliance lives. Because the new vie w does not pay sufficient heed to the distinction between a change in facts and a change in the rule that applies to those facts, it is unable to explain why either its concerns about nominally retroactive rules ought no t apply to protect reliance interests or, conversely, why a norm of no minal retroactivity ought not apply to changes in positive law. For what is quintessential to a nominally retroactive change, but is typically abse nt in a nominally prospective change, is that circumstances have not changed at all: The past is what it is.


The new transitions literature does not appear to o ffer particularly robust policy recommendations. For the most part, the recommendations are either consistent with those that a proponent o f the old view might offer, or questionable. I have argued that the principal reason for the failings of the literature is that it does not pay heed to the specific conditions under which legal transitions occur. These conditions ar e importantly conventional, and because they are conventional, th e preconditions to the conventions must be satisfied if the conventions ar e to persist; if the preconditions do not persist, then law itself is ca lled into question. The old view gives expression to the preconditions in recog nizing, even if not always with sufficient precision, the relevance of the distinction between nominally retroactive and nominally prospective leg al change, or between law-making and law discernment. [141] In something of an irony, the analogy the new transition literature uses to justify a par ticular set of norms holds only because these conventionally established condi tions obtain and thereby appear natural, but the conditions themselves would not obtain if the rules the new transitions literature offers for many case s became transition norms.

The considerations developed here also suggest that if the authority of the lawmaker did not depend for the most part on the convention that the lawmaker expresses the will of a consenting body, m ost of the difficulties described in this Article would disappear. In that case, transition norms would not affect the conditions that make action un der law possible, because they would not call into question the regim e that individuals must presuppose in order for law and institutions that d epend on law, such as markets, to exist. However, the connection between political and legal authority on one hand, and the will of the individu als subject to law on the other, is more than incidental to the existence of the legal regime. The connection means that transition norms are both mor e limited and more flexible than the recent transitions literature has recognized. They are more limited because they must respect the idea that law derives from an electorate that is self-governing. Both nominal re troactivity in many circumstances, and immediate nominal prospectivity in fewer circumstances, are inconsistent with this idea. Tr ansition norms are more flexible because they may vary according to the typ e of legal transition that different institutions, based upon different conven tions, engage in.

In one sense, these criticisms can be traced back t o a fundamental tenet of the new transitions literature, articulate d first by Graetz, developed by Kaplow, and given perhaps its most comprehensive explication by Shaviro. This is the idea that new rules have a re troactive effect when they have an impact on the future consequences of decisi ons made prior to the adoption of the rule. The idea that such transitio ns “reach back into the past,” in Shaviro’s formulation, implies that one c annot meaningfully distinguish between legal transitions that apply by their terms to completed conduct and those that apply to the present and fut ure consequences of completed conduct—conduct that projects itself into , and assumes the risks of, future developments, whether they be changed ma terial circumstances or different lawmakers with different policy judgme nts. It is on the basis of this view that the new transitions literature downp lays, if it does not dismiss, the distinction between nominal retroactiv ity and nominal prospectivity that has retroactive effect. The con siderations developed in this Article explain why it is sensible to draw a d istinction in kind between forward-looking reforms and nominally retroactive o nes, and why one might want to reserve the term retroactive for the latter as well as, perhaps, for future consequences of past actions where the d istinction between past and future is irrelevant—no change in facts and no change in lawmaker. Future consequences always depend upon future condi tions, whereas past consequences and conduct depend upon nothing; they are complete. It is often appropriate to impute to investors an assumpt ion of the risk that the future will differ from the past, but not that the past does not turn out to be what it was.

The considerations developed here also suggest that the newer transitions literature has not offered much by way of advance over the older view. I have argued that two pillars of the new vi ew – the ideas that legal transitions are closely akin to market changes and that an appeal to reliance interests is circular – are importantly mistaken. Once one removes these pillars, it is not clear that much remains of the n ew view’s general prescriptive edifice. Whether a rule of no relief is appropriate in the legislative setting would seem to depend, as it alw ays has, on whether it made sense for those subject to the old rule to rel y on its persistence. In other words, it is a question that cannot be answer ed apart from an examination of the facts in any particular case. S imilarly, whether a difference in default rule ought to obtain between legislative and judicial action would seem to depend, as it always has, on t he significance of the distinction between making law and finding it. [142]

Where does all of this leave us? Prior to the deve lopment of the new view, the general orientation of the transition s literature seems to have been that the question of the appropriate transitio n rule was to be answered on a case-by-case basis. [143] In other words, the idea that there might an optimal transition norm seems not to have been entertained. The newer literature seeks, instead, to develop a framework f or analysis that largely eschews any examination of the particular circumsta nces of any given change in favor of a search for transition norms. The arguments developed in this Article suggest, however, that the question of the optimal transition norm is no more meaningful than is the question, “What is the best law?” Neither can be answered in the abstract, because ne ither is a question of norms. Rather, each is a question that must be approached in light of circumstances, and addressed to the new challenges as they arise.


