Retroactivity and Prospectivity of Judgments in American Law

Abstract

In every American jurisdiction, new rules of law announced by a court are presumed to have retrospective effect – that is, they are presumed to apply to events occurring before the date of judgment. There are, however, exceptions in certain cases where a court believes that application of the new rule will upset serious and reasonable reliance on the prior state of the law. This chapter summarizes these exceptional cases. It shows that the proper occasions for issuing exclusively or partially prospective judgments have varied over time and that there are still substantial differences in approach according to the particular jurisdiction and the kind of law under consideration. The chapter concludes with a brief survey of some of the still unresolved jurisprudential and constitutional problems raised by recognition of the power of courts to issue non-retroactive judgments.

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

Appellate decisions generally consist of two elements – the resolution of a dispute and a statement of law explaining that resolution (Stone 1985, p. 188). The resolution of the dispute is necessarily only retroactive – a judge cannot resolve a case before it arises. Therefore, this report discusses prospectivity and retroactivity only with respect to the general statements of law explaining those resolutions. Because such statements are a court’s best explanation of the legal rules governing those facts, it follows that they should also apply to cases with the same facts that arise after the judgment. Every reasoned judgment, therefore, is always at least prospective. But, if a court thought its reasoning was appropriate for facts that arose before the judgment in the one case before it, why should it not apply that reasoning in other cases based on facts that occurred before that judgment? In this sense, it seems logical to apply the legal rules announced in a judgment retroactively as well.

The idea that rules declared on Day Two govern people’s actions on Day One raises immediate alarms. In American law, as in most law, retroactive rules are disfavored. The United States Supreme Court expressed the prevailing attitude thus:

[T]he presumption against retroactive legislation is deeply rooted in our jurisprudence, and embodies a legal doctrine centuries older than our Republic. Elementary considerations of fairness dictate that individuals should have an opportunity to know what the law is and to conform their conduct accordingly; settled expectations should not be lightly disrupted. For that reason, the “principle that the legal effect of conduct should ordinarily be assessed under the law that existed when the conduct took place has timeless and universal appeal.” In a free, dynamic society, creativity in both commercial and artistic endeavors is fostered by a rule of law that gives people confidence about the legal consequences of their actions.

Landgraf v. USI Film Prods., 511 U.S. 244, 265–66 (1994) (quoting Kaiser Aluminum & Chem. Corp. v. Bonjorno, 494 U.S. 827, 855 (1990), (Scalia, J., concurring) (citations omitted)). Of course, even statutes written to apply only prospectively have a necessarily retroactive effect. People reasonably assume that the law will stay as it is when planning some activities, so any change in the law will upset some settled expectations (Fisch 1997, p. 1087; Alexander and Sherwin 2001, p. 152). But, apart from this inevitable effect, American courts interpreting legislation indulge a strong presumption against retroactivity. Landgraf (1994, p. 265).

Several provisions of the Unites States Constitution are motivated, at least in part, by concerns about the evils of retroactive law. Article I, Sections 9 and 10 prohibit “ex post facto” laws,¹ although the Supreme Court has interpreted those provisions to prohibit only laws imposing or increasing criminal penalties on past conduct. Calder v. Bull, 3 U.S. (3 Dall.) 386 (1798). Article I, Sections 9 and 10 also prohibit “bills of attainder” or legislative declarations of criminal guilt. Article I, Section 10, Clause 1 forbids states – but not the federal government – from “impairing the Obligation of Contracts.” The Fifth Amendment prohibits the taking of property without just compensation.² Finally, the Fifth and Fourteenth Amendments prohibit state and federal governments from depriving any person of “life, liberty or property without due process of law.” Due process has been construed to require that every law be a rational means of achieving a legitimate public purpose. Williamson v. Lee Optical of Okla., Inc., 348 U.S. 483, 488 (1954). In meeting this requirement, “retroactive legislation [has] to meet a burden not faced by legislation that has only future effects.” Pension Benefit Guar. Corp. v. R.A. Gray & Co., 467 U.S. 717, 730 (1984). But it is enough if the retroactivity of the law is itself rationally justified. Pension Benefit Guar. Corp., (1984, p. 730).

Notwithstanding this well-established hostility to retroactive legislation, the presumption is quite the opposite when it comes to the judgments of courts. The strong presumption is that statements of law contained in judgments – even when they announce new rules or overturn old ones – apply to conduct predating that judgment.

This seeming inconsistency derives from the “declaratory” theory of adjudication – the historical view of legislatures as making new law and of courts as finding and declaring pre-existing law.³ The theory had its roots in the Blackstonian understanding of judgments as merely “the principal and most authoritative evidence” of a law with a prior and independent existence (Blackstone 1765, Vol. 1, p. *69). Under this view, even when a court announced a new common law doctrine, it was merely restating objective reason – or “established custom” or “divine law”, something which had always existed and had never varied (Blackstone 1765, Vol. 1, pp. *69–70). Courts, that is, are assumed to engage in interpretive not creative acts.⁴ Joseph Story, a preeminent early American legal authority, embraced this idea with enthusiasm. Legal rules, he claimed, were “antecedent” to judicial decisions and the latter were valuable only for “their supposed conformity to those rules” (Story 1852, pp. 503, 506). The theory that legislatures make new rules while courts simply recognize controlling rules that have been there all along explains why rules announced by the legislature have only prospective effect while those announced by judges have retroactive effect as well. Partly for this reason, the constitutional provisions that prohibit retroactive laws generally have been held inapplicable to judicial acts.⁵

Modern jurisprudence, of course, has largely debunked this simple picture and has recognized an inevitable law-making power in courts.⁶ From this recognition, however, it should follow that the retroactivity of rules arising from adjudication is as worrisome as that associated with legislation. Once we recognize that courts also may formulate “new rules,” we must account for cases in which people have acted in substantial and reasonable reliance on the previous state of law. Why not apply a new judicial rule, like a new legislative rule, only to the future?

An example from 1848 illustrates the point. Bingham v. Miller, 17 Ohio 445 (1848). The defendant in a contract action pled that she was a feme covert and thus immune to legal action. In response, the plaintiff introduced into evidence an act of the Ohio legislature dissolving the defendant’s claimed marriage. The Supreme Court of Ohio held that legislative divorce was unconstitutional under the doctrine of separation of powers. Nevertheless it affirmed the judgment for the plaintiff (Bingham 1848, p. 447). The legislature had exercised this power for over 40 years and to:

declare all the consequences resulting from [legislative divorces] void, is pregnant with fearful consequences. If it affected only the rights of property, we should not hesitate; but second marriages have been contracted, and children born, and it would bastardize all these, although born under the sanction of apparent wedlock, authorized by an act of the legislature before they were born …. On account of these children, and for them only, we hesitate …. [W]e are constrained to content ourselves with simply declaring that the exercise of the power of granting divorces, on the part of the legislature, is unwarranted and unconstitutional …. We trust we have said enough to vindicate the constitution, and feel confident that no department of state has any disposition to violate it, and that the evil will cease.

Bingham (1848, pp. 448–49).

In the early twentieth century, as the force of the declaratory theory began to wane, the idea of limiting judgments’ effect to future transactions was increasingly proposed as a reasonable approach when courts created new legal rules and especially when they overruled established precedent. In 1921, Chief Judge (as he then was) Benjamin Cardozo noted that in some cases:

when the hardship [of the retroactive effect of judge-made law] is felt to be too great or to be unnecessary, retrospective operation is withheld …. It may be hard to square such a ruling with abstract dogmas and definitions. When so much else that a court does, is done with retroactive force, why draw the line here? The answer is, I think, that the line is drawn here, because the injustice and oppression of a refusal to draw it would be so great as to be intolerable.

(Cardozo 1921, pp. 146–47).⁷ The practice had become prominent enough by 1931 that an article in the American Bar Association Journal proposed that legislatures explicitly authorize courts to declare that new judge-made rules would operate only prospectively.

In 1932, the constitutionality of such “prospective overruling” was challenged in the United States Supreme Court. Great N. Ry. Co. v. Sunburst Oil & Refining Co., 287 U.S. 358 (1932). The Montana Supreme Court had previously held that when the Montana Railroad Commission reversed a determination of freight charges’ reasonableness, shippers could recover the excess amounts paid under that determination. In this case, the Montana court overruled that holding but applied the old rule to the parties before it and allowed the shipper to recover the unreasonable charges. The railroad argued that this deprived it of property without due process of law because it had been forced to refund the payments by virtue of an interpretation of the statute now acknowledged to be wrong. Great N. Ry. Co. (1932, pp. 359–61). In Great Northern Railway Co. v. Sunburst Oil & Refining Co., the Supreme Court rejected this argument in a unanimous decision written by now Justice Cardozo:

A state in defining the limits of adherence to precedent may make a choice for itself between the principle of forward operation and that of relation backward. It may say that decisions of its highest court, though later overruled, are law none the less for intermediate transactions. Indeed, there are cases intimating, too broadly, that it must give them that effect; but never has doubt been expressed that it may so treat them if it pleases, whenever injustice or hardship will thereby be averted.

