Test Claimants in the Franked Investment Income Group Litigation v Commissioners of Inland Revenue and another

before

Lord Hope, Deputy President
Lord Walker
Lord Brown
Lord Clarke
Lord Dyson
Lord Sumption
Lord Reed

JUDGMENT GIVEN ON

23 May 2012

Heard on 21, 22, 23, 27, 28 and 29 February 2012

Appellant Graham Aaronson QC Laurence Rabinowitz QC David Cavender QC (Instructed by Dorsey & Whitney (Europe) LLP)

Respondent David Ewart QC Rupert Baldry QC Andrew Burrows QC Kelyn Bacon Sarah Ford (Instructed by Solicitor for Her Majesty’s Revenue and Customs)

LORD HOPE

1. Very substantial judgments have been prepared in this case by Lord Walker, Lord Reed and Lord Sumption, to each of which I pay tribute. I wish in this short introduction to do two things. First, I shall say a bit about the background, to assist the reader in understanding at the outset what the issues are and to provide a guide to the passages in those judgments where they are dealt with. Second, I shall indicate briefly what my opinion is on each of them. I will however have to say a bit more about the one issue on which the court is divided: the DMG remedy/section 320 issue: see para 11, below. As it raises a question of EU law and the division of opinion shows that the answer to it is not acte clair, it is plain that it will need to be the subject of a reference to the Court of Justice for a preliminary ruling under article 267 TFEU.

The proceedings

2. As Henderson J explained at the outset of his judgment [2008] EWHC 2893 (Ch), [2009] STC 254, para 1, the Franked Investment Income (“FII”) Group Litigation with which these proceedings are concerned was established by a group litigation order on 8 October 2003. The test claimants are all companies which belong to groups which have UK-resident parents and also have foreign subsidiaries, both in the European Union and elsewhere. In the broadest terms, the purpose of the litigation was to determine various questions of law arising from the tax treatment of dividends received by UK-resident companies from non-resident subsidiaries, as compared with the treatment of dividends paid and received within wholly UK-resident groups of companies. The provisions giving rise to these questions related to the system of advance corporation tax (“ACT”) and to the taxation of dividend income from non-resident sources under section 18 (Schedule D, Case V) of the Income and Corporation Taxes Act 1988 (“the ICTA”) (“the DV provisions”). The relevant provisions of the ICTA have since been amended, ACT was abolished for distributions made on or after 5 April 1999 and the DV provisions were repealed for dividend income received on or after 1 April 2009. But the problems created by their existence in the past have not gone away.

3. The test claimants’ case is that the differences between their tax treatment and that of wholly UK-resident groups of companies breached article 43 (freedom of establishment) and article 56 (free movement of capital) of the EC Treaty (now articles 49 and 63 of the Treaty on the Functioning of the European Union) and their predecessor articles, and that these breaches have caused them loss dating back, at least in some cases, to the accession of the UK to the European Economic

Community signed at Brussels on 22 January 1972 and the introduction of ACT in April 1973. Their arguments are directed in part to issues of domestic law. But they are also directed to the extensive case law resulting from the application by the Court of Justice of the European Communities and, since the coming into force of the Lisbon Treaty, the Court of Justice of the European Union of principles of Community law to domestic tax systems, including an earlier reference in this case: Test Claimants in the FII Group Litigation v Inland Revenue Comrs (Case C- 446/04) [2007] STC 326. They raise difficult issues, and very large amounts of money are at stake. Henderson J was told that the maximum amount of the claims advanced in the FII Group Litigation was of the order of £5 billion.

4. The issues with which Henderson J had to deal were grouped by him under four headings: see [2009] STC 254, para 7. These were (1) the lawfulness of the UK rules imposing corporation tax on dividends received by UK parent companies from subsidiaries resident in other EU member states and, in some contexts, from subsidiaries in third countries, (2) the lawfulness of UK rules charging ACT on the onward distribution by UK-resident companies of dividend income received from such subsidiaries, (3) the lawfulness of rules applicable to dividends payable out of distributable foreign profits which permitted an election to be made to treat such income as foreign income dividends (“FIDs”) and (4) a number of fundamental questions relating to remedies.

