There are basically three options:
1. On the basis that the constitution was always the higher law, and therefore the statute was always unauthorized, the decision could be retroactive -- everyone should be treated as if it had never been enacted.
2. On the basis that constitutional jurisprudence is really a different form of legislation, and legislation should usually be prospective (so as not to overturn reasonable reliance and other rule-of-law values), then the decision could be prospective. Past transactions would be left undisturbed, and the new constitutional rule would take effect after the decision.
3. The courts could recognize that the legislatures are better able to make this transitional determination, and give them room to decide in most cases.
The Supreme Court of Canada has not been good at coming up with a coherent answer as to which option should be chosen. In theory, #1 is the default rule for constitutional remedies. However, most cases with bigger social implications have, in fact, been given a suspended declaration of invalidity -- which doesn't take effect until sometime after the decision, and which would seemingly be inconsistent with retroactivity.
In addition, there is a doctrine against actions for damages on the basis of acts which were subsequently declared unconstitutional. Until recently, the same applied to actions for restitution.
The Pithlord will declare a preference for Option #3. Let the legislatures decide how they want to deal with the temporal issues of unconstitutionality (within reason, of course). Unfortunately, here we come up against the attitude of judicial supremacy.
Option #3 does require a default rule in the event that the legislatures don't act. Here, I would say that the best rule is retroactivity subject to ordinary limitation statutes.
2007 has seen two cases on the issue of the temporal effect of unconstitutionality. I hate to say it, but all the Court seems to have accomplished is to compound the confusion.
In Kingstreet, the issue was whether a provincial tax found to be "indirect" (and therefore unconstitutional) should be returned. In Canadian constitutional law, only the federal government can levy indirect taxes, which are defined as those for which the incidence falls on somebody other than the legal payor. (A difficulty with this test is that modern economic thought tells us that this applies to all taxes. Not surprisingly, the case law is therefore a bit confused.) Back in the seventies, the Court held that provincial laws preventing civil suits for unconstitutional taxes were unconstitutional. In the eighties, noting the potential fiscal difficulties, it then created a common law rule that unconstitutional taxes could not be recovered. In 2007, it now abolished both rules, leaving only the limitation period as a limit on recovery.
So far, this might be seen as an implementation of the Pithlord's preferred option. The retroactive effect of a declaration that a tax is unconsitutional will be up to the legislature, with a default rule of retroactivity.
Hislop involved a class action by survivors of same-sex couples claiming survivorship benefits under the Canada Pension Plan. The background is that the Supreme Court of Canada narrowly upheld different treatment of same-sex and heterosexual common law couples in federal pension plans in 1995. Then, in 1999, another Court decision convinced governments that they were going to lose if they continued to distinguish between heterosexual and same-sex couples. The federal government enacted specific transition provisions to determine both the earliest date of death for eligibility, and when the payments would start.
In Hislop, the Court gave with one hand and took away with the other. The Court struck down the legislated transitional provisions, but simultaneously decided that its prior ruling would not be retroactive. The general discussion of remedy is quite confused -- the general rule is supposedly retroactivity of remedy, but the prospectivity exception swallows it. There is some quite open discussion of the law-making function of judges in the Charter era:
In substance, the position of the appellants is predicated on the traditional – often called Blackstonian – view that judges never make law, but merely discover it. In this perspective, the courts are said to apply the law as it really was or has been rediscovered. As a consequence of the declaration of nullity, the appellants claim that they are entitled to the full benefits of the law, in conformity with an understanding of the Constitution, which is deemed to have never changed.
When the Court is declaring the law as it has existed, then the Blackstonian approach is appropriate and retroactive relief should be granted. On the other hand, when a court is developing new law within the broad confines of the Constitution, it may be appropriate to limit the retroactive effect of its judgment.
This is highly confused. Any court ruling creates new law in the sense that it can be used as an authoritative statement of positive law in the future. And courts may apply old constitutional principles to new situations. But the principles themselves are not supposed to change.
In any event, and however you try to address the counter-majoritiarian difficulty inherent in judicial review, the distinction between "declaring the law as it has existed" and "developing new law" will be impossible to draw. And making remedies -- which, after all, are the only things litigants care about -- depend on such philosophical considerations is madness.
Obviously, we are going to see a lot more on this as the lower courts try to figure out what they're supposed to do.