Great N. Ry. Co. (1932, p. 364) (emphasis added and citations omitted). Justice Cardozo described the “declaratory” understanding of adjudication as merely one of several permissible approaches. A state court might:

hold to the ancient dogma that the law declared by its courts had a Platonic or ideal existence before the act of declaration, in which event the discredited declaration will be viewed as if it had never been …. The choice for any state may be determined by the juristic philosophy of the judges of her courts, their conceptions of law, its origin and nature …. [W]e may say of the earlier decision that it has not been overruled at all. It has been translated into a judgment of affirmance and recognized as law anew. Accompanying the recognition is a prophecy, which may or may not be realized in conduct, that transactions arising in the future will be governed by a different rule …. [W]e are not at liberty, for anything contained in the constitution of the United States, to thrust upon those courts a different conception either of the binding force of precedent or of the meaning of the judicial process

Great N. Ry. Co. (1932, pp. 365–66). This decision put to rest any constitutional concerns with state courts’ prospective judgments.

The balance of this chapter will examine how courts have responded to this possibility and attempt to summarize the state of the law. Like any attempt to describe American law in general, it will be complicated by the federal character of the jurisdiction. The law of prospectivity and retroactivity can be and very often is different from state to state. My summary account of state law, therefore, must be taken as more indicative than definitive. In addition, I will describe the development and current state of the subject in connection with federal law – the law of the United States. Although it will be apparent that the division is in some ways artificial, I will also divide the treatment between judgments of civil law and criminal law.

Prospective Judgments in Civil Law

State Law

Statements of law contained in a judgment are uniformly presumed to apply to events predating that judgment. The negative impact of such retroactive application of judicially created rules must be substantial before a court will consider limiting the rules to future cases. It follows that, at a minimum, a judicial decision must create a genuinely new rule of law before it may be even be a candidate for prospective-only application. Only then does a judgment “create[] an interregnum during which social relations have been conducted within institutional arrangements subsequently determined to be legally vulnerable” (Currier 1965, p. 240). As noted, the clearest case is when a court explicitly overrules a prior decision but it may also be present when a court formulates a rule for the first time.

When deciding whether to depart from the default of full retroactivity, the primary consideration is the nature and degree of the parties’ reasonable reliance on the prior state of the law. This consideration “can hardly be overemphasized.” Beavers v. Johnson Controls World Servs., Inc., 881 P. 2d 1376, 1384 (N.M. 1994). Courts view some fields of law – such as contract and property – as especially likely to induce such reliance. Apart from the injustice of erasing or devaluing rights deemed to have already “vested” in their holders (Traynor 1977, p. 544), these are fields where individuals may have actually paid attention to existing rules of law, perhaps even consulted legal advisers, before engaging in a given transaction (Traynor 1977, p. 544; Currier 1965, p. 242; Eisenberg 1988, p. 122).

By contrast, new rules of tort law seldom upset significant reliance interests. “Ordinarily,” for example, “persons who drive carelessly do not do so in conscious reliance upon some rule of law” (Fairchild 1967–1968, p. 261).⁸ The scope of tort liability may, however, affect some decisions on whether and how much insurance a party obtains, as well as that party’s decision to investigate an incident for which it might be held liable. Thus, courts often make decisions eliminating tort immunity for municipalities and charitable institutions prospective-only.⁹ It should be noted that in calculating the reliance that justifies making judicial decisions non-retroactive, courts almost always consider categories of cases; not the presence or absence of reliance by the particular parties before the court (Schaefer 1967, pp. 642–43).¹⁰

The serious costs that full retroactivity can impose have been emphasized in certain (federal) constitutional cases. Striking down a longstanding and deeply entrenched public program or institution as unconstitutional may dramatically upset the lives of many people. While seldom described in such terms, the United States Supreme Court’s judgment in Brown v. Board of Education ordering the dismantling of racially segregated schools but only “with all deliberate speed” might be understood as such a case. Brown v. Board of Education, 349 U.S. 294, 300–01 (1955). That decision made school systems in 20 states and the District of Columbia illegal. Immediately eliminating this illegality would have radically disrupted the lives of tens of thousands of students, parents, teachers, and employees. As it turned out, implementation of the Brown judgment went on for decades (Claus and Kay 2010, pp. 496–500; Currier 1965, pp. 231–32).

Likewise, when courts found that the composition of many state legislatures violated the Equal Protection Clause of the Fourteenth Amendment of the United States Constitution under the Supreme Court’s “one person-one vote” doctrine, a question arose as to the validity of laws passed by the mal-apportioned legislatures. Courts uniformly held that the mal-apportioned legislatures could make valid laws until a conforming legislature could be convened.¹¹ Similarly, the Supreme Court refused to invalidate completed actions of the Federal Elections Commission and the federal bankruptcy courts even though both had been found illegally constituted.¹²

In each case where prospective operation is suggested, there are necessarily competing considerations. Retroactive application might undermine reasonable actions taken in reliance on the former law. On the other hand, the very content of the judgment declares the new rule to be superior to the old one. Prospective-only operation, therefore, entails a decision to apply an inferior rule to prior transactions. In accommodating the relevant factors, many state courts have settled on some variation of a test formulated by the United States Supreme Court in its 1971 decision in Chevron Oil Co. v. Huson, 404 U.S. 97 (1971).¹³ The test considers three factors:

• 1. whether the decision to be applied non-retroactively establishes a new principle of law, either by overruling clear past precedent or by deciding an issue of first impression;

• 2. if, in light the new rule’s purpose and effect, retrospective operation would further or retard its operation; and

• 3. the extent of the inequity imposed by retroactive application, namely the injustice or hardship that would be caused by retroactive application.¹⁴

A court examines these factors, it must be stressed, against the background presumption that retroactivity is “overwhelmingly the norm.” James B. Beam Distilling Co. v. Georgia, 501 U.S. 529, 535 (1991) (opinion of Souter, J.). Thus, a litigant seeking prospective-only application must firmly convince a court that each factor favors such a decision.

Once we have established the possibility that a court may limit the retroactive application of its judgment, other questions arise. We might expect such a decision would mean that the new rule is to apply only to primary conduct occurring after the date the decision is announced and to no conduct occurring before that date. While that is sometimes the case, there are other possibilities. A court might make a new norm partly retroactive, applying it to some but not all prior events. For example, when the Connecticut Supreme Court expanded an enterprise’s “slip and fall” tort liability to include injuries caused by a foreseeably unsafe “mode of operation,” it applied its holding to “all future cases and, as a general rule, to all previously filed cases in which the trial has not yet commenced ….” Kelly v. Stop & Shop, Inc., 918 A.2d. 249, 265 n.9 (Conn. 2007).¹⁵ The court apparently concluded that the costs of adjusting to the new rule would not be excessive if litigation had not yet reached the trial stage.

The simplest approach of starting the rule running at the moment of decision has, for reasons which will become apparent, been labeled “pure prospectivity.” Neither the litigant in the case announcing the new rule nor any other person whose claim is based on prior events will be subject to the new rule. James B. Beam (1991, p. 536).¹⁶ Pure prospectivity makes clear the two distinct activities of an appellate court: articulating the law and deciding the case in controversy. Since the new rule plays no role in determining the outcome of the litigation, it is technically dicta and, as such, communicates only a prediction of what the law will be (Schaefer 1982, p. 22). It is at best, as Justice Cardozo recognized in the Sunburst decision, only a “prophecy, which may or may not be realized in conduct ….” Great N. Ry. Co. (1932, p. 366). A court might even postpone the moment the rule becomes applicable to some date further in the future. This variation is sometimes called “prospective-prospective overruling.” In these cases, a court may reason that parties affected by the new rule need additional time to adjust their behavior. So, when the Wisconsin Supreme Court abrogated the doctrine of governmental immunity from tort liability on June 5, 1962, it held the “effective date of the abolition of the rule” would be July 15, 1962 in order “[t]o enable the various public bodies to make financial arrangements to meet the new liability.” Holytz v. City of Milwaukee, 115 N.W.2d 618, 626 (Wis. 1962).¹⁷ When, later the same year, the Minnesota Supreme Court reached a similar conclusion, it expressed its “intention to overrule the doctrine of sovereign tort immunity as a defense with respect to tort claims … arising after the next Minnesota Legislature adjourns, subject to any statutes which now or hereafter limit or regulate the prosecution of such claims.” Spanel v. Mounds View Sch. Dist. No. 621, 118 N.W.2d 795, 803 (Minn. 1962). This both allowed institutions to buy liability insurance and gave the legislature a chance to craft an alternative liability regime that would accommodate the special interests of the public entities. Spanel (1962, pp. 803–04).¹⁸ The “alternatives of time and method are almost limitless.” (Rogers 1968, p. 57).

Judgments applying “pure prospectivity,” appear to be relatively infrequent. Much more commonly, a court applies the new rule to the litigants in the instant case but “then return[s] to the old one with respect to all other[] [cases] arising on facts predating the pronouncement.” James B. Beam (1991, p. 537). This course is sometimes called “selective prospectivity.” James B. Beam (1991, pp. 537–38). In part, this practice is motivated by a desire to connect a judgment’s statements of law to the particular controversy before the court (Eisenberg 1988, p 131; Moschzisker 1924, pp. 426–27). More prominent is the worry that not granting the benefit of the new rule to the party arguing for it in the case in which it is announced would discourage other litigants from advancing claims that would change existing law. It would, therefore, deprive the legal system of the law-reform benefits that derive from judicial consideration of those claims (Ghatan 2010, pp. 180–81).¹⁹ Critics have questioned this premise. The fact that courts maintain retroactive application in the great majority of cases is enough to motivate most litigants. Some parties, moreover, will have a continuing interest in the legal rule so that even if they fail to benefit in the first case, they will profit from its adoption in future ones (Rogers 1968, p. 49).

