Unlike the US where the question of temporal effects of judicial rulings was considered early on, other major jurisdictions in the world, essentially from civil law tradition, addressed this issue much later. The prevalent narrative in most civil law jurisdictions has always been that, unlike parliamentary legislation, judicial decisions are not proper sources of law and therefore do not create legal rules. Since the power to make substantive law is vested exclusively in the legislature, civilian courts cannot make law but are bound to decide cases according to the best understanding of the law established by legislation and custom. This sharp distinction operated between courts’ decisions and legislative enactments has always carried with it the consequence that, whereas new legislation does not operate retrospectively, new judicial rulings are essentially retroactive. Furthermore, in civil law systems, where there is no doctrine of stare decisis and precedents are not formally binding, it is more difficult to know when a change has taken place since jurisprudence arises out of an accumulation or repetition of decisions in the same direction. Therefore, the precise moment when a judicial rule or interpretation has been modified is often difficult to determine. Overruling decisions are generally easier to identify in common law systems where judicial rulings are given official status through the operation of the doctrine of stare decisis; in such circumstances a single judgment is sufficient enough to give rise to a ruling with binding effect for the future.⁷ Having said that, even in common law systems where precedents are considered to be proper sources of law, the declaratory theory derived from Blackstone’s famous dictum that judges do not create law but merely discover it had the effect to hamper the reflection about the temporal effect of judicial decisions.⁸ And, even though the traditional declaratory approach has not remained unchallenged in modern time, there is still a deep seated belief that courts have only the power to grant retroactive relief, only the legislature is entrusted with the power to fashion new laws for the future.⁹ It is clear from the foregoing that in a system where the declaratory theory remains persuasive and judicial rulings operate retrospectively there is little chance for the doctrine of prospective overruling to take root.

One might be tempted to draw from these general observations the conclusion that the diversity of approaches towards precedents has influenced the way individual legal systems deal with this issue. Whereas this is to a certain extent true, it also appears that the categorizations and distinctions made in various jurisdictions transcend the traditional division between common and civil law systems. In fact, the decision as to the backward or forward application of judicial rulings is primarily dependent on the nature and factual circumstances of the case at hand and is mainly based on considerations of convenience or on sentiment of justice; and most of the time the outcome of a particular dispute rests on the balancing of the diverse interests involved rather than on a rigorous application of established criteria.

There is a strong argument that in the event of a statute being silent about the temporal effect of its provisions (b) it should be for Parliament, not judges, to remedy this defect. However, the practice of the courts on the subject of prospective effect does not offer a neat distinction between judicial rulings dealing with statutory law and those concerned with judge-made law.¹⁰ Therefore, in the following discussion the expression ‘prospective overruling’ will be used in both instances.

The forms prospective overruling may take include, first, pure prospective overruling.¹¹

Judges adopt prospective overruling in its ‘purest’ form when they declare that a new precedent is confined to future cases arising from events occurring after the announcement of the new holding; the dispute at hand being governed by the old ruling. This generally occurs in circumstances where the immediate application of the new ruling would be particularly harsh on the parties before the court. In such circumstances the principle of legitimate expectation in the continuing application of the previous case law would be particularly at risk. This model will be typically used in cases where the protection of public rights or civil liberties is at stake. A fairly common illustration is when a court overrules a past precedent by giving a new interpretation on statutory time limitations for a particular class of actions with the consequence that such a change would deprive a party to a pending case from having his case heard in court if applied immediately. Therefore, if as the consequence of such a ruling the plaintiff’s action would be time barred, the court may apply the new interpretation prospectively, thus preventing the plaintiff’s action to be denied as inadmissible. This has happened notably in the context of time limit for actions for defamation. For example, in France, the Court of Cassation took upon itself to overrule prospectively a former interpretation of a time-limitation rule for libel in a case where a radio station was sued for breach of the principle of presumption of innocence against a lawyer charged for professional misconduct.¹² In this case, not applying prospective overruling would have denied the defendant in the case to seek remedy in court and thus deprived her of her right to a fair trial within the meaning of article 6 § 1 of the European Convention on Human Rights.¹³ Similar solutions can be observed in other jurisdictions where an issue of time limitation or availability of review is raised in a case together with a breach of a fundamental right.¹⁴

However, despite the foregoing, pure prospectivity remains an exceptional device for three compelling reasons. One is that, if used too often, it would hamper the normal course of legal development through case-law. In some jurisdictions the courts themselves stress this point by declaring in the text of their judgment that the appellant has no vested rights to courts decisions remaining unchanged.¹⁵ Secondly, litigants would have no incentive to sue or appeal if they knew in advance that overruling would not improve their situation. Finally, for a court to merely announcing a new rule without applying it to the case at hand is equivalent to a mere dictum and thus faces the objection that in so doing judges act as legislators. This objection is considered further in part 3 of this chapter.