5. He held that it followed from the judgment of the ECJ under the earlier reference that the UK rules on corporation tax on overseas dividends were not compatible with Community law as regards dividends from subsidiaries resident in other member states, and that the UK legislative scheme as regards FIDs also breached Community law. A further reference was however required in relation to two of the issues relating to liability: paras 138, 197. As for the issues relating to remedies, it was common ground that two types of restitutionary remedies are available in domestic law: a claim for restitution of tax unlawfully demanded under the principle established in Woolwich Equitable Building Society v Inland Revenue Comrs [1993] AC 70 (“Woolwich”), and the claim for tax wrongly paid under a mistake which was recognised in Deutsche Morgan Grenfell Group plc v Inland Revenue Comrs [2006] UKHL 49, [2007] 1 AC 558 (“DMG”).

6. Henderson J held that, under the principle laid down in Amministrazione delle Finanze dello Stato v SpA San Giorgio (Case 199/82) [1983] ECR 3595 (“San Giorgio”), EU law required there to be an effective remedy for monies paid in respect of the tax that was unlawfully charged. The test claims were properly to be classified in English law as claims in restitution based on a mistake of law. The Woolwich cause of action (which is now time-barred), for which mistake was not a necessary ingredient, was likely to play a subsidiary role in such cases: para 260. It was not open to the Revenue to rely on section 320 of the Finance Act 2004 (“Section 320 FA 2004”) or section 107 of the Finance Act 2007 (“Section 107 FA

2007”) to exclude DMG mistake claims, as these provisions purported to curtail the extended limitation period under section 32(1)(c) of the Limitation Act 1980 without notice and without providing any transitional arrangements to protect the right under Community law. But the test claimants had failed to establish any sufficiently serious breach to entitle them to damages.

7. The case then proceeded to the Court of Appeal (Arden, Stanley Burnton and Etherton LJJ): [2010] EWCA Civ 103, [2010] STC 1251. The various issues were made the subject of an agreed list which the court amended and to which it gave numbers. They were identified in an index at the beginning of the judgment, to which reference may be made. Issues 1 to 10 related to liability. Issues 11 to 23 were concerned with remedy. The Court of Appeal was divided as to the meaning of para 54 of the judgment of the ECJ with respect to one of the test claimants’ submissions on liability, so it held that a reference should be made on that issue. On all but one of the other issues relating to liability it agreed with the judge. On four issues relating to remedy the appeal by the Revenue was allowed. Differing from the judge, it held that the Woolwich restitution remedy was a sufficient remedy as EU law does not require that there must also be a remedy based on mistake (issue 12); that the Woolwich restitution remedy met the requirements of EU law and was not affected by sections 320 FA 2004 and 107 FA 2007 (issues 20 and 21); and that section 33(2A) of the Taxes Management Act 1970 (“TMA”) (issue 23), which excludes relief under that section where Case V corporation tax has been paid under a mistake, applied to an assessment based on a provision that infringed Community law as a conforming interpretation could be given to it. Issue 22, as to whether section 32(1)(c) of the Limitation Act 1980 applied to a Woolwich claim, was not argued before the judge. But it was argued before the Court of Appeal, which held that it could not be given that wider meaning.

8. Applications for permission to appeal to the Supreme Court were lodged by both parties. On 8 November 2010 the panel refused permission on the issue as to which the Court of Appeal decided that there should be a reference, and it remitted another issue relating to liability to the management judge to frame a reference on that point also. The time limit for making an application for permission on a number of other issues, including issue 22, was extended until the references had been determined by the ECJ and its rulings applied by the Court of Appeal. But permission to appeal was given on four issues relating to remedy: issues 12, 20, 21 and 23. Shortly before the hearing of the appeal permission was given to the claimants for issue 22 to be argued also.

The issues

9. The parties are agreed that the issues in the appeal are best expressed as follows:

“(1) Could Parliament lawfully curtail without notice the extended limitation period under section 32(1)(c) of the Limitation Act 1980 for the mistake cause of action (section 320 FA 2004) and cancel claims made using that cause of action for the extended period (section 107 FA 2007)? In particular:

1. Would a Woolwich restitution remedy be a sufficient remedy for the repayment claims brought on the basis of EU law (Court of Appeal issue 12)?

2. Whether or not a Woolwich restitution remedy would be a sufficient remedy, does EU law protect the claims which were made in mistake; and, specifically, did the curtailment without notice of the extended limitation period for mistake claims (section 320 FA 2004) and the cancellation of such claims in respect of the extended period (section 107 FA 2007) infringe the EU law principles of effectiveness, legal certainty, legitimate expectations and rule of law (Court of Appeal issues 20 and 21)?

(2) Are the restitution and damages remedies sought by the test claimants in respect of corporation tax paid under section 18 (Schedule D, Case V) of the ICTA 1988 excluded by virtue of the statutory provisions for recovery of overpaid tax in section 33 of the Taxes Management Act 1970 (Court of Appeal issue 23)?”