  1.  Seminal treatments include those of Michael Graetz, Legal Transitions: The Case of Retroactivity in Income Tax Revision, 126 U. PA. L. REV. 47 (1977) [hereinafter Graetz,Legal Transitions], and Louis Kaplow, An Economic Analysis of Legal Transitions, 99 HARV. L. REV. 509 (1986) [hereinafter Kaplow, Economic Analysis]. More recently, Daniel Shaviro has provided a comprehensive analysis of legal transitions with an emphasis on tax transitions. See DANIEL SHAVIRO, WHEN RULES CHANGE: AN ECONOMIC AND POLITICAL ANALYSIS OF TRANSITION RELIEF AND RETROACTIVITY (2000). Additional extensive discussion is found in 13 J. CONTEMP. LEGAL ISSUES, the first issue of which compiles papers delivered at a symposium on legal transitions.
  2. See Louis Kaplow, Transition Policy: A Conceptual Framework, 13 J. CONTEMP.LEGAL ISSUES 161, 163–68 (2003) [hereinafter Kaplow, Transition Policy].
  3. See, e.g., Kyle D. Logue, Legal Transitions, Rational Expectations, and Legal Progress, 13 J. CONTEMP. LEGAL ISSUES 211, 216 n.5 (2003) [hereinafter Logue, Legal Transitions].
  4. See, e.g., Kaplow, Economic Analysis, supra note 1, at 551 (“The economic analysis presented thus far generally favors a transition policy of nominally prospective implementation of changes in government policy with no transitional relief.”); SHAVIRO, supra note 1, at 229 (generally proposing no transition relief for “policy change retroactive taxes”—i.e., for transitions where the new steady-state rule has different content that the prior one).
  5. See, e.g., Saul Levmore, Changes, Anticipations, and Reparations, 99 COLUM. L. REV. 1657, 1658 (1999) [hereinafter Levmore, Changes].
  6. See Kaplow, Economic Analysis, supra note 1, at 517 (citing, among others, “changes in monetary and fiscal policy” as instances of legal transitions).
  7. Id. at 512.
  8. See SHAVIRO, supra note 1, at 25–26.
  9. Others have criticized the recent transitions literature as unhelpful, though generally on different grounds. See, e.g., Michael Doran, Legislative Compromise and Tax Transition Policy, 74 U. CHI. L. REV. 545 (2007). In a similar vein, Mark Ramseyer and Minoru Nakazato have opposed the now dominant view that transition relief is typically a bad idea by arguing that a requirement of grandfather relief when the law changes enables Congress to insulate itself from lobbying groups interested in retaining prior, presumably worse, law. See Mark Ramseyer & Minoru Nakazato, Tax Transitions and the Protection Racket: A Reply to Professors Graetz and Kaplow, 75 VA. L. REV. 1155 (1989).
  10. For each of these questions, the issue is what to do about the consequences of the transition that result from its having been unanticipated, not about the gains or losses that different individuals experience as a result of the new legal rule on a steady-state basis. Thus, the fact that labor may be less favorably taxed under a consumption tax than under an income tax does not by itself create a transition issue on the shift from an income tax to a consumption tax. However, to the extent that individuals, prior to the transition, allocated more of their resources to the future production of labor income on the assumption that the income tax would continue indefinitely, a transition issue does arise on the shift to a consumption tax. See Kaplow, Economic Analysis, supra note 1, at 516.
  11. I.R.C. § 103 (2010). Graetz made this example the focus of his analysis, and subsequent authors have done so as well. See Graetz, Legal Transitions, supra note 1, passim; Kaplow, Economic Analysis, supra note 1, at 515–16; Saul Levmore, The Case for Retroactive Taxation, 22 J. LEGAL STUD. 265, 266 n.3 (1993) [hereinafter Levmore, Retroactive Taxation] (“The tax-exempt bond example is a favorite of the legal transition literature.”); Kyle D. Logue, Tax Transitions, Opportunistic Retroactivity, and the Benefits of Government Precommitment, 94 MICH. L. REV. 1129, 1133 [hereinafter Logue, Tax Transitions] (“The classic illustration of these concepts involves the repeal of the exemption for interest on state and local bonds.”); SHAVIRO, supra note 1, at 5–8.
  12. See Graetz, Legal Transitions, supra note 1, at 54–57, for a detailed discussion of the consequences of such an un-grandfathered repeal.
  13. See Graetz, Legal Transitions, supra note 1, at 52–53, for a discussion of possible effective dates.
  14. SHAVIRO, supra note 1, at 2–3.
  15. Not uncommonly, changes to positive law or to “legislative” rules are nominally retroactive to the date of announcement in the Congressional Record or the Federal Register. See, for example, Edwin S. Cohen, The Administration’s Interim Program of Tax Reform and Tax Relief, 47 TAXES 325 (1969), for a discussion of an example of this type of retroactivity. In the typical case, such changes address perceived failures of the pre-change regime to give effect to the intent of Congress or the relevant agency on the legal question in issue. Such changes, though legislative in form, are more akin to interpretive clarifications of prior law than to a genuine shift in policy.