Other forms of prospective overruling are more limited and selective in their departure from the normal effect of court decisions. A common variation of prospective overruling is what has been termed limited pure prospectivity or qualified prospective overruling or selective prospectivity, whereby a new ruling applies not only to future cases but also to the instant case (ex nunc) but return to the old rule for all cases predating this decision including cases still open for review. A significant drawback with this model is that the new precedent does not necessarily (although it might) apply to other similar cases pending before the courts and is thus tantamount to inequality of treatment between litigants in similar position. This cannot be a satisfactory outcome in view that equality of application of the law is a manifestation of the principle of legal certainty as well as being a component part of the rule of law.¹⁶

In view of the foregoing criticisms addressed to prospective overruling, would a better approach to the question be to abandon the term prospective and use instead the phrase non-retroactive overruling as has been done in some jurisdictions both in their judicial practice and academic writing? This seems to be a better description of what a court actually does when confronted with the temporal effect of its decision. Non-retroactivity entails acting upon the backward application of a new principle of law in a way which fits the particulars of the situation in dispute. Seen from this angle, it becomes apparent that a court determines the outcome in relation to particular facts. Non-retroactive overruling thus becomes a judicial tool fashioned to mitigate the adverse consequences of judicial changes and a proper method of deciding cases. Presented this way it appears to be more consistent with what is expected from judges and therefore is most prone to promote consensus between judicial activists and those in favour of judicial restraint. Non-retroactivity is now examined in more detail.

Judges tend to proceed pragmatically when issues of prospective application arise. The idea of justice and the practical administration of society prevail over formal logic. Most of the time justification for non-retroactivity takes the form of a set of policy considerations raised by each particular dispute courts have to resolve. The principles of reliance, legal certainty, legitimate expectations and fairness are commonly cited in civil cases to support non-retroactivity; similarly, fair warning and due process of law are used in criminal proceedings; in the area of public law, the potential disruption in the running of public services justifies that constitutional rulings of invalidity do not operate retrospectively.

Deeper concern about the jurisdictional or theoretical basis of the ruling that operates prospectively may sometimes lead to the articulation of a number of proper factors or set of guidelines provided by the court itself to limit retroactivity. A typical illustration is the three factor retroactivity test laid down in 1971 by the US Supreme Court in Chevron Oil Co v. Huson. This test requires a three-part analysis as described by Justice Stewart in his opinion:

In our cases dealing with the non-retroactivity question, we have generally considered three separate factors. First, the decision to be applied non-retroactively must establish a new principle of law, either by overruling clear past precedent on which litigants may have relied, or by deciding an issue of first impression whose resolution was not clearly foreshadowed. Second, it has been stressed that “we 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.” Finally, we have weighed the inequity imposed by retroactive application, for “where a decision of this Court could produce substantial inequitable results if applied retroactively, there is ample basis in our cases for avoiding the ‘injustice or hardship’ by a holding of non-retroactivity.”¹⁷

A second illustration is provided by the European Court of Justice. In R (Bidar) v. Ealing London Borough Council where the Court sitting in Grand Chamber reiterated its basic approach that in defined circumstances it may exceptionally limit the temporal effect of a ruling:

The court has taken that step only in quite specific circumstances, where there was a risk of serious economic repercussions owing in particular to the large number of legal relationships entered into in good faith on the basis of rules considered to be validly in force and where it appeared that both individuals and national authorities had been led into adopting practices which did not comply with Community legislation by reason of objective, significant uncertainty regarding the implications of Community provisions, to which the conduct of other member states or the Commission may even have contributed …¹⁸

A final example of proposed guidelines in respect of prospective effect is the list of recommendations made by the special working committee set up in the early 2000s by the French Court of Cassation. In its Report to the Court the working group suggested that, in narrowly defined circumstances, decisions of the Court might be applied ‘non-retroactively’¹⁹ Without setting out any formal factors or criteria to be taken into account when considering whether a new ruling by the Court should apply retrospectively or not, the committee nevertheless recommended that the Court should limit the retrospective temporal effect of its ruling where there was (i) a strong motive of general public interest or (ii) a manifest disproportion between the general benefits attached to the retrospective effect of a court ruling (e.g. the fact that persons in like cases are treated equally) and the potential unfairness such a retrospective change in the law would occasion to the parties involved. The working group further recommended procedural safeguards in so far as prospective overruling could only be applied by the Court of Cassation itself which, for this purpose, should, first identify clearly and explicitly the meaning and scope of its new ruling in the case at hand and, secondly, allow each party to the case to put forward their respective views on whether to overrule a previous decision retrospectively or prospectively.