In addition to doubts about its incentive effect, critics of selective prospectivity emphasize the inevitable inequity that results (Eisenberg 1988, p. 129). The best known example of this defect is the 1959 decision of the Supreme Court of Illinois in Molitor v. Kaneland Community Unit. District 302, 163 N.E.2d 89 (Ill. 1959). The plaintiff was 1 of 14 school children suffering burns and other injuries when, due to the negligence of its driver, a school bus struck a culvert and exploded into flames. The Supreme Court used the case to reconsider and to abolish the tort immunity of school districts. Molitor (1959, pp. 89–98). It noted, however, that retrospective application of the decision would work a hardship on school districts that may have failed to secure adequate insurance or to investigate prior accidents on the assumption they could not be held responsible for them. It decided that the new liability would apply only in “cases arising out of future occurrences.” It made an exception, however, for “the plaintiff in the instant case.” It cited the two standard reasons for such an exception: if it failed to apply the new rule, the “announcement would amount to mere dictum”; and, “more important,” it would “deprive appellant of any benefit from his effort and expense” and eliminate any “incentive to appeal the upholding of precedent.” Molitor (1959, pp. 97–98).

The unattractive consequences of this solution became apparent when seven other children hurt in the same accident – including three of the first plaintiff’s siblings – sought relief. Molitor v. Kaneland Cmty Unit. Dist. 302, 182 N.E.2d 145, 146–47 (Ill. 1962). Originally all eight children had filed a single complaint. But only one child, randomly chosen, was named in the first appeal to the Supreme Court. The trial court, relying on the Supreme Court’s explicit exception for only “the plaintiff in the instant case,” dismissed the complaints. The Supreme Court reversed since it “now appears the [first] appeal was treated by the parties as a test case ….” Molitor (1962, pp. 145–46). The facts of this case highlight the arbitrary quality of selective prospectivity. The Court’s second decision eliminated the inequity for those involved in the same accident but left in place the different treatment accorded every other victim of municipal negligence who was injured before the date of the first decision.²⁰

Federal Courts

Despite these concerns, most state courts do maintain the option of non-retroactivity. The situation in federal courts is more complicated. After an initial period of infrequent and uncritical use of non-retroactivity, the United States Supreme Court systematized its approach in 1971 by articulating the three-factor Chevron Oil test, cited above, for deciding whether to apply a judgment non-retroactively. Then, in the early 1990s, the Supreme Court reversed course and held that federal courts must always apply their judgments retroactively. The following is a brief summary of that evolution.

A set of cases in the nineteenth century recognized – indeed, appeared to require – non-retroactive application of judge-made changes in state law insofar as that law was applied in federal court litigation founded on “diversity jurisdiction,” providing a federal forum where the parties to a controversy reside in different states. U.S. Const. art. III, § 2 (creating diversity jurisdiction). At that time, federal judges in diversity cases had developed and applied their own federal common law,²¹ but deferred to state courts’ interpretations of enacted state law, i.e. statutes and constitutions. In an 1847 diversity case appealed from the federal court in Mississippi, however, the Supreme Court decided that it should defer to state courts’ interpretations of enacted state law only prospectively. Rowan v. Runnels, 46 U.S. (5 How.) 134, 139 (1847). The Supreme Court had previously held, in the absence of any state court interpretation on the point, that a provision of the Mississippi constitution prohibiting the sale of slaves was ineffective without state legislation implementing it. After the contract at issue had been made, however, the Supreme Court of Mississippi held the provision self-executing. Rowan (1847, pp. 134–35). The United States Supreme Court agreed that federal courts should conform to state court interpretations “from the time they are made.” “But we ought not to give them a retroactive effect, and allow them to render invalid contracts entered into with citizens of other States, which in the judgment of this court were lawfully made.” To do so would render the independent federal diversity jurisdiction “utterly useless and nugatory.” Rowan (1847, p. 139). The dissenting opinion highlighted the anomaly of such a holding, arguing that it “gives to the Constitution of Mississippi different meanings at different periods of its existence ….” Rowan (1847, p. 140) (Daniel, J., dissenting). This approach was followed in several other federal diversity cases dealing with the validity of bonds issued by local governments under an authority that had first been confirmed by decisions of the state courts but subsequently denied under changed interpretations of state constitutions.²²

The underlying concern about the unfairness of retroactive decisions evident in these cases, as well as in the state court decisions already canvassed, also surfaced in connection with federal court judgments applying federal law. On three occasions in the 1960s, perhaps influenced by its decisions refusing to apply new rules of criminal procedure retroactively,²³ the Supreme Court refused to give its holdings in civil cases retroactive effect.²⁴ Only in the 1971 case of Chevron Oil Co. v. Huson, however, did the modern Supreme Court consider the issue of prospectivity in depth. The plaintiff had sustained a personal injury while at work on Chevron’s off-shore drilling platform. Recovery for such claims was governed by a federal statute, the Outer Continental Shelf Land Act, which specified no statute of limitations. Most courts that had addressed the issue had held that the limitations period was controlled by the equitable doctrine of laches. Chevron Oil (1971, pp. 98–99). Then, in a 1969 decision, after Huson had filed his complaint, the Supreme Court rejected those cases and interpreted the Act to borrow the neighboring state’s personal injury limitations period. Rodrigue v. Aetna Cas. & Sur. Co., 395 U.S. 352 (1969). For Huson, that was Louisiana and its 1 year statute of limitations now barred his claim. Chevron Oil (1971, p. 99).

But the Supreme Court held that its 1969 decision “should not be invoked to require application of the Louisiana time limitation retroactively to [Huson].” Chevron Oil (1971, p. 100). The Court went on to elaborate the three-factor test already mentioned: (i) the rule was genuinely new; (ii) retroactive application was not necessary to further the operation of that rule; and (iii) retroactivity “could produce substantial inequitable results.” Chevron Oil (1971, pp. 106–07). In this case, each factor favored prospective-only application. Chevron Oil (1971, p. 107). Eight Justices joined this opinion.²⁵ As already noted, the Chevron Oil test soon became the standard way of deciding prospectivity questions in state courts.²⁶

Three decisions in the early 1990s, however, reversed the adoption of the Chevron Oil test in federal courts. By this time, the Supreme Court, as will be discussed below, had retreated from the idea that it could limit the retroactive effect of decisions creating new constitutional rules of criminal procedure.²⁷ The three civil decisions each dealt with the question of whether taxpayers were entitled to a refund of state taxes paid under a statute later held unconstitutional. In the first, the Supreme Court held that taxpayers were not entitled to a full refund. Am. Trucking Ass’ns, Inc. v. Smith, 496 U.S. 167 (1990). Four justices applied the Chevron Oil test, observing that state authorities had reasonably supposed the taxes valid when imposed and that refunds “could deplete the state treasury [and] threaten[] the State’s current operation and future plans.” Am. Trucking (1990, p. 182) (opinion of O’Connor, J.). Four dissenting justices, however, objected to the very idea that courts could apply two different laws to identical controversies simply because they arose at different times. Am. Trucking (1990, pp. 205–06) (Stevens, J., dissenting). They read Chevron Oil narrowly, confining it to a court’s power to adjust “rules of tolling, laches, and waiver” for equitable reasons. Am. Trucking (1990, p. 221). Chevron Oil did not “alter the principle that consummated transactions are analyzed under the best current understanding of the law at the time of decision ….” Am. Trucking (1990, p. 222) (Stevens, J., dissenting). Ultimately, the Court denied the refunds because the ninth judge, Justice Scalia, believed that the state tax in question had been and continued to be constitutional. Am. Trucking (1990, pp. 204–05) (Scalia, J., concurring in the judgment). But he made clear in his concurrence that he agreed with the dissenters on prospectivity. “Since the Constitution does not change from year to year; since it does not conform to our decisions, but our decisions are supposed to conform to it; the notion that our interpretation of the Constitution in a particular decision could take prospective form does not make sense.” Am. Trucking (1990, p. 201) (Scalia, J., concurring in the judgment).

The writing was now on the wall. The next year, in the second unconstitutional state tax case, the Court issued five separate opinions, none with the support of more than three justices. But again, one could count five votes for the proposition that, when the Court decided a constitutional issue and applied it to the parties at bar, it must apply that holding to any other cases still open. James B. Beam (1991, pp. 543–44) (opinion of Souter, J.); James B. Beam (1991, pp. 548–49) (Scalia, J., concurring in the judgment).

Finally, in the third case, the Court produced a single majority opinion expressing the new understanding:

When this Court applies a rule of federal law to the parties before it, that rule is the controlling interpretation of federal law and must be given full retroactive effect in all cases still open on direct review and as to all events, regardless of whether such events predate or postdate our announcement of the rule …. [W]e now prohibit the erection of selective temporal barriers to the application of federal law in noncriminal cases.