  16. See, e.g., Martin S. Feldstein, On the Theory of Tax Reform, 6 J. PUB. ECON. 77, 93–94 (1976).
  17. See Graetz, Legal Transitions, supra note 1, at 49–50, 73–79; Kaplow, Economic Analysis, supra note 1, at 515.
  18. See Graetz, Legal Transitions, supra note 1, at 63–87.
  19. Id. at 56–57.
  20. Id. at 63–87.
  21. In particular, actors would accept or decline the “invitation” based in part on their assessments of the likelihood that preference would be repealed at a time or in a way that would adversely affect their investment. Whether and when such a repeal would occur often would be no more predictable by the enacting legislature than by those to whom it applied. Graetz, Legal Transitions, supra note 1, at 64–66.
  22. For an extension of the argument, see Kaplow, Transition Policy, supra note 2, at 170.
  23. Graetz’s criticisms of the reliance theory included the following: (1) the extent of reliance depends on individual assessments of the probability of legal change that can vary from person to person and that in the end provide no stable benchmark for assessing whether the reliance is reasonable; (2) the reasonableness of reliance depends on developments that change the calculation of whether reliance is reasonable; (3) apparently formal distinctions (such as whether the possibility of future repeal is made express or not) in the framing of legislation can affect the reasonableness of reliance; (4) relief tends to protect only a subset of those who relied on the old law; and (5) individual reliance on the possibility that a provision will be enacted does not seem different from reliance on a provision that is enacted and then quickly repealed, yet reliance in the former case does not entitle individuals to relief when the provision is not enacted. Graetz, Legal Transitions, supra note 1, at 74–78.
  24. See Graetz, Legal Transitions, supra note 1, at 49–50.
  25. See Logue, Legal Transitions, supra note 3, at 216–20 (reviewing the new view approach to legal transitions).
  26. See Kaplow, Economic Analysis, supra note 1, at 517. The “government action” qualification is arguably not explicit, though Kaplow discusses it at various points in the exposition. See, e.g., id. at 517–18 (discussing changes in “government policy”). Further, if the qualification is not assumed, it becomes impossible to distinguish legal transitions from other kinds of transitions or events, such as those brought about by private actors. For a criticism of the definition as overbroad to the point of including virtually all activity that takes place or has consequences over time, see Frederick Schauer, Legal Development and the Problem of Systemic Transition, 13 J. CONTEMP. LEGAL ISSUES 261, 263–65 (2003). See also discussion infra Part III.A.
  27. The requirement of unexpectedness also implies that many transitions in the conventional sense do not raise transition issues. For example, a change to tax rates scheduled to take effect in one year does not raise a transition issue.
  28. See Kaplow, Economic Analysis, supra note 1, at 517; see also Kaplow, Transition Policy, supra note 2, at 165 n.5.
  29. See SHAVIRO, supra note 1, at 25–26.
  30. See Logue, Legal Transitions, supra note 3, at 211–12.
  31. Id. at 250.
  32. See Kaplow, Economic Analysis, supra note 1, at 516; see also Logue, Legal Transitions, supra note 3, at 239–42.
  33. See Graetz, Legal Transitions, supra note 1, at 63–65 (noting that the law may change in unpredictable ways as tastes and other conditions change).
  34. SHAVIRO, supra note 1, at 112.
  35. Kaplow, Transition Policy, supra note 2, at 177.
  36. See Kaplow, Economic Analysis, supra note 1, at 515–19. Following Kaplow, I use the term “market risk” to include “all risks—including the risk of natural events such as floods—not caused by uncertainty concerning future government policy.” Id. at 533 n.62.
  37. See id. at 541–42.
  38. Whether nominal prospectivity constitutes transition relief is a point of contention. See infra Part IV.
  39. See Kaplow, Transition Policy, supra note 2, at 177–79.
  40. One might object that forcing such individuals to bear these costs unfairly penalizes those who may lack the means to purchase insurance or to rebuild, possibly through no fault of their own. The usual response is that such hardships are due to preexisting distributive inequities that should be dealt with in other ways, such as through taxand- transfer programs that are not keyed to specific losses. Programs of this nature do not effect distributive objectives by means of inefficient subsidies. See id.
  41. Graetz himself raised this point, but it has been more fully developed in the later literature. See Graetz, Legal Transitions, supra note 1, at 65–66.