To that there must be added the following:

“(3) Does section 32(1)(c) of the Limitation Act 1980 apply to a claim for a Woolwich restitution remedy (Court of Appeal issue 22)?”

As Lord Walker explains in para 35 below, a further issue became apparent as the parties’ submissions on issues 12, 20 and 21 have developed which can be expressed as follows:

“(4) Does the Woolwich restitution remedy apply only to tax that is demanded by the Revenue, and not to tax such as ACT which is payable on a return; and, if so, what amounts to a demand?”

10. In the judgments that follow:
11. Issue (4), above, the question whether a Woolwich claim arises only where a demand has been made by the Revenue, is dealt with by Lord Walker in paras 64- 83 and by Lord Sumption in paras 171-174.

2. Issue (3), above (Court of Appeal issue 22), as to whether section 32(1)(c) of the Limitation Act 1980 should be widely construed so as to give a Woolwich restitution remedy the benefit of the extended limitation period, is dealt with by Lord Walker in paras 42-63 and by Lord Sumption in paras 177-185.

3. Issue (2), above (Court of Appeal issue 23), as to whether section 33 of the TMA is incompatible with EU law because it excludes the test claimants’ right of action at common law, is dealt with by Lord Walker in paras 116-119 and by Lord Sumption in paras 204-205.

I agree, for all the reasons they give, that each of these three distinct issues should be answered in the negative. I would uphold the judgment of the Court of Appeal on issues (3) and (4) and, because it should not be read as excluding rights of action for the recovery of tax charged contrary to EU law, I would allow the appeal on issue (2) as to the meaning of section 33 of the TMA.

The DMG remedy/section 320 issue

11. The remaining issue (issue (1), above) is an issue of EU law. The background is provided by the ruling of the Grand Chamber that it is for the domestic legal systems of each member state to lay down the detailed procedural rules governing actions for safeguarding rights which individuals derive from Community law, and that the national courts and tribunals before which claims are brought are obliged to ensure that individuals should have an effective legal remedy enabling them to obtain reimbursement of the tax unlawfully levied by a member state or withheld by it directly against that tax: Test Claimants in the FII Group Litigation v Inland Revenue Comrs (Case C-446/04) [2007] STC 326, paras 202-203.

12. It follows from the answers given to issues (3) and (4) that this issue must be approached on the basis that a Woolwich claim would have been available had it been brought in time. But it has been excluded by the expiry of the limitation period. The test claimants are left therefore with their DMG mistake claim. It has the benefit of the extended limitation period, but the Revenue say that it has been excluded by section 320 FA 2004 and section 107 FA 2007.
13. As Lord Walker explains in para 38, the question is whether EU law requires only that the member state must make available an adequate remedy which meets the principles of effectiveness and equivalence, or whether it requires every remedy recognised in domestic law to be available so that the taxpayer may obtain the benefit of any special advantages that this may offer on the question of limitation. The position in domestic law is not now in doubt. In DMG it was held that the taxpayer was entitled to take advantage of the remedy which was most advantageous to him. The fact that a Woolwich claim was not available because it was subject to a shorter limitation period did not prevent him from pursuing his mistake claim if his interests were best suited by doing so.

14. This issue can be broken down into three questions: (1) would Woolwich on its own provide a remedy for the test claimants’ San Giorgio claims which satisfies the requirements of the EU principles of effectiveness and equivalence? (2) were those principles, and the principle which protects legitimate expectations, infringed by section 320 FA 2004, which curtailed without notice the extended limitation period for mistake claims? (3) were these principles infringed by the retrospective cancellation of such claims by section 107 FA 2007 in respect of the extended period?

15. Lord Walker and Lord Sumption are agreed that section 107 FA 2007 was contrary to EU law, although they do not reach that conclusion by the same route. This is because they disagree on the primary issue as to whether Woolwich on its own was sufficient to meet the requirements of effectiveness and equivalence. Having reached the view that it was not, Lord Walker holds that section 320 FA 2004 was not compatible with EU law as it infringed those principles and maybe that it infringed the principle of legitimate expectations too: para 114-115. Lord Sumption disagrees. He holds that the Woolwich remedy on its own with a normal limitation period was an effective way of asserting the test claimants’ EU right, that there was no obligation on the UK to maintain a concurrent right and that, for this reason and because the test claimants could not have had a legitimate expectation that they would have the benefit of the extended limitation period, section 320 FA 2004 was lawful: paras 198-202. But, because the circumstances had changed and they had acquired a legitimate expectation by 2006, it was contrary to that principle for that expectation to be defeated by section 107 FA 2007. Like Lord Walker (see para 115), I agree with Lord Sumption’s reasoning in para 203 as to section 107 FA 2007.