Harper v. Va. Dep’t of Taxation, 509 U.S. 86, 97 (1993). Although it expressed a view of adjudication distinctly hostile to any form of non-retroactivity, the majority opinion actually forbade only “selective prospectivity,” in which a court refuses to apply a new rule to other cases concerning conduct predating the court’s judgment, but does apply it to the parties before it. It did not, therefore, overrule Chevron Oil, which was an instance of “pure prospectivity” in which the plaintiff had been given the benefit of the earlier limitations period.²⁸ The United States Court of Appeals for the Ninth Circuit recently held that, in the absence of an explicit holding that Chevron Oil had been overruled, it was still bound to apply new rules purely prospectively when the three Chevron Oil factors so required. Nunez-Reyes v. Holder, 646 F. 3d 684, 690–95 (9th Cir. 2011); (Fisch 1997, p. 1062).

The Supreme Court’s decisions retreating from prospective judgments show the influence of the factors already discussed that have worried courts and commentators about the practice. A central concern is a necessary departure from what was understood as the essential judicial role. This understanding reflected, at some level, the Blackstonian view of adjudication. This was most explicit in the separate opinions of Justice Scalia. The judge’s job, he asserted, “is to say what the law is, not to prescribe what it shall be …. [A prospective holding] presupposes a view of our decisions as creating the law, as opposed to declaring what the law already is.” Am. Trucking (1990, p. 201) (Scalia, J., concurring in the judgment); (James B. Beam 1991, p. 549 (Scalia, J., concurring)); (Harper 1993, pp. 106–07 (Scalia, J., concurring)).²⁹

Closely related was regard for the constitutional imperative that federal courts adjudicate only real “cases or controversies.” U.S. Const. art. III, § 2, cl. 1. Some have argued that this precludes a federal court from pronouncing on a legal issue unless it is necessary to resolve the dispute at bar. A purely prospective holding, the enunciation of a “‘prophecy, which may or may not be realized in conduct’ … would be beyond the authority of a federal court” (Yale Law Journal Note 1962, p. 932). This is doubtful as a matter of constitutional interpretation (Currier 1965, pp. 216–20). But the Supreme Court had appeared to accept the reasoning in an earlier criminal procedure case. Stovall v. Denno, 388 U.S. 293 (1967). It declined to apply a new rule retroactively though it conceded that it had applied it to the parties in the case first announcing it. In that first case, retroactive application was an “unavoidable consequence of the necessity that constitutional adjudications not stand as mere dictum [and of] [s]ound policies of decision-making, rooted in the command of Article III of the Constitution that we resolve issues solely in concrete cases or controversies ….” Stovall (1967, pp. 300–01) (Roosevelt 1999, pp. 1111–12).

In sum, if the limits of constitutional federal jurisdiction obliged a federal court to apply a new rule, at least to the party in the case announcing it, then the only kind of prospectivity available was “selective prospectivity.” The Supreme Court, however, became unwilling to accept the inequity of making the applicability of a rule turn on the arbitrary matter of which case happened to reach the Court first (Shannon 2003, p. 866). In an earlier case, Justice Harlan had protested the consequences of this policy:

Simply fishing one case from the stream of appellate review, using it as a vehicle for pronouncing new constitutional standards, and then permitting a stream of similar cases subsequently to flow by unaffected by that new rule constitute an indefensible departure from this model of judicial review.

Mackey v. United States, 401 U.S. 667, 679 (1971) (Harlan, J., concurring and dissenting). These arguments, which had already convinced a majority of the Court in the field of criminal procedure, ultimately led the Court to establish a policy of “full retroactivity” in the adjudication of federal civil cases. Harper (1993, p. 97).

It is important to recall that the development just traced is applicable only to changes in federal law. As already noted, state courts applying state law retain the option of prospective-only effect when declaring new rules. Such a practice, moreover, continues to be constitutionally permissible under the rule of Great Northern Railway Co. v. Sunburst Oil & Refining Co. (1932). These courts have generally rejected the reasoning of the United States Supreme Court’s post-Chevron Oil cases. Indeed, the Chevron Oil analysis remains the most common test in state jurisdictions for deciding whether to apply a new rule retroactively – notwithstanding its abandonment by the Court that created it.³⁰ When, however, state courts apply a new judge-made rule of federal law, the Supremacy Clause of the federal Constitution requires that they apply it retroactively in accordance with the holdings of the United States Supreme Court. Reynoldsville Casket Co. (1995, p, 754), Harper (1993, p. 100).

The Limits of Retroactivity

Whether in state or federal court, there are necessary limits to the ordinary retroactive application of judicial pronouncements. Although the United States Supreme Court has said that “a rule of federal law, once announced and applied … must be given full retroactive effect by all courts adjudicating federal law,” it restricts that command to “cases still open on direct review.” Harper (1993, pp. 96–97). No one suggests that a new rule requires courts to re-open and re-decide every case ever litigated to which a new rule might apply. A rule’s retroactivity does not extend to cases that have proceeded to:

such a degree of finality that the rights of the parties should be considered frozen …. [T]hat moment should be when the transaction is beyond challenge either because the statute of limitations has run or the rights of the parties have been fixed by litigation and have become res judicata.

United States v. Estate of Donnelly, 397 U.S. 286, 296 (1970) (Harlan, J., concurring).³¹

This limit is illustrated by Chicot County Drainage District v. Baxter State Bank, 308 U.S. 371 (1940). Bondholders whose rights had been reduced under a federal statute subsequently declared unconstitutional sought to recover the full amount originally due. The Supreme Court noted that a 1936 District Court proceeding – in which the validity of the governing law was not raised – had confirmed a prior adjustment and had never been appealed. Chicot Cnty. Drainage Dist. (1940, pp. 372–74). The law’s constitutionality was thus res judicata and could not be raised in a collateral proceeding. Chicot Cnty. Drainage Dist. (1940, p. 378). As the Court put it in a later case, “the res judicata consequences of a final, unappealed judgment on the merits [are not] altered by the fact that the judgment may have been wrong or rested on a legal principle subsequently overruled in another case.” Fed. Dep’t Stores, Inc. v. Moitie, 452 U.S. 394, 398 (1981).

The limits of retroactivity are grounded in strong practical policy. “A contrary rule,” as one state court noted, “would produce chaos in the legal system, as judgments could be continually opened and reopened with every fluctuation in the law.” Quantum Res. Mgm’t, L.L.C. v. Pirate Lake Oil Corp., 112 So. 3d 209, 217 (La. 2013), cert. denied sub nom. Haydel v. Zodiac Corp., 134 S. Ct. 197 (2013).³² A nineteenth century Supreme Court decision put the matter powerfully:

[T]he maintenance of public order, the repose of society, and the quiet of families, require that what has been definitely determined by competent tribunals shall be accepted as irrefragable legal truth. So deeply is this principle implanted in [our] jurisprudence, that commentators upon it have said, that res judicata renders white that which is black, and straight that which is crooked.

Jeter v. Hewitt, 63 U.S. (22 How.) 352, 364 (1859).³³ This concern for finality qualifies all of the United States Supreme Court decisions endorsing the general principle of retroactivity. Justice Souter conceded that “one might deem the distinction arbitrary, …. why should someone whose failure has otherwise become final not enjoy the next day’s new rule, from which victory would otherwise spring?” James B. Beam (1991, p. 541) (opinion of Souter, J.). Such equity, however, “could only be purchased at the expense of another principle. Public policy dictates that there be an end of litigation …. Finality must thus delimit equality in a temporal sense, and we must accept as a fact that the argument for uniformity loses force over time.” James B. Beam (1991, p. 542) (opinion of Souter, J., quoting Moitie (1981, p. 401) (internal quotation marks omitted)).

The policy of finality in civil litigation is not absolute. In exceptional cases, parties may collaterally attack otherwise final judgments – but only if the case is truly exceptional. Both the Restatement (Second) of Judgments and the Federal Rules of Civil Procedure articulate such a safety valve.

Section 73(2) of the Restatement states that a judgment “may be set aside or modified if … “[t]here has been such a substantial change in the circumstances that giving continued effect to the judgment is unjust.” Noting that this principle has sometimes been applied to cases where a later decision changes the law applied in an earlier (but unrelated) judgment, however, Comment (c) to this section labels such decisions “a misinterpretation of the rule and a very unsound policy.” Restatement of the Law, Second: Judgements 2d. (1980).