  42. Levmore, Changes, supra note 5, at 1662.
  43. See Logue, Legal Transitions, supra note 3, at 217 n.10 for authorities.
  44. SHAVIRO, supra note 1.
  45. See id. at 64–91.
  46. As an example of refinement, Shaviro disaggregates the idea of a transition tax (or subsidy) from that of transition risk. The former is the component of a transition reflecting an expected overall reduction (increase) in returns from an anticipated rule change offered without transition relief, while the latter is simply the effect that uncertainty about the future has on asset values, even though on an ex ante basis the expected value of an asset subject to a future rule change remains constant. See SHAVIRO, supra note 1, at 27–32.
  47. “This book’s framework is utilitarianism.” Id. at 4.
  48. See id. at 104.
  49. See id. at 104–10.
  50. Id. at 104. Shaviro’s analysis thus may be thought of as developing observations first made by Graetz and later developed by Kaplow and others on this point.
  51. Id. at 104–05.
  52. Although Shaviro does not expressly state that new rules of the road should be accompanied by transition relief, he elsewhere argues for transition relief for “accounting rule changes”—that is, changes that do not involve an alteration of legal entitlements on a steady-state basis. Id. at 53–54 (explaining accounting rule changes), 102–03 (arguing for transition relief for such changes). Rules of the road would seem to fall into this category.
  53. Id. at 105–07.
  54. Id.; see supra Part II.B.
  55. SHAVIRO, supra note 1, at 106.
  56. Id.
  57. Id.
  58. Id. at 107.
  59. One example is Stephen R. Munzer, Retroactive Law, 6 J. LEGAL STUD. 373 (1977).
  60. Kaplow, Transition Policy, supra note 2, at 165.
  61. SHAVIRO, supra note 1, at 64–91.
  62. Stephen Munzer introduced a similar distinction in an early article, though it is not precisely the same as that here. See Munzer, supra note 59, at 388 (“The legal status of an act depends not on something inherent to that act, but on the relation the act has to the law or laws governing its status.”). The entitlement-transition rule distinction also is arguably implicit in Eric Chason’s treatment of legal transitions. See Eric Chason, The Economic Ambiguity (and Possible Irrelevance) of Tax Transition Rules, 22 VA. TAX REV. 615, 620 (2003) (defining transition rules in temporal terms and characterizing the definitions as formal).
  63. Kaplow analyzes the question in terms of changes to the scheduled legal landscape rather than to future entitlement rules, a formulation broad enough to capture situations in which future law features planned changes to entitlement rules. Kaplow, Transition Policy, supra note 2, at 165 n.5. I dispense with a consideration of known future variation in law, because an evaluation of it would require introduction of considerable complexity without altering the basic argument. The complexity results from having to distinguish between entitlements that change where, by hypothesis, the change does not raise a transition issue, from entitlement changes that do.
  64. The meta-character of the transition question has been noted in the literature. See, e.g., id. at 169 (“[T]he [subject of transition] involves a sort of meta-problem.”). The literature also has addressed the still higher-order question of what the appropriate transition to the appropriate transition policy should be. See Kaplow, Economic Analysis, supra note 1, at 557–60; Kirk J. Stark, The Elusive Transition to a Tax Transition Policy, 13 AM. J. TAX POL. 145 (1996).
  65. Again, this definition was needed in order to support a conception of legal transitions capacious enough to include all alterations to legal entitlements, no matter their provenance, as the new view literature requires, while excluding changes to the value of investments that result from events that fall short of actual rule changes. A more conventional understanding of legal transitions would view them as changes to positive law (and perhaps “legislative regulations”) and therefore could bundle together what I have called the entitlement rule and the transition rule in the definition of legal transition.
  66. One might counter that the rule obtained by the later-enacted or -adopted rule, rather than the earlier, constitutes a transition rule, but this rule simply specifies the source of law, not the determination of when law that emanates from that source applies. I thank Scott Shapiro for alerting me to this issue.
  67. See Munzer, supra note 59, at 374 (“Under [the declaratory theory], if a court ‘overrules’ an earlier decision, it does no more than declare what the law has always been and apply it to the case at hand.”). There are serious difficulties with the declaratory theory, but the notion that judicial interpretation offers a pronouncement of what the statute or rule was at the time of conduct in question need not rely on the declaratory theory.
  68. See, e.g., U.S. v. Kubrick, 444 U.S. 111, 117 (1979) (identifying finality concerns that support statutes of limitations).
  69. Whether default norms operate in specified contexts to limit the effect of change temporally is a separate issue that merely goes to expectations that have developed and the reasons for them; it does not go to the nature of legal transitions themselves. For example, many legal changes ordinarily have retroactive effect. The most common example is a new judicial interpretation of any pre-existing legal rule or doctrine. See SHAVIRO, supra note 1, at 112–15.
