Canada v. Craig, 2012 SCC 43 was just released. It holds that a lower court may not overrule a higher court but may point out why the higher court’s precedent is problematic. It then deals with a court reversing itself on a prior precedent:
A. Should the Tax Court and the Federal Court of Appeal Have Followed Moldowan or Gunn?
 There is no doubt that Dickson J.’s interpretation of s. 13(1) in Moldowan, is a precedent binding on the Federal Court of Appeal and the Tax Court of Canada. While Gunn agreed with much of what Dickson J. wrote in Moldowan, on the crucial question of whether farming as a source of income could be subordinate to another source and still avoid the loss deduction limitation of s. 31(1), Gunn departed from Moldowan, a precedent binding on the Federal Court of Appeal.
 One of the fallouts from Gunn is that it left the Tax Court of Canada and the Federal Court of Appeal itself in the difficult position of facing two inconsistent precedents and having to decide which one to follow. The uncertainty which the application of precedent is intended to preclude is seen in the decisions since Gunn, in which the Tax Court has acknowledged Moldowan as the leading case while also feeling bound to follow Gunn: Stackhouse v. R., 2007 TCC 146,  3 C.T.C. 2402, Falkener v. R., 2007 TCC 514,  2 C.T.C. 2231, Loyens v. R., 2008 TCC 486,  1 C.T.C. 2547, Johnson v. The Queen, 2009 TCC 383, 2009 D.T.C. 1245, Scharfe v. The Queen, 2010 TCC 39, 2010 D.T.C. 1078, and Turbide v. The Queen, 2011 TCC 371, 2011 D.T.C. 1347. And of course the Federal Court of Appeal followed Gunn in the instant case.
 It may be that Gunn departed from Moldowan because of the extensive criticism of Moldowan. Indeed, Dickson J. himself acknowledged that the section was “an awkwardly worded and intractable section and the source of much debate”. Further, that provision had not come before the Supreme Court for review in the three decades since Moldowan was decided.
 But regardless of the explanation, what the court in this case ought to have done was to have written reasons as to why Moldowan was problematic, in the way that the reasons in Gunn did, rather than purporting to overrule it.
 The Federal Court of Appeal, on the basis of its prior decision in Miller v. Canada (Attorney General), 2002 FCA 370, 220 D.L.R. (4th) 149, in which that court reaffirmed the rule that it would normally be bound by its own previous decisions, followed Gunn, and not Moldowan. The application of Miller and the question of whether the Federal Court of Appeal should have followedGunn simply did not arise, in view of the Moldowan Supreme Court precedent.
 The Federal Court of Appeal’s purported overruling of Moldowan does not, however, affect the merits of this appeal or the core question of whether Moldowan should in fact be overruled.
B. Should This Court Overrule Moldowan?
 The question of whether this Court should overrule one of its own prior decisions was addressed recently in
 Nonetheless, this Court has overruled its own decisions on a number of occasions. (See R. v. Chaulk,  3 S.C.R. 1303, at p. 1353, per Lamer C.J., for the majority; R. v. B. (K.G.),  1 S.C.R. 740; R. v. Robinson,  1 S.C.R. 683.) However, the Court must be satisfied based on compelling reasons that the precedent was wrongly decided and should be overruled. (SeeR. v. Salituro,  3 S.C.R. 654, at p. 665; Minister of Indian Affairs and Northern Development v. Ranville,  2 S.C.R. 518, at p. 527; Hamstra (Guardian ad litem of) v.
 Courts must proceed with caution when deciding to overrule a prior decision. In Queensland v. Commonwealth (1977), 139 C.L.R. 585 (H.C.A.), at p. 599, Justice Gibbs articulated the required approach succinctly:
No Justice is entitled to ignore the decisions and reasoning of his predecessors, and to arrive at his own judgment as though the pages of the law reports were blank, or as though the authority of a decision did not survive beyond the rising of the Court. A justice, unlike a legislator, cannot introduce a programme of reform which sets at nought decisions formerly made and principles formerly established. It is only after the most careful and respectful consideration of the earlier decision, and after giving due weight to all the circumstances, that a Justice may give effect to his own opinions in preference to an earlier decision of the Court.
 The vertical convention of precedent is not at issue with respect to the decision as to whether the Supreme Court should overrule one of its own precedents. Rather, in making this decision the Supreme Court engages in a balancing exercise between the two important values of correctness and certainty. The Court must ask whether it is preferable to adhere to an incorrect precedent to maintain certainty, or to correct the error. Indeed, because judicial discretion is being exercised, the courts have set down, and academics have suggested, a plethora of criteria for courts to consider in deciding between upholding precedent and correcting error. (See R. v. Bernard,  2 S.C.R. 833, at pp. 850-61, Chaulk, at p. 1353, Henry, at paras. 45-46.)