Rule 60(b) of the Federal Rules of Civil Procedure specifies five grounds for “relief from a final judgment,” none of which speak directly to a change in the governing law. Fed. R. Civ. P. 60(b)(1)-(5). A sixth merely refers to “any other reason that justifies relief.” Fed. R. Civ. P. 60(b)(6). The rare cases in which the Supreme Court considered applying Rule 60(b)(6) in connection with a change in governing law are inconclusive.³⁴ One judge has described the state of Rule 60(b)(6) jurisprudence as characterized by “a strong current of unwillingness to reopen judgments but with some wriggle room for future arguments.” Norgaard v. DePuy Orthopaedics, Inc., 121 F.3d 1074, 1078 (7th Cir. 1997) (Easterbrook, Cir. J.).³⁵ Still, other lower federal courts have referred to the rule as “a grand reservoir or [sic] equitable power,” Radack v. Norwegian Am. Line Agency, Inc., 318 F.2d 538, 542 (2d. Cir. 1963),³⁶ and have assumed that an “intervening change of controlling law” may justify exercising it. Marrero Pichardo v. Ashcroft, 374 F.3d 46, 56 (2d Cir. 2004) (quoting Virgin Atl. Airways, Ltd. v. Nat’l Mediation Bd., 956 F.2d 1245, 1255 (2d Cir. 1992)). In one case, a federal district judge allowed a plaintiff to re-open an unappealed judgment that had rejected his claim that state tuition grants to racially segregated private schools violated the Equal Protection Clause. Griffin v. State Bd. of Educ., 296 F. Supp. 1178 (E.D. Va. 1969). After the first judgment, the Supreme Court had pronounced a different and stricter test for evaluating such programs. Relying on Rule 60(b)(6), the District Court concluded that, in light of this “substantial change in the law … to continue [the first judgment’s] efficacy would be unjust to those initially and now affected by the order.” This was especially true when the litigation affected matters of “public import.” Griffin (1969, pp. 1180–82).³⁷

This kind of relief is exceptional. When courts note the possibility of modifying a final judgment, they always stress the need to show particularly compelling reasons. The United States Court of Appeals for the First Circuit summed up the prevailing attitude:

[T]he case law is very hostile to using a mistake of state law, still less a change in state common law, as grounds for a motion to reopen a final judgment under Rule 60(b)(6). Although the door is not quite closed, there is good sense—as well as much precedent—to make this the rarest of possibilities. Decisions constantly are being made by judges which, if reassessed in light of later precedent, might have been made differently; but a final judgment normally ends the quarrel. Indeed, the common law could not safely develop if the latest evolution in doctrine became the standard for measuring previously resolved claims. The finality of judgments protects against this kind of retroactive lawmaking.

Biggins v. Hazen Paper Co., 111 F.3d 205, 212 (1st Cir. 1997) (citations omitted) (paragraphs combined).³⁸

Notwithstanding the occasional exception, then, it is fair to say that the presumptive – and in federal courts nearly compulsory – retroactive effect of civil judgments reaches back only to controversies still open to judicial resolution. At some point adjudication comes to an end and unsuccessful civil litigants are denied the solace of newer and friendlier law.

In criminal cases, however, where a defendant remains in custody, the finality of a conviction is not so unqualified. The continuing possibility of collateral attack so long as a defendant remains in custody has been critical in shaping the law of the retroactivity and prospectivity of judicial decisions. This is the subject of the next section.

Prospective Judgments in Criminal Law

Substantive Liability

Discomfort with retroactive law has been most acute in connection with retroactive criminal liability. Hence the United States Constitution explicitly prohibits all ex post facto criminal laws.³⁹ The values underlying these worries are not entirely clear. The reliance interest, so prominent in civil prospectivity jurisprudence, may play a role if an actor is likely to consult the criminal law before acting. But criminal acts, like most tortious acts, are seldom the subject of self-conscious reliance on the law (Krent 1996, pp. 2160–63). The objection to ex post facto criminality seems premised on some more rudimentary sense of fairness (Traynor 1977, pp. 548–49).

As noted, the Supreme Court has held that the constitutional limitation on “ex post facto laws” refers only to legislation; not to judicial acts.⁴⁰ Concern about retroactivity is at its nadir when judicial action contracts the scope of criminal behavior. Consequently, when a criminal statute is held unconstitutional, even a final judgment of conviction is deemed void and may be subject to collateral attack⁴¹ (Traynor 1977, p. 553, n.54) (citing Ex parte Siebold, 100 U.S. 371, 376–77 (1879)). “[C]onventional notions of finality of litigation have no place where life or liberty is at stake and infringement of constitutional rights is alleged” Traynor (1977, p. 555) (quoting Sanders v. United States, 373 U.S. 1, 8 (1963)). More serious issues arise when courts interpret criminal law to criminalize acts that appeared lawful when committed. Such cases seem to raise problems identical to those underlying the ban on ex post facto legislation. As a result, courts have strained to find ways to apply these new interpretations only prospectively (Rogers 1968, p. 64).

In State v. Jones, the Supreme Court of New Mexico changed its construction of the state statute criminalizing lotteries. State v. Jones, 107 P.2d 324 (N.M. 1940). The defendants had previously been convicted under the statute for holding a “bank night” promotion at a movie theater but on appeal the Supreme Court held that such bank nights were not lotteries. Jones (1940, p. 325). When the same defendants were prosecuted a second time for the same offense, the Supreme Court rejected its former interpretation and held that bank nights were lotteries. Since, however, the defendants “did only that which this court declared, even if erroneously, to be within the law[,] …. [t]he plainest principles of justice demand that [the new interpretation] should [be given only prospective effect] ….” The court, therefore, quashed the information and declared that its new view of the statute would be observed only “in cases having their origin in acts and conduct occurring subsequent to the effective date of this decision.” Jones (1940, p. 329).⁴²

The Supreme Court of the United States dealt with a similar problem in James v. United States, a prosecution for tax evasion based on the defendant’s failure to report embezzled funds as income. James v. United States, 366 U.S. 213 (1961). An earlier case, Commissioner v. Wilcox, had held that embezzled funds were not income for these purposes. Commissioner v. Wilcox, 327 U.S. 404, 410 (1946). Three justices thought Wilcox continued to be good law and would have dismissed the prosecution on that basis. James (1961, p. 248) (Whittaker, J., concurring in part and dissenting in part). Three different justices would have overruled Wilcox and remanded for a new trial. James (1961, p. 241) (Clark, J., concurring in part and dissenting in part); James (1961, p. 241) (Harlan, J., concurring in part and dissenting in part). A third set of three justices would have overruled Wilcox and dismissed this case because the existence of Wilcox made it impossible to attribute to the defendant the “willfulness” necessary to sustain a conviction. James (1961, pp. 221–22 (opinion of Warren, C.J.). The net result of this division was that the Court overruled Wilcox but did not apply its new interpretation to the case before it or to any tax returns before the date of the decision. James (1961, p. 222). It should be noted, however, that six justices rejected the idea that the Court could apply its interpretation of the criminal law prospectively only. In his opinion, Justice Harlan claimed that “[o]nly in the most metaphorical sense has the law changed: the decisions of this Court have changed, and the decisions of a court interpreting the acts of a legislature have never been subject to the same [ex post facto] limitations which are imposed on legislatures ….” James (1961, p. 247) (Harlan, J., concurring in part and dissenting in part). In his separate opinion, Justice Black was even more direct:

We realize that there is a doctrine with wide support to the effect that under some circumstances courts should make their decisions as to what the law is apply only prospectively. Objections to such a judicial procedure, however, seem to us to have peculiar force in the field of criminal law. In the first place, a criminal statute that is so ambiguous in scope that an interpretation of it brings about totally unexpected results, thereby subjecting people to penalties and punishments for conduct which they could not know was criminal under existing law, raises serious questions of unconstitutional vagueness. Moreover, for a court to interpret a criminal statute in such a way as to make punishment for past conduct under it so unfair and unjust that the interpretation should be given only prospective application seems to us to be the creation of a judicial crime that Congress might not want to create.

James (1961, pp. 224–25) (Black, J., concurring in part and dissenting in part).

These cases demonstrate how judges try to avoid the retroactive application of a new, broader construction of a criminal statute. When it feels unable to apply such devices, the United States Supreme Court has adopted Justice Black’s suggestion that “subjecting people to penalties and punishments for conduct which they could not know was criminal under existing law” is a conviction based on an unknowable law, which deprives a defendant of “due process of law” and so violates the Fifth or Fourteenth Amendment. James (1961, p. 224). That was the Court’s holding in Bouie v. City of Columbia, 378 U.S. 347 (1964). The South Carolina Supreme Court had adopted a new and surprising interpretation of the state’s criminal trespass law to sustain the conviction of civil rights demonstrators conducting a “sit-in” at a segregated lunch counter. Bouie (1964, pp. 349–50). The United States Supreme Court held that this “unforeseeable and retroactive judicial expansion of narrow and precise statutory language” was a “deprivation of the right of fair warning.” Such an interpretation, if “applied retroactively, operates precisely like an ex post facto law …. If a state legislature is barred by the Ex Post Facto Clause from passing such a law, it must follow that a State Supreme Court is barred by the Due Process Clause from achieving precisely the same result by judicial construction.”⁴³ Bouie (1964, pp. 352–54) (citing Hall 1960, p. 61).

The Supreme Court qualified this statement in Rogers v. Tennessee, 532 U.S. 451 (2001). Tennessee courts had previously construed the state murder statute to incorporate the common law requirement that a victim die within a year and a day of the defendant’s act. In this case, the Tennessee Supreme Court abolished that requirement and upheld the challenged murder conviction. Rogers (2001, p. 455). In affirming, the United States Supreme Court denied that identical limits constrained legislation and judicial interpretation. Unlike legislatures, courts must routinely clarify and reinterpret prior holdings. This is especially true for the application of common law rules that “presuppose[] a measure of evolution that is incompatible with stringent application of ex post facto principles.” Due process prohibits only new interpretations that are “unexpected and indefensible by reference to the law which had been expressed prior to the conduct in issue.”⁴⁴ Rogers (2001, pp. 461–62). In this case, the “year and a day rule” had never been successfully invoked in Tennessee and it had been in retreat in other American jurisdictions for decades. Rogers (2001, p. 455). Thus, its abolition was not so “unexpected and indefensible” as to deprive the defendant of due process of law. As long as a defendant had “fair warning” that this kind of change might occur, a court was entitled to give it retroactive effect. Rogers (2001, pp. 462–64).