  70. See, e.g., id. at 112 (“It has long been a truism that ‘statutes operate only prospectively, while judicial decisions operate retrospectively.’” (quoting Rivers v. Roadway Express, Inc., 511 U.S. 298, 311–12 (1994))). Again, the notion that the absence of an express transition rule implies that a default transition rule is in effect follows from the demands of the welfarist conceptual apparatus. A more conventional understanding of the concept of a legal transition would permit one to view a transition rule as in effect only when the effective date departed from the legislative default. This understanding, however, would be based on the notion that a change to positive law is a different kind of thing from a new interpretation of existing law. If one is prepared to make that distinction, then one may consistently go on to view the change from the pre- to the post-enactment regime as intrinsic to a legal transition, and therefore to view transition rules as applying only when an exception is made to the general rule of date-of-enactment effectiveness.
  71. U.S. CONST., art. I, § 9, cl. 3.
  72. Logue, Legal Transitions, supra note 3, at 211; c.f. Kaplow, Economic Analysis, supra note 1, at 524 (discussing a product discovered to be unsafe on the basis of new scientific evidence).
  73. Logue, Legal Transitions, supra note 3, at 239–41.
  74. Id.
  75. Id. at 239–40.
  76. Kaplow, Transition Policy, supra note 2, at 177. For other statements, see Michael J. Graetz, Retroactivity Revisited, 98 HARV. L. REV. 1820, 1825 (1985) [hereinafter Graetz, Retroactivity Revisited] (“The costs of uncertainty in the law do not seem necessarily different in kind or in magnitude from the costs of uncertainty in markets . . . .”); see also Logue, Legal Transitions, supra note 3, at 221 (noting the reasonableness of the market analogy assuming investors are rational).
  77. Kaplow, Economic Analysis, supra note 1, at 520. See also Kaplow, Transition Policy, supra note 2, at 179 (“Because government-created risk is analogous in relevant respects to market and natural risks and because government mitigation of gains and losses is inefficient with regard to the latter, it follows that it must be inefficient with regard to the former.”); Chason, supra note 62, at 616–17 (describing the Graetz-Kaplow view as based on the analogy between the risk of repeal and the risk of market-based or casualty losses).
  78. Kaplow, Transition Policy, supra note 2, at 177; see also supra Part IV.A.
  79. Note that Kaplow himself does not accept the analogy fully, for if transition uncertainty were “analytically equivalent to general market uncertainty,” Kaplow, Economic Analysis, supra note 1, at 520, the notion that it might occasionally be appropriate to make certain new rules nominally retroactive would be false, contrary to Kaplow’s assertion. Id. at 551 (“Sometimes new legal rules should be made fully retroactive: they should be applied to time periods before the enactment date, even as to investments no longer in existence.”).
  80. The quality of the new entitlement as better or worse than the old one and the differing possibilities for compromise of the legal actor’s efforts to act for a public benefit, depending on the legal actor implementing the change, are the two bases that writers in the new view have identified as reasons for different transition norms. Thus, Kaplow acknowledges that different norms for different types of government actor may be appropriate, but he does not ground the observation in the different natures of what the actor does; rather the focus is on the susceptibility of different government actors to different forms of suboptimal behavior. Kaplow, Transition Policy, supra note 2, at 190–200.
  81. Levmore, Changes, supra note 5, at 1672. To be clear, here I abstract from considerations relating to “good government” concerns that might counsel in favor of different norms for different government actors. See supra note 80. Rather the question is whether the distinctive types of activity that different government decision-makers engage in might support different transition norms for these actors. On this point the literature is silent.
  82. See, e.g., Levmore, Retroactive Taxation, supra note 11, at 277–78; SHAVIRO, supra note 1, at 21–22.
  83. For a general description of the incentive effects of income and consumption taxes, see Joseph Bankman and David A. Weisbach, The Superiority of an Ideal Consumption Tax Over an Ideal Income Tax, 58 STAN. L. REV. 1413 (2006).
  84. Kaplow, Economic Analysis, supra note 1, at 551 (arguing that a “new rule” should be applied retroactively when the new rule is better tailored than the old rule to the circumstances that previously obtained).
  85. See Harold M. Hochman, Rule Change and Transitional Equity, in REDISTRIBUTION THROUGH PUBLIC CHOICE 320, 331 (Hochman and Peterson eds., 1974) (noting the problem of violation of consent that arises in the similar case in which earlier generations set fixed rules for later ones).
  86. This observation may help to explain the courts’ oft-stated view that retroactivity that extends to the beginning of the current legislative period raises no constitutional question, but retroactivity extending much beyond it does. See U.S. v. Carlton, 512 U.S. 26, 38 (1994) (O’Connor, J., concurring) (“A period of retroactivity longer than the year preceding the legislative session in which the law was enacted would raise, in my view, serious constitutional questions.”).
  87. The claim is similar to Rousseau’s observation in On the Social Contract that legitimate rule must be unanimously authorized. See JEAN-JACQUES ROUSSEAU, ON THE SOCIAL CONTRACT 52 (Roger D. Masters ed., Judith R. Masters trans., 1978) (1762) (“The law of majority rule is itself an established convention, and presupposes unanimity at least once.”).