16. On the primary issue however, like Lord Reed, I agree with Lord Walker. I would take as my starting point the fact that in domestic law two types of restitutionary remedies are available and that the taxpayer is entitled to take advantage of the remedy that is most advantageous to him: a claim for restitution of tax unlawfully demanded under the principle established in Woolwich, and the claim for tax paid under a mistake of law which was recognised in DMG. It is, of

course, true that DMG had not yet reached the House of Lords when section 320 FA 2004 was enacted. But the common law rule that money which had been paid under a mistake of law was not recoverable had already been rejected. It was rejected in Scotland in Morgan Guaranty Trust Co of New York v Lothian Regional Council 1995 SC 151, for reasons that were special to Scots law, and in South Africa in Willis Faber Enthoven (Pty) Ltd v Receiver of Revenue 1992 (4) SA 202. But it had also been rejected by the common law in Canada: see the dissenting opinion of Dickson J, with which Laskin CJ agreed, in Hydro Electric Commission of Township of Nepean v Ontario Hydro [1982] 1 SCR 347, 357-370. Dickson J’s opinion was adopted by La Forest J, with whom Lamer, Wilson and L’Heureux-Dubé agreed on this point, in Air Canada v British Columbia [1989] SCR 1161. The same result was reached in Australia in David Securities Pty Ltd v Commonwealth Bank of Australia (1992) 175 CLR 353. Then in Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349 the House of Lords held that the rule could no longer be maintained, and that it should be recognised that there was a general right to recover money paid under a mistake, whether of fact or law.

17. It was contended for the Inland Revenue Commissioners in DMG that the general right of recovery did not apply in the case of payments made under a mistake of law to the revenue. But this topic had already been the subject of comment by one of the most distinguished and influential scholars on the law of restitution, the late Professor Peter Birks. He declared that, unless displaced by statute, causes of action good against private citizens are no less good against public bodies: see his essay (in the volume Essays on Restitution (1990), edited by Professor P D Finn) entitled “Restitution from the Executive: a Tercentenary Footnote to the Bill of Rights”, at p 174. He also made the point that, if in Woolwich the building society had made a mistake of fact, it would undoubtedly have entitled the society to restitution of the money it paid to the revenue in consequence of its mistake, just as it plainly would have been had the transaction been with a private citizen. The decision of Park J at first instance in DMG [2003] 4 All ER 645, [2003] STC 1017, in which he upheld the taxpayer’s claim for repayment of tax wrongly paid under a mistake of law with an extended limitation period, should be seen against this background.

18. As Henderson J observed in para 406 of his judgment, it was not possible to predict with any confidence what the outcome would be of the appeals in DMG that were to follow. But I think that it would be going too far to say that Park J’s judgment was bound ultimately to be set aside. The fact that on 8 September 2003, less than two months after Park J’s judgment was delivered on 18 July 2003, the Paymaster General announced the introduction of what was to become section 320 FA 2004, and said that it was to affect proceedings issued on or after that date, suggests that the revenue had at least some expectation that it would not be successful in achieving that result. Like Lord Walker (see para 108), I think that the suggestion that the Court of Appeal’s decision was just a “bump in the road”

understates the strength of the arguments in support of its appeal. But I cannot agree with Lord Sumption (see paras 200-201) that it was unrealistic for there to have been a reasonable expectation by that date that the right of recovery on the ground of mistake with an extended limitation period would be upheld. My own view lies between these two extremes.

19. I share Lord Walker’s view that it would have been helpful to have had the view of the judge on this issue: para 112. But I also think that in para 243 Lord Reed has identified the right way to look at it, which does not require anything more than we already know. One must ask oneself what the test claimants were entitled to expect when they made their claims based on mistake. There was no certainty at that time when section 320 FA 2004 was enacted that their claims based on mistake would succeed. But those claims were undoubtedly arguable, as the subsequent ruling by the House of Lords in DMG [2007] 1 AC 558 made clear. They were entitled to expect that the question whether their claims based on mistake were well founded would be decided by the courts, as there was a real issue to be tried. They were also entitled to expect, according to the principle of legal certainty, that this entitlement would not be removed from them by the state by the introduction without notice of a limitation period that was not fixed in advance: see ACF Chemiefarma v Commission of the European Communities (Case 41/69) [1970] ECR 661, para 19; Marks & Spencer plc v Customs and Excise Comrs (Case C-62/00) [2003] QB 866, para 39.