New Constitutional Rules of Criminal Procedure

While American courts generally apply reductions in criminal liability retroactively and enlargements of that liability prospectively, the rules governing the application of changes in criminal procedure are much more complicated.

In the 1960s, the United States Supreme Court decided a number of cases dramatically enlarging the rights of criminal defendants, including the right to counsel,⁴⁵ to remain silent,⁴⁶ to fair procedures in identification by witnesses,⁴⁷ and to exclude improperly secured evidence.⁴⁸ The cumulative effect has more than once been described as a “revolution” (Kamisar 1995, p. 1).⁴⁹ Judged under these new constitutional standards, many prior convictions would be invalid. During the same period, the Supreme Court also expanded the opportunities to attack a constitutionally defective state court conviction in federal court through a petition for a writ of habeas corpus.⁵⁰

The sum of these developments was ominous for a regime of thorough retroactivity. Thousands of incarcerated people were now in a position to reopen and possibly reverse their criminal convictions. One solution was to hold the new rules of criminal procedure at least partly non-retroactive. The issue came before the Supreme Court in 1965 in Linkletter v. Walker, 381 U.S. 618 (1965). Three years before, in Mapp v. Ohio, the Supreme Court had held that the Fourth Amendment made improperly seized evidence inadmissible in criminal prosecutions in state courts. Mapp (1961, p. 660). Linkletter was convicted of burglary in Louisiana in 1959 based, in part, on evidence subsequently held to be illegal. The state Supreme Court affirmed his conviction in 1960. After the decision in Mapp, Linkletter petitioned for a writ of habeas corpus. Linkletter (1965, p. 621). The United States Supreme Court held that the rule of Mapp was not retroactive. Linkletter (1965, p. 640). The Court cited some of the civil cases discussed above for the proposition that “the Constitution neither prohibits nor requires retrospective effect.” “[W]e must … weigh the merits and demerits in each case by looking to the prior history of the rule in question, its purpose and effect, and whether retrospective operation will further or retard its operation.” Linkletter (1965, pp. 629). Here, since Mapp was intended as a “deterrent to lawless police action” its purpose would not be “advanced by making the rule retrospective.” Past police misconduct could not be undone “by releasing the prisoners involved.” Moreover, states’ reliance on prior law was due the same respect that private reliance was given when new rules changed civil liability.⁵¹ And given the number of potential petitioners, retroactive application “would tax the administration of justice to the utmost.” Linkletter (1965, pp. 636–37).

The Court later rephrased the proper approach in a formula:

The criteria guiding resolution of the question implicates [sic] (a) the purpose to be served by the new standards, (b) the extent of the reliance by law enforcement authorities on the old standards, and (c) the effect on the administration of justice of a retroactive application of the new standards.

Stovall v. Denno, 388 U.S. 293, 297 (1967).⁵² Linkletter itself denied retroactive operation only to cases that were already final when the new rule was announced. Linkletter (1965, p. 622).⁵³ In subsequent cases, however, the Court took a broader view of its authority to limit the retroactive effect of new constitutional rules. In Stovall v. Denno, the Court held that a new rule excluding evidence of a police-arranged eyewitness identification if counsel were absent would “affect only those cases and all future cases which involve confrontations … conducted” after the new rule was announced. Stovall (1967, p. 296). Thus, it would apply neither to finally decided cases nor to some cases still open on direct review:

[N]o distinction is justified between convictions now final, as in the instant case, and convictions at various stages of trial and direct review. We regard the factors of reliance and burden on the administration of justice as entitled to such overriding significance as to make that distinction unsupportable.

Stovall (1967, p. 300).

The Court set various effective dates for the applicability of new rules of criminal procedure. A rule excluding a defendant’s statements made without adequate warnings about the right to counsel would apply only to cases in which the trial began after the new rule was announced. Johnson v. New Jersey, 384 U.S. 719, 721 (1966) (concerning retroactivity of Escobedo v. Illinois, 378 U.S. 478 (1966), and Miranda v. Arizona, 384 U.S. 436 (1966)). The rule that electronic surveillance was a “search” or “seizure” subject to the Fourth Amendment would apply “only to cases in which … .[the] electronic surveillance [was] conducted after” the judgment pronouncing the new rule.⁵⁴ Desist v. United States, 394 U.S. 244, 254 (1969) (concerning retroactivity of Katz v. United States, 389 U.S. 347 (1967)). New rules on permissible searches incident to arrest would apply only to searches occurring after the promulgation of those rules. Williams v. United States, 401 U.S. 646, 656 (1971) (plurality opinion) (concerning retroactivity of Chimel v. California, 395 U.S. 752 (1969)).⁵⁵

The era of partial retroactivity, however, was short-lived. Starting in the late 1960s, Justice Harlan, who had joined some of the early opinions, issued a series of powerful dissents to the Court’s decisions. He objected to a perceived departure from the Court’s judicial role. He was especially offended by the Court’s record of what we have called “selective prospectivity.” Typically, the Court decided the constitutional question in one case and applied its new rule – necessarily retroactively – to the parties at bar. It decided the more general retroactivity question in a later case. This resulted in an intolerable inequity.⁵⁶ Justice Harlan thus advocated the full retroactivity of constitutional judgments. But by this he meant only their application to cases that not yet final in the sense that the defendants had exhausted all available appeals. This meant that, with limited exceptions, a petition for a writ of habeas corpus could not be based on a new rule of criminal procedure. Mackey v. United States, 401 U.S. 667, 690 (1971) (Harlan, J., concurring and dissenting).

A majority of the Court adopted these arguments after Justice Harlan had left the bench. The 1982 case of United States v. Johnson, 457 U.S. 537 (1982),⁵⁷ concerned the retroactivity of a 1980 case⁵⁸ holding that the Fourth Amendment prohibits warrantless entry into a residence to make a routine arrest. The Court held that its decisions should usually be “applied retroactively to all convictions that were not yet final at the time the decision was rendered.” Johnson (1966, p. 573).⁵⁹ In a significant limitation, however, the Court permitted non-retroactive application of new rules that were “clear break[s] with the past.” Johnson (1966, p. 558). In these cases, “prospectivity [was] arguably the proper course.” Johnson (1966, p. 589) (quoting Williams v. United States, 401 U.S. 646, 659 (1971)). In 1987, however, in Griffith v. Kentucky, 479 U.S. 314 (1987), the Court abandoned this “clear break” exception when it held that a new rule⁶⁰ limiting a prosecutor’s ability to use peremptory challenges based on a potential juror’s race would apply retroactively notwithstanding its admitted novelty. The Court thought the reasons for adhering to a rule of retroactivity were no less applicable to “clear break” cases than to less drastic changes in criminal procedure. Current federal law on the retroactivity of new decisions on questions of criminal procedure has thus reverted to the rule of Linkletter v. Walker: “[A] new rule for the conduct of criminal prosecutions is to be applied retroactively to all cases, state or federal, pending on direct review or not yet final ….” Griffith (1987, pp. 326–28).⁶¹

The key moment for cutting off the retroactive effect of judgments announcing new constitutional rules of criminal procedure, therefore, is when state court convictions have become “final” – i.e., when there is no further opportunity for direct appellate review either in the state courts or by writ of certiorari to the United States Supreme Court. As in civil litigation, this point has been defended, in part, by the need to bring proceedings to some identifiable close. There must, Justice Harlan had argued, “be a visible end to the litigable aspect of the criminal process …. If law, criminal or otherwise, is worth having and enforcing, it must at some time provide a definitive answer to the question litigants present or else it never provides an answer at all.” Mackey (1971, pp. 690–91) (Harlan, J., concurring and dissenting).

Res judicata, however, did not shield a final criminal judgment from any challenge as it would a final civil judgment. “Because of habeas corpus and similar writs,…. [a] criminal judgment of conviction does not enjoy the same degree of finality until the defendant has been executed, died in prison, or been released.” (Currier 1965, pp. 258–59). A habeas petition technically initiates a new civil action “for the enforcement of the right to personal liberty ….” Fay (1963, p. 423).⁶² Expressly limiting retroactive application to cases “on direct review” was therefore essential to avoid potentially re-opening the conviction of every defendant still in custody. And because the exclusionary rule at issue in Linkletter was, on the Court’s reasoning, unrelated to the merits of the prosecution, applying it retroactively would result in the “wholesale release of the guilty victims.” Linkletter (1965, p. 637).⁶³ If it did nothing else, limiting the retroactive effect of new criminal procedure rules to cases on direct review imposed a quantitative limit on the resulting disruption (Fallon and Meltzer 1991, p. 1815).⁶⁴

Nonetheless, refusing to apply a new rule of criminal procedure to a class of defendants incarcerated as a result of trials in which those new rules were not observed necessarily involved an arbitrary element. The Court’s approach in Linkletter kept “all people in jail who were unfortunate enough to have had their unconstitutional convictions affirmed before June 19, 1961.” Linkletter (1965, p. 641) (Black, J., dissenting). And as the dissent was quick to point out, Linkletter had committed his offense before the defendant in Mapp, who had been released under that case’s new exclusionary rule. Linkletter (1965, pp. 641–42) (Black, J., dissenting). If the courts had not delayed in resolving the Linkletter appeal, then he would have reached the Supreme Court first and been released under the new exclusionary rule. “Too many irrelevant considerations,” noted one commentator, “including the common cold, bear upon the rate of progress of a case through the judicial system” Schaefer (1967, p. 645).⁶⁵

Limiting the retroactive effect of new criminal procedure rules to cases on direct – not habeas – review has been defended not so much as an appropriate limit on retroactivity but rather as a necessary aspect of the restricted purpose of habeas corpus in federal courts. On this account, habeas did not exist to correct errors but to ensure that state courts adhered to the applicable federal standards of criminal justice. For this purpose, it generally sufficed that criminal prosecutions conformed to the law in effect at the time of the trial. Mackey (1971, pp. 691–92 (Harlan, J., concurring and dissenting); Roosevelt 1999, pp. 1093–94).