  88. One could refine the point as follows: Certain tacit rules conceivably could be derived, but they must derive from features of a relevant bargaining situation such that no agent in that situation could rationally oppose them; identifying rules of this sort is the subject of social contract theory. See, e.g., JOHN RAWLS, A THEORY OF JUSTICE (1971). It is difficult to see how one such rule would ever include an authorization for a legislature to override a previous legislature with respect to the rules the previous legislature adopted for the period during which it was to govern, unless the terms permitting such an override were already expressly agreed to before the override occurred.
  89. See supra Part II.B.
  90. See Kaplow, Economic Analysis, supra note 1, at 515–16; Levmore, Retroactive Taxation, supra note 11, at 266–78. See also Stark, supra note 64, at 154 (“Given individuals’ forward-looking behavior, the difference between prospective laws and retroactive ones is really just a matter of degree, not of kind.” (summarizing Graetz’s view)).
  91. See Graetz, Retroactivity Revisited, supra note 76, at 1826.
  92. In some cases, nominal prospectivity does raise difficulties similar to those raised by the nominal retroactivity case. In particular, where there is no meaningful change between the circumstances of enactment and those of repeal, failure to provide some form of transition relief starts to look like nominal retroactivity. These reasons are explored in the next Part.
  93. See Munzer, supra note 59, at 391 (“A retroactive law makes it difficult or impossible to act with knowledge of the applicable law. The strongest way of putting this is to say that if a law is retroactive, it logically cannot guide behavior with respect to the act whose legal status is altered.”).
  94. It should be noted, however, that if the law had this character, it is still not clear that any incentive problem would exist. If government’s authority is predicated on its wisdom, then subjects should simply follow its commands. Rapid changes in the law would not supply a basis for reticence about taking advantage of preferences, because the changes would be correct.
  95. For purposes of the discussion, I assume that parties would continue to have an incentive to bring suit even where relief depended upon adoption of a novel and therefore nominally prospective interpretation of existing law, since the question here relates to the different matter of the larger consequences of a prospectivity norm. The principle could be implemented by providing for a government payment to the aggrieved party where the court found for the plaintiff on the basis of a novel interpretation of the law.
  96. See Munzer, supra note 59, at 374–75; SHAVIRO, supra note 1, at 112.
  97. A salient example is the California Supreme Court’s adoption of a strict liability standard for product defects. Greenman v. Yuba Power Prods., Inc., 59 Cal. 2d 57 (1963).
  98. See SHAVIRO, supra note 1, at 112 (“To some extent, the nominally retroactive application of judicial decisions is logically inherent in what we call the judicial function.”); see also, e.g., Rivers v. Roadway Express, 511 U.S. 298, 312–13 (1994) (“It is this Court’s responsibility to say what a statute means, and once the Court has spoken, it is the duty of other courts to respect that understanding of the governing rule of law. A judicial construction of a statute is an authoritative statement of what the statute meant before as well as after the decision of the case giving rise to that construction.”); Id. at 313 n.12 (“[W]hen this Court construes a statute, it is explaining its understanding of what the statute has meant continuously since the date when it became law.”).
  99. Kaplow, Economic Analysis, supra note 1, at 551. See also Logue, Tax Transitions, supra note 11, at 1154–58 (distinguishing the case for transition relief on the basis of whether Congress has acted opportunistically or not, and on the basis of the leastcost avoider in the case of transitions motivated by the discovery of new information).
  100. See, e.g., Scott J. Shapiro, The “Hart-Dworkin” Debate: A Short Guide for the Perplexed 41–52 (University of Michigan Law School Working Paper No. 77) (developing a positivist account of disagreement over theoretical legal principles), available at (last visited May 14, 2010).
  101. Id. at 16.
  102. Marbury v. Madison, 5 U.S. (1 Cranch) 137, 177 (1803) contains probably the most famous articulation of this principle in the constitutional law context. See also WILLIAM BLACKSTONE, COMMENTARIES ON THE LAWS OF ENGLAND 69 (1765) (stating that the court’s duty was not to “pronounce a new law, but to maintain and expound the old one.”), quoted in Munzer, supra note 59, at 374.
  103. For sample discussions of non-socially optimizing conduct, see generally SHAVIRO, supra note 1, at 64–91.
  104. There appears to be some recognition of this claim in the literature. For example, in his review of Shaviro’s book on legal transitions, Logue notes that the “concrete policy payoff” of Shaviro’s extended argument is ultimately disappointing. Kyle D. Logue, If Taxpayers Can’t Be Fooled, Maybe Congress Can: A Public Choice Perspective on the Tax Transition Debate, 67 U. CHI. L. REV. 1507, 1509 (2000) (internal quotation marks omitted). Similarly, Barbara Fried notes that the philosophical literature grounded in “luck egalitarianism,” which is not significantly distinguishable in orientation from the “old view,” has reached more or less the same set of policy recommendations as the new view. Barbara Fried, Ex Ante/Ex Post, 13 J. CONTEMP. LEGAL ISSUES 123, 124 (2003) (“The organizing premise of luck egalitarianism . . . that a just state would equalize the effects of brute luck on individual endowments, but leave individuals to bear the consequences (good and bad) of their choices . . . has led luck egalitarians to more or less the same conclusion that ‘rational expectations’ economists have reached on welfarist grounds . . . .”).