Special difficulties have been encountered in constitutional cases where a constitutional court strikes down legislation, or a longstanding program, or institution, as being unconstitutional.²⁰ Such declarations of invalidity may dramatically upset the running of public services or jeopardize the legitimate expectations of a category of citizens if they are given full retroactive effect. Two striking examples can be given to illustrate this point. One is the American case of Brown v. Board of Education where the US Supreme Court ordered in 1955 the dismantling of racially segregated schools in several states. Removing retroactively illegal schools under this new ruling would have affected the lives of thousands of pupils, parents, teachers and employees.²¹ Similarly, in the 1985 Canadian Manitoba Language Reference case, where the Supreme Court held that the Constitution required that the province of Manitoba legislation be enacted in English and in French, the Court ruling had the potential effect to invalidate all of the statute law of the province which, following the common law tradition, was only enacted in English. Thus, applying the declaration of invalidity retroactively would have left the province without laws and posed serious disruption in the legal system.²²

In order to avoid undesirable consequences in similar circumstances of invalidity, a first solution consists of applying prospectively the declaration of invalidity to cases in which the issue was raised as well as to future cases. As a consequence, despite the fact of the statute being deemed not to have existed at all, the decision of invalidity will not fully operate retroactively. Many authors have pointed out the conceptual difficulty here. Indeed, where a ruling of unconstitutionality is applied prospectively it necessarily means that the courts are upholding an unconstitutional law, albeit only for a limited period of time.²³

A slightly different approach from prospective effect is the suspension of the declaration of invalidity until a certain date, thereby allowing the legislature to enact valid legislation during the defined period.²⁴ Suspension of the nullified provisions for a defined period entails the maintaining of these provisions, or some of them, in the legal system in order to prevent a legal vacuum.²⁵ It is interesting to note that in the European Court of Justice tax case of Banco Popolare di Cremona v Agenzia Entrate Ufficio Cremona, Advocate General Jacobs proposed the suspension approach in respect of the Court’s rulings, suggesting that the retrospective and prospective effect of a ruling of the Court might be subject to a temporal limitation that the ruling should not take effect until a future date, namely, when the State had had a reasonable opportunity to introduce new legislation.²⁶

Contrary to the prospective and suspensory approach, a more orthodox view militates in favour of invalidity ab initio (ex tunc) each time a statute is found unconstitutional. In this respect, Irish law is of particular interest in that it highlights the particular dilemma posed by unconstitutional statutes where judges are faced with a choice between two unsatisfactory options; one being to declare the unconstitutional statute void ab initio, which may lead to unjust and chaotic consequences; a second option consisting of limiting the retrospective effect of the declaration of constitutional invalidity which runs counter the principle that unconstitutional law cannot be effective. Such a difficulty was manifest in two Irish leading cases, Murphy v. Attorney General and A v. Governor of Arbour Hill Prison, where the issues raised by invalidity were considered at length.²⁷ A possible way to escape such a theoretical conundrum would be to follow the solution frequently adopted by the German Constitutional Court whereby, instead of annulling the norm with immediate consequential retroactive effect, judges deliver a mere declaration of incompatibility subject to a future date before which no litigant may rely on the incompatibility in any claim against the State. In practice this has the same effect as a suspension order but, in theory, it is more consistent with the division of law making authority in so far as the court does not directly address or deal itself with the validity of the norm; the legislature is ultimately in charge of removing the norm from the statute book.

The Irish cases of Murphy and A further highlighted the problem posed by a potential, albeit limited, right to redress for harm caused pursuant to unconstitutional legislation, especially in overpaid taxation cases such as Murphy. Since a finding of unconstitutionality operates erga omnes (in relation to all), its benefit not being confined to the litigant in the case at hand, it may lead to further abundant litigation and have potential catastrophic consequences in the event of full redress being granted.²⁸ This would not be the case with the other above-mentioned models of declaration of invalidity since limiting a declaration of unconstitutionality to prospective effect only has the consequence of denying a remedy.

More generally, such difficulties in dealing with declarations of invalidity may have adverse consequences on the upholding of the rule of law in a legal system. Thus, it has been argued that if a finding of unconstitutionality had these devastating consequences for society in general and the legal system in particular which the courts found themselves unable to control, then this would inevitably impact on the practical willingness of the courts to make such a finding of unconstitutionality.²⁹