The Supreme Court crystallized its new approach in 1989 in Teague v. Lane, 489 U.S. 288 (1989). The plurality opinion in that case restated the background rule of full retroactivity of judgments on direct appeal expounded the previous term in Griffith v. Kentucky. Teague (1989, p. 304) (citing Griffith (1987, p. 314)). It then elaborated the limits of habeas corpus, including the inadmissibility of relying on rules not yet formulated at the time of the conviction in question. Teague (1989, pp. 308–09) (opinion of O’Connor, J.). “The relevant frame of reference,” it emphasized, “is not the purpose of the new rule whose benefit the [defendant] seeks, but instead the purposes for which the writ of habeas corpus is made available.” Teague (1989, p. 306) (quoting Mackey (1971, p. 682) (Harlan, J., concurring in part and dissenting in part)). The Court then established what a leading treatise refers to as a “quite differently structured doctrine of non-retroactivity”: A rule declared only after a conviction was final could not be relied on in a subsequent collateral proceeding, subject to two narrow exceptions (LaFave et al. 2000, p. 866). A dissenting Justice summarized them accurately:

Any time a federal habeas petitioner’s claim, if successful, would result in the announcement of a new rule of law, … it may only be adjudicated if that rule would [1] plac[e] certain kinds of primary, private individual conduct beyond the power of the criminal law-making authority to proscribe, or [2] if it would mandate new procedures without which the likelihood of an accurate conviction is seriously diminished.

Teague (1989, p. 330) (Brennan, J., dissenting) (citations omitted) (internal quotation marks omitted). Although some parts of this analysis commanded the assent of only four justices in Teague, a majority of the Court approved it the following year. Penry v. Lynaugh, 492 U.S. 302 (1989). Teague’s rules are now treated “as the settled guidelines for determining what law applies on habeas review” (LaFave et al. 2000, p. 880).

Teague’s restriction on the use of “new rules” on collateral review has been interpreted very broadly. A new rule need not be the kind of “clean break” briefly significant under United States v. Johnson.⁶⁶ According to the Teague plurality, “a case announces a new rule if the result was not dictated by precedent existing at the time the defendant’s conviction became final.” Teague (1989, p. 301). Consequently, even though a judgment is carefully and plausibly explained as an application of existing law, it may still be a “new rule.” It qualifies, according to a later decision, so long as its outcome “was susceptible to debate among reasonable minds.” Butler v. McKellar, 494 U.S. 407, 415 (1990).⁶⁷ “[A]ny reading beyond the narrowest reasonable reading of [applicable] precedent … can readily be viewed as a ‘new rule’” (LaFave et al. 2000, p. 882).

This understanding of eligible new rules is reinforced by the Court’s parsimonious reading of Teague’s two exceptions. The first concerned new rules that placed “certain kinds of primary, private individual conduct beyond the power of the criminal law-making authority to proscribe.” Teague (1989, p. 311) (quoting Mackey (1971, p. 692) (Harlan, J., concurring in part and dissenting in part)). This falls within the well-established doctrine that decisions narrowing the scope of criminal liability should be fully retroactive.⁶⁸ Therefore, a court recently held that it was proper to reconsider a final unappealed conviction for gun possession after the Supreme Court found the relevant law unconstitutional. Magnus v. United States, 11 A.3d 237, 243–46 (D.C. 2011).⁶⁹ This exception also allows collateral review when a court reinterprets a criminal statute to exclude a petitioner’s conduct. Bousley v. United States, 523 U.S. 614, 620–21 (1998).

The second exception permits a habeas court to review an otherwise final conviction if it were obtained in violation of a later-formulated “watershed rule[] of criminal procedure,” non-observance of which would result in “the likelihood of an accurate conviction [being] seriously diminished.” Teague (1989, p. 313).

In order to qualify as watershed, a new rule must meet two requirements. First, the rule must be necessary to prevent an impermissibly large risk of an inaccurate conviction. Second, the rule must alter our understanding of the bedrock procedural elements essential to the fairness of a proceeding.

Whorton v. Bockting, 549 U.S. 406, 418 (2007) (citations omitted) (internal quotation marks omitted). This exception has turned out to be extremely limited in practice. The Court had already recognized that the great bulk of the new procedures mandated in the rights revolution of the 1960s did not measurably enhance the truth-finding aspects of a criminal prosecution. They were aimed, rather, at deterring unconstitutional police misconduct. Linkletter (1965, p. 637). The Supreme Court has identified only one case whose rule would satisfy this criterion – Gideon v. Wainwright, which mandated legal representation at public expense for indigent defendants. Whorton (2007, p. 419). By contrast, the Court has declined to allow a death-row prisoner to challenge his execution based on a Supreme Court judgment – announced after his sentence had been pronounced – requiring that the aggravating factor essential to impose the death penalty be determined by the jury and not the judge. Schiro v. Summerlin, 542 U.S. 348, 352–56 (2004).

In 1996, Congress enacted the Antiterrorism and Effective Death Penalty Act, Pub. L. No. 104–132, 110 Stat. 1214 (codified in scattered sections of 28 U.S.C.), which restricted the right of state court defendants to challenge their convictions by collateral review in federal court, creating, among other limitations, strict time limits. 28 U.S.C. § 2244(d) (2012). It also specified that if a particular claim had been “adjudicated on the merits in state court proceedings,” a federal court could not issue a writ of habeas corpus unless the state adjudication either: (i) “resulted in a decision that was contrary to, or involved an unreasonable application of, clearly established Federal law, as determined by the Supreme Court of the United States”; or (ii) “resulted in a decision that was based on an unreasonable determination of the facts in light of the evidence presented in the State court proceeding.” 28 U.S.C. § 2254(d)(1)-(2) (2012). This requirement roughly mirrors the plurality approach in Teague. Notably, the “clearly established federal law” to which the state court decision must conform refers to law at the time of that state court decision; not to subsequently declared changes in the required procedures. Lockyer v. Andrade, 538 U.S. 63, 71 (2003).

The Act and the Court’s doctrine are not precisely congruent, however. For example, the “new rule” that bars habeas relief under Teague is one announced after the petitioner’s case became final. Teague (1989, p. 310). But the “clearly established Federal law” to which a state court decision must conform to bar relief under the statute is that existing at the time of the relevant decision, even if the law is changed before the conviction becomes final – so that it might have been properly applied in reviewing the decision under Griffith and Teague. Greene v. Fisher, 132 S. Ct. 38, 44–45 (2011). Both sets of limitations – of Teague and of the Act – must be overcome before a federal district court may grant a habeas petition. Horn v. Banks, 536 U.S. 266, 271 (2002) (per curiam) (“[I]n addition to performing any analysis required by AEDPA, a federal court considering a habeas petition must conduct a threshold Teague analysis when the issue is properly raised by the state.”).⁷⁰

The foregoing discussion concerns only the limits of retroactive application of new constitutional rules of criminal procedure in collateral review of state criminal convictions in federal courts. State law also typically allows collateral attacks on convictions even after direct review is no longer available. State courts are free to apply new rules of criminal procedure on such review, even if a federal court could not. Danforth v. Minnesota, 552 U.S. 264, 295 (2008). State courts, in fact, apply a range of approaches when deciding whether to apply such law. Many have adopted the federal approach articulated in Teague.⁷¹ Other jurisdictions have kept the three factor test adopted – and now rejected – by the United States Supreme Court in Linkletter v. Walker and Stovall v. Denno.⁷²

The Problems of Prospective Adjudication

As this summary indicates, the history – and indeed, the current status – of the once widely accepted idea that judicial pronouncements of law are thoroughly retroactive has been more than a little complicated. Currently, for civil cases, most state courts examine the relevant factors favoring or disfavoring prospective application on a case-by-case basis. Federal courts must apply any new rule of law to all cases still pending on direct review when the rule is declared. Almost all courts will find a way to apply a judicially-narrowed rule of criminal liability retroactively but will refuse to do the same when the scope of liability has been broadened. The United States Supreme Court, after a period when it made some new rules of criminal procedure selectively prospective, has now settled on a “firm rule of retroactivity,” Landgraf (1994, p. 278 n.32), binding all federal courts. This retroactivity, however, reaches back only to cases in which direct appeals remain available. Reconsideration of a conviction in light of a subsequently announced procedural rule by way of collateral attack is permitted only within the narrow exceptions defined in Teague v. Lane and the AEDPA. ⁷³ State courts collaterally reviewing a judgment are not bound by Teague or the AEDPA and apply a variety of approaches. These divergent standards show that the retroactivity and prospectivity of judicially created law remains profoundly controversial in American jurisprudence.