20. The crucial question, however, is whether the retrospective application of that limitation period to claims based on mistake was in conformity with the principles of equivalence and effectiveness, as explained by the Grand Chamber in its judgment in these proceedings: Case C-446/04 [2007] STC 404, para 203. I accept, of course, that the Woolwich remedy on its own was an effective way of vindicating the San Giorgio right. But what about the principle of equivalence which, as Lord Reed points out in para 218, is a complementary requirement? The Woolwich remedy was not the only remedy in domestic law, as it was held in DMG that a taxpayer who wrongly paid tax under a mistake of law is entitled to a restitutionary remedy against the revenue. The theory is that judicial decisions must be taken to declare the law that applies to the case with retrospective effect, whenever the events that gave rise to the claim occurred. So, in the events that have happened, the DMG remedy must be taken to have been always available. It is not just a mirror image of the remedy that is afforded under Woolwich. Both remedies lead to the same result. But they are different remedies founded upon different principles and they are subject to different limitation periods. There may be other differences, depending on the facts and circumstances of each case.

21. There is no obvious way of deciding which of these two remedies must be adopted if only one can be allowed. Is it to be held the claimant is under an obligation, if both are available, to select the remedy which best suits his

opponent? This would be an odd result, as I said in DMG [2007] 1 AC 558, para

51. For the reasons which I gave in that paragraph, I think that domestic law must reject this idea because it has no basis in principle. In fairness, the claimant ought to be free to choose the remedy that best suits his case. The principle of equivalence requires that the rules regulating the right to recover taxes levied in breach of EU law must be no less favourable than those governing similar domestic actions. So it seems to me that it must follow, if the means of recovering of taxes levied contrary to EU law are to match those in domestic law, that both remedies should be available.

Conclusion

22. For these reasons, and those given more fully by Lord Reed, I agree with Lord Walker’s analysis. I would hold that Parliament could not lawfully curtail without notice the extended limitation period under section 32(1)(c) of the Limitation Act 1980 for the mistake cause of action by section 320 FA 2004. I agree with both Lord Walker and Lord Sumption that it could not cancel claims made using that cause of action for the extended period by section 107 FA 2007. The question whether there was a legitimate expectation of bringing an action of the kind that was excluded by that section does not raise any issue of EU law. So I do not think that there are grounds for seeking a reference on that point.

23. I recognise however that, as there is a division of opinion among us as to whether EU law requires that both remedies should be available to the test claimants so that they can choose the remedy that best suits their case for reimbursement, the answer to that question cannot be regarded as acte clair. I would therefore invite the parties to prepare in draft the question or questions on which they suggest a preliminary ruling should be sought from the CJEU, and a brief note of the submissions that each party would wish to be included in the reference. I would also invite their views as to whether this reference should be combined with the references that are to be made on the other issues, or whether it should be submitted separately.

LORD WALKER

Introduction

24. This appeal is a further stage, but by no means the last stage, in complex and protracted group litigation, designated as Test Claimants in the FII [franked investment income] group litigation. In this group litigation, and other parallel group litigation proceedings, numerous issues have been raised as to whether

features of the UK corporation tax regime infringe EU law, and as to the remedies available to companies which claim to have been financially disadvantaged in various ways by such infringements. These proceedings have already resulted in two references to the Court of Justice.

25. Since the Court of Justice’s judgment on the first reference¹ wide language extending the scope of the sections to actions not expressed to be brought on the grounds of mistake. So the apparently self-inflicted injury which the test claimants invite would seem to require an amendment to the pleadings, and even then (if the section 32(1)(c) argument succeeds) the Woolwich claim would remain with a six- year limitation period, which is what it has always been assumed to have. Nevertheless, the section 32(1)(c) point is an important point of law that has been fully argued, and so it should be addressed.

    Section 32(1)(c)

    42. Section 32(1) of the Limitation Act 1980 provides:

    “Subject to [provisions not now material], where in the case of any action for which a period of limitation is prescribed by this Act, either –

    1. the action is based upon the fraud of the defendant; or
    2. any fact relevant to the plaintiff’s right of action has been deliberately concealed from him by the defendant; or
    3. the action is for relief from the consequences of a mistake;

    the period of limitation shall not begin to run until the plaintiff has discovered the fraud, concealment or mistake (as the case may be) or could with reasonable diligence have discovered it.