  105. See supra Part IV.
  106. See, e.g., Kaplow, Economic Analysis, supra note 1, at 522 n.26 (citing commentary).
  107. See supra Part III.
  108. Kaplow, Economic Analysis, supra note 1, at 531. See also Graetz, Legal Transitions, supra note 1, at 77–78; Graetz, Retroactivity Revisited, supra note 76, at 1823–24; and SHAVIRO, supra note 1, at 2–3, for further discussion of this issue. Kaplow also considers and rejects other arguments that might support transition relief. These include forcing the government to internalize harms, more fairly distributing the burdens and benefits of government, and maintaining horizontal equity. Kaplow, Transition Policy, supra note 2, at 170. I do not discuss them here.
  109. See supra Part III.
  110. For purposes of this discussion, I bracket the point made earlier that Kaplow does not articulate a well-defined concept of transition relief. See supra Part III.A.
  111. See supra Part IV.A.
  112. See supra Part III.
  113. In essence, in Kaplow’s view the most that could be said about the expectations argument is that it raises the question of the proper transition to the proper transition regime. Kaplow, Economic Analysis, supra note 1, at 557–60. Kaplow notes that the questions this issue raises are distinct from those of the proper “steady state” transition policy. Since it is currently reasonable to expect transition relief, it may not be appropriate to move to another transition norm without offering second-order transition relief. Id.
  114. Graetz, Retroactivity Revisited, supra note 76, at 1823 (“In short, the argument asserts that people have a right to protection merely because either they now expect such protection or they expected such protection when they entered into a transaction; their expectations allegedly create a right and their asserted rights legitimate their expectations.”); Kaplow, Transition Policy, supra note 2, at 170 (“[I]t is increasingly recognized that arguments based on actors’ reliance on and expectations concerning existing law are circular and otherwise deficient.”).
  115. Graetz, Legal Transitions, supra note 1, at 74 (quoting Cohen, supra note 15, at 327).
  116. Id.
  117. Graetz omits the first word of the first sentence quoted: “These.” It refers to what Cohen calls “the major tax preferences.” Cohen, supra note 15, at 327. They include such items as accelerated depreciation deductions, tax incentives for oil exploration, and charitable-contribution deductions. Id.
  118. See, e.g., id. at 336 (“We believe that the distortions of income tax liability involved in [so-called ABC transactions involving mineral production payments], and increasing utilization in various extractive industries, indicate that these distortions should be terminated promptly . . . . We recommend, therefore, that this provision be enacted as promptly as possible and be effective with respect to transactions consummated, or covered by a binding contract entered into, on or after April 22, 1969 [i.e., the date Cohen delivered the statement to the House Ways and Means Committee]. The industries involved have had adequate notice that the tax treatment of production payments was under reconsideration.”).
  119. Id. at 346–47.
  120. Kaplow, Economic Analysis, supra note 1, at 522 (footnotes omitted).
  121. Id. at 525. See also Kaplow, Transition Policy, supra note 2, at 170 (“Although commonplace in most transition settings, it is increasingly recognized that arguments based on actors’ reliance on and expectations concerning existing law are circular and otherwise deficient.”).
  122. For example, three of the four commentaries that Kaplow cites in support of the pervasiveness of the reliance norm are expressly concerned with the validity of reliance arguments against nominally retroactive legislation. See Kaplow, Economic Analysis, supra note 1, at 522 n.26 (citing Committee on Tax Policy, Tax Section, New York State Bar Association, Retroactivity of Tax Legislation, 29 TAX LAW. 21, 22–23 (1975); Note, Setting Effective Dates for Tax Legislation: A Rule of Prospectivity, 84 HARV. L. REV. 436, 439–42 (1970); Alan S. Novick & Ralph I. Petersberger, Retroactivity in Federal Taxation: Part II, 37 TAXES 499, 499–504 (1959). The fourth commentary adduces reliance as a basis for compensation in the case of prospective changes to zoning laws. Id. (citing Developments in the Law: Zoning, 91 HARV. L. REV. 1427, 1494–96 (1978)). Even here, however, the cited author notes that the strength of the reliance argument varies with the nature of the zoning ordinance on which it is based. Id. at 1495–96. In their counterpart discussions of the old view of reliance, Graetz and Logue cite some of the same literature. See Graetz, Legal Transitions, supra note 1, at 73 n.78; Logue, Tax Transitions, supra note 11, at 1136 n.27.
  123. See, e.g., Cohen, supra note 15, at 346 (advocating an effective date of the date of announcement before committee for a provision designed to correct an error in the tax law).