The disagreement has often been expressed in terms of the practice’s relation to the doctrine of stare decisis. The declaration of a genuinely new rule is, by definition, a break with the discipline of stare decisis. ⁷⁴ Still, its advocates have argued that prospective overruling is supported by that doctrine’s core purposes:

At its core, stare decisis allows those affected by the law to order their affairs without fear that the established law upon which they rely will suddenly be pulled out from under them. A decision not to apply a new rule retroactively is based on principles of stare decisis. By not applying a law-changing decision retroactively, a court respects the settled expectations that have built up around the old law.

James B. Beam (1991, pp. 551–52) (O’Connor, J., dissenting).

The practice, therefore, “ensures realization of the goals underlying the principle of stare decisis without inhibiting the overturning of socially incongruent doctrine because of a concern for these values” (Auerbach 1991, p. 567).⁷⁵ This argument, however, turns out to be two-edged sword. The very capacity of a prospective ruling to minimize the upsetting of justified reliance may remove one of the greatest incentives to adhere to precedent:

By announcing new rules prospectively or by applying them selectively, a court may dodge the stare decisis bullet by avoiding the disruption of settled expectations that otherwise prevents us from disturbing our settled precedents. Because it forces us to consider the disruption that our new decisional rules cause, retroactivity combines with stare decisis to prevent us from altering the law each time the opportunity presents itself.

James B. Beam (1991, p. 548) (opinion of Blackmun, J.).⁷⁶

By reducing the costs of overruling, non-retroactivity may have contributed to an increasing tendency to overrule controlling precedents and, therefore, on balance, to increase instability in the law, thus undermining the principal goal of stare decisis (Wistrich 2012, pp. 765–70).⁷⁷ “Prospective decisionmaking,” according to Justice Scalia, “is the handmaid of judicial activism, and the born enemy of stare decisis.” Harper (1993, p. 105) (Scalia, J., concurring). By reducing the jolts caused by judicial actions, prospective overruling may encourage and legitimate judicial legislation, an enterprise basically inconsistent with the limited role of courts as interpreters and appliers of pre-existing law.⁷⁸

In one nineteenth-century judgment the United States Supreme Court declared that a new interpretation of a statute “is to all intents and purposes the same in its effect on contracts as an amendment of the law by means of a legislative enactment.” Douglass (1879, p. 687). This has been a criticism of non-retroactive judgments as long as they have been rendered in American courts.⁷⁹ It was also a prominent theme in the Supreme Court’s debate on the practice in the latter part of the twentieth century. Justice Harlan accused those willing to apply a new rule of criminal procedure only prospectively of feeling “free to act, in effect, like a legislature, making its new constitutional rules wholly or partially retroactive or only prospective as it deems wise.” Mackey (1971, p, 677) (Harlan, J., concurring and dissenting).

This disapproval, of course, assumes that it is improper for courts to make law. It depends, that is, at some level, on the ancient declaratory theory of adjudication. Indeed, in a concurring opinion endorsing full retroactivity, Justice Scalia quoted extensively from Blackstone. Harper (1993, pp. 106–07) (Scalia, J., concurring).⁸⁰ But this was a theory which the advocates of limited retroactivity had already rejected, indeed ridiculed for decades.⁸¹ In some versions, the advocacy of non-retroactive judgments was part and parcel of the American legal realist critique of formalist jurisprudence. One commentator was pleased at the prospect that “the more courts begin to utilize prospective overruling the more it will become obvious that the judge is, in fact, inescapably a judicial legislator” (Levy 1960, p. 16).⁸² And in an unusually candid judicial recognition of the realist position, Justice O’Connor defended non-retroactive judgments by citing Marbury v. Madison: “[P]recisely because this Court has ‘the power to ‘say what the law is,” when the Court changes its mind, the law changes with it.” James B. Beam (1991, p. 550) (O’Connor, J., dissenting) (citations omitted) (quoting Marbury v. Madison, 5 U.S. (1 Cranch) 137 (1803).⁸³ By the late twentieth century, this was a position with which it was difficult to argue and it undermined the claim, prominently advanced by Justice Harlan, that non-retroactivity was somehow inconsistent with the nature of adjudication. Mackey (1971, pp. 677–81) (Harlan, J., concurring and dissenting).

To the extent that the practice of giving judgments only prospective effect reflects modern recognition of the law-making power of judges, however, we might expect it to be employed differently depending on the underlying source of the law being applied. The demise of the declaratory theory led first to the conclusion that the judicial creation and modification of rules of common law were inevitably exercises of judicial legislation. It was with respect to the common law that Holmes recognized that “judges do and must legislate, but they can do so only interstitially ….” S. Pac. Co. v. Jensen, 244 U.S. 205, 221 (1917).⁸⁴ The idea that enacted law does not pre-exist judicial cases invoking it, however, is markedly harder to sustain. It might follow that courts should limit the applicability in time of their common law judgments but not those bottomed on statutes or constitutions (Roosevelt 1999, pp. 1076, 1107).

A few cases support this intuition. As recently as 2010, the Massachusetts Supreme Judicial Court held that ‘[w]here a decision does not announce new common-law rules or rights but rather construes a statute, no analysis of retroactive or prospective effect is required because at issue is the meaning of the statute since its enactment.” In re McIntire, 936 N.E.2d 424, 428 (Mass 2010).⁸⁵ For the most part, however, neither courts nor commentators have suggested that the source of the law at issue was much of a consideration with respect to the temporal effect of a judgment. Cardozo thought there was no “adequate distinction” between changes of rulings concerning statutes or common law. The choice of whether to apply a new rule prospectively was not governed “by metaphysical conceptions of the nature of judge-made law, nor by … the division of governmental powers, but by considerations of convenience, of utility, and of the deepest sentiments of justice” (Cardozo 1921, pp. 148–49). True to that position, when he wrote the Supreme Court’s Sunburst opinion, upholding the constitutionality of prospective overruling by state courts, he noted that the “alternative is the same whether the subject of the new decision is common law or statute.” Great N. Ry. Co. (1932, p. 365) (citations omitted).⁸⁶

In fact, some observers noted that prospective rulings have been more common in the case of new statutory interpretations than in the case of new common law rules (Hart and Sacks 1994, p. 604; Durgala 1962–63, pp. 54–56). The justification for such a priority has never been thoroughly explained. In Sunburst, Justice Cardozo assumed that the decision to apply judgments retroactively or prospectively – in whatever kind of case – was an aspect of the doctrine of stare decisis and that doctrine was itself a part of the common law and, therefore, within the authority of the judges. Great N. Ry. Co. (1932, p. 366). One writer has suggested that individuals are more likely to rely on statutory or constitutional rights than common law rights and are therefore entitled to a greater degree of protection (Rogers 1968, p. 54).

A recent decision of the United States Court of Appeals for the Sixth Circuit addressed the “seemingly compelling” argument that a state court was without power to treat a statutory interpretation as anything but fully retroactive. Volpe v. Trim, 708 F.3d 688, 702 (6th Cir. 2013). The court noted that legislatures often write broad statutes and rely on the courts to refine and apply them. Volpe (2013, p. 702). It was thus proper for a court to make “readjustments” in light of experience. Volpe (2013, p. 702). In these circumstances, “the judicial development of the legislatively-created concept is little different from the development of judicially-announced law” and, therefore, a court could properly consider whether its interpretation should apply retroactively. Volpe (2013, p. 702).

In sum, the current confused state of the law on the possibility of limited retroactivity of judgments demonstrates a persistent and possibly irresoluble tension in the American view of law and of the roles of legal institutions. The separation of powers is a fundamental dogma of the constitutional system. But it assumes that we are able to identify with some precision what distinguishes “legislative” from “judicial” functions (Currier 1965, pp. 221–22). The declaratory view of adjudication fits comfortably with that assumption. But the idea that the content of the law exists prior to and independent of its application by the courts seems to have been irretrievably lost. Even in the case of enacted written law, modern notions of interpretation have blurred the line between legislation and adjudication (Kay 2007, pp. 243–49). In these circumstances, what can it mean to complain that a prospective-only ruling is inconsistent with the judicial role? The difficulty is illustrated by a passage from one of Justice Scalia’s separate opinions in the Supreme Court’s series of cases effecting the transition from employment of limited retroactivity to a “firm rule of retroactivity.” Landgraf (1994, p. 278, n. 32):

I am not so naive (nor do I think our forebears were) as to be unaware that judges in a real sense “make” law. But they make it as judges make it, which is to say as though they were “finding” it—discerning what the law is, rather than decreeing what it is today changed to, or what it will tomorrow be.

James B. Beam (1991, p. 549) (Scalia, J., concurring). In his separate opinion in the same case, Justice White pounced on this obscure description of the proper role of courts:

[E]ven though the Justice is not naive enough (nor does he think the Framers were naive enough) to be unaware that judges in a real sense “make” law, he suggests that judges (in an unreal sense, I suppose) should never concede that they do and must claim that they do no more than discover it, hence suggesting that there are citizens who are naive enough to believe them.

James B. Beam (1991, p. 546) (White, J., concurring). There is no more contested issue in American constitutional law than the propriety of independent policy-making by courts (Kay 2007). Prospective judgments dramatically spotlight that controversy and it is not surprising that it has been a difficult and contentious issue for courts and commentators alike. It will be impossible, however, to arrive at a coherent and generally accepted approach to the retroactive or prospective application of new judicial declarations of law until there is an equally well-accepted definition of the proper allocation of lawmaking authority.