  124. Graetz remarks that “[t]he evaluation of claims for transitional relief should be guided not by the existence of expectations, but rather by some independent normativevision of what people should be entitled to expect.” Graetz, Retroactivity Revisited, supra note 76, at 1823. The statement is true, but without more, it does not demonstrate that reliance arguments may not succeed in explaining why transition relief should be offered in certain circumstances.
  125. Although most ordinary income is taxed in the year earned, and hence does not raise a transition question assuming there is no nominal retroactivity to a later change in law, a long-term contract entered into under prior law would raise a transition issue. Thus, a taxpayer who entered into a multi-year contract when the applicable marginal was 25% would not be thought entitled to have that rate grandfathered if rates went to 35% before the term of the contract.
  126. See Note, Setting Effective Dates for Tax Legislation, 84 HARV. L. REV. 436, 438–42 (1970) (noting retroactive aspects of tax legislation).
  127. The reliance argument is sometimes framed as resulting from a notion of quasicontract between the government and private actors. It is better framed as resulting from a quasi-promise, both because reliance damages can arise out of promissory obligations and because there is no analog to consideration in contract that runs from those subject to legal rules to those who promulgate them.
  128. Indeed one could reasonably characterize the central task of modern political theory beginning with Hobbes as finding a way to ensure that and which statements of legal rules may be relied upon. See, e.g., THOMAS HOBBES, LEVIATHAN 129 (Michael Oakeshott ed., 1962) (arguing that security is possible only in civil society in which a single legislator has the power to announce and enforce the law); IMMANUEL KANT, PERPETUAL PEACE AND OTHER ESSAYS 79 (Humphrey trans.1983) (arguing that the concept of right requires a sole legislator possessing unlimited power to impose coercive law as long as it is not selfcontradictory as law).
  129. SHAVIRO, supra note 1, at 48.
  130. I.R.C. § 174 (2010).
  131. See supra Part III.B.
  132. This result is a statement of the so-called Cary Brown Theorem. Cary Brown, Business Income Taxation and Investment Incentives, in INCOME, EMPLOYMENT AND PUBLIC POLICY: ESSAYS IN HONOR OF ALVIN H. HANSEN 300 (1948). It depends upon certain simplifying assumptions that are not relevant to the present analysis, such as that the relevant tax rates are the same at all times and that the investor can make full, current use of the deduction for the initial investment. Kaplow discusses this equivalence, as well as the economic equivalence between a nongrandfathered repeal of benefits that had been anticipated to accrue in the future and the same repeal coupled with taxing back benefits that had been provided as an up-front subsidy. See Kaplow, Transition Policy, supra note 2, at 188–89. See also SHAVIRO, supra note 1, at 218 (noting the equivalence in the methods for the municipal bond case).
  133. Consider a taxpayer in the 30% tax bracket who earns $100 and uses the after-tax proceeds to purchase a ten-year municipal bond that pays 10% annually. Under the interest exclusion regime, she pays $30 in tax on her earnings, purchases a $70 bond, receives $7 annually in untaxed interest, and receives her $70 back at term tax-free. Under the deduction regime, she pays no tax on the $100 of earnings, purchases a $100 bond, earns $10 annual interest on which she pays $3 tax, and receives her $100 back at term, at which time she pays $30 in tax.
  134. See generally Logue, Tax Transitions, supra note 11, at 1181–94, for a discussion of methods by which Congress might bind itself to transition relief.
  135. SHAVIRO, supra note 1, at 106. Shaviro uses the quoted language to describe the definition of nominal retroactivity.
  136. Kaplow, for example, notes that where changes to the facts (as opposed to a reevaluation of the wisdom of a rule) give rise to a legal transition, retroactive effectiveness would not be appropriate. Kaplow, Economic Analysis, supra note 1, at 551. See, e.g., supra note 131 and accompanying text.
  137. See supra Part IV.B.
  138. Thus, one could imagine a case in which a late-term Congress passes a suite of relief provisions all of which materialize over ten or more years, and where the reason for passing the provisions in this form was to lock in current legislative judgments. This example has a parallel in the case of executive action, where an executive about to leave office issues or finalizes a large number of orders and regulations to entrench its policy preferences. See Nina Mendelson, Agency Burrowing: Entrenching Policy and Personnel Before a ‘ew President Arrives, 78 N.Y.U. L. REV. 557 (2003).
  139. See, e.g., Levmore, Retroactive Taxation, supra note 11, at 278 (observing that repeated use of retroactive taxation would undermine its usefulness as a surprise to raise tax revenue efficiently).
  140. See supra Part II.A.
  141. See, e.g., Novick & Petersberger, supra note 122, and authorities cited therein.
  142. See Munzer, supra note 59, at 374–75.
  143. This is the general assumption that Cohen makes. See Novick & Petersberger, supra note 122, at 500.