    References in this subsection to the defendant include references to the defendant’s agent and to any person through whom the defendant claims and his agent.”

    It replaces (with a minor amendment to section 32(1)(b)) provisions first enacted in section 26 of the Limitation Act 1939, in which section 26(c) was in the same terms as section 32(1)(c). The change in the law made in 1939 was recommended by the Law Review Committee (chaired by Lord Wright MR) in its Fifth Interim Report, (Statutes of Limitation) (1936) (Cmd 5334). Indeed the expression “relief from the consequences of a mistake” appears three times in para 23 of the report, dealing with this topic. The recommendation was that in such cases the equitable rule (that time should run only from when the mistake was, or could with reasonable diligence have been, discovered) should apply to claims which were formerly within the exclusive jurisdiction of common law courts (as opposed to being within equitable or concurrent jurisdiction).

    43. The previous state of the law was established by the decision of Hamilton J (later Lord Sumner) in Baker v Courage & Co [1910] 1 KB 56. The facts were that the plaintiff was the former owner of a public house who had in 1896 been mistakenly overpaid by £1,000 on the sale of his leasehold public house to the defendants, who were brewers. The plaintiff then deposited £9,000 at interest with the defendants. In 1909 he wished to withdraw the last of the deposit (standing, as it happens, at £1,000) but the defendants, on reviewing the position, discovered their mistake and refused to return the money. When sued they pleaded set-off and made a counterclaim, both of which were opposed as statute-barred.
    44. Hamilton J referred (at p 62) to the purely equitable claim made in Brooksbank v Smith (1836) 2 Y & C Ex 58, a decision of Alderson B sitting in the equity side of the Court of Exchequer. Hamilton J said that Brooksbank v Smith

    “was a case to which the Statute of Limitations did not apply; and the rule which was there laid down was one which in my opinion cannot be transferred to cases like the present, to which the statute does directly apply. In dealing with the latter class of cases, Courts of Equity were just as much bound by the statute as were Courts of Common Law.”

    In any event, he went on, the brewers had had the means of knowing the truth throughout, if they had chosen to look at the sale contract and examine their books of account. He also rejected a second contention that time did not start to run until notice of the mistake (that is, the overpayment of £1,000 in 1896) had been given to the plaintiff and a demand had been made.

    45. It is common ground that section 26(c) of the Limitation Act 1939 was intended to reverse the first point of principle (though not, on the facts, the result) in Baker v Courage & Co. The issue is how much further the change in the law was meant to go. The leading case on that point is the decision of Pearson J in Phillips-Higgins v Harper [1954] 1 QB 411. It was fully argued, and the argument is fully reported. Professor Andrew Burrows has noted that there was an unsuccessful appeal on the facts by the defendant, briefly reported at p 420, but no cross-appeal on the limitation point. The decision of Pearson J has been followed by the Court of Appeal, apparently with little or no oral argument on the point, in Malkin v Birmingham City Council (unreported) 12 January 2000, a claim for breach of statutory duty. The judgment of the Court of Appeal in this case recorded [2010] STC 1251, para 242 that Mr Ewart (for HMRC) “very generously” did not submit that the Court of Appeal was bound by Malkin. In any event the Court of Appeal, after full argument, accepted Phillips-Higgins and Malkin as correct. It did so after considering the history and language of section 32(1)(c), and the reasoning in the judgment of Pearson J (a long passage from which is set out at para 240). But for the general importance of the point, it might be sufficient to say that the Court of Appeal was right, and for the right reasons.
    46. Phillips-Higgins v Harper was an action by a woman solicitor who had been employed as an assistant by a sole practitioner, Mr Harper, between 1938 and 1950, when she became a salaried partner. Her employment was, on her case, at a basic salary supplemented by an annual sum to bring her total remuneration up to one-third of the net profits of Mr Harper’s practice. Mr Harper contended that (until 1948) the bargain was to supplement her remuneration to one-quarter of the net profits as determined by his accountant, and he pleaded the Limitation Act 1939. The judgment is reported verbatim only on this point, but it is recorded (at p 413) that Pearson J found: “(1) that the original fraction of the relevant profit figure to which the plaintiff was entitled was one-third, and that that fraction had been reduced to one-quarter by the defendant by private instructions to his accountant and that the plaintiff did not know and did not consent to the reduction;
    2. that the plaintiff’s contention as to the relevant profit figure was correct; and
    3. that there had been no intention on the part of the plaintiff to agree the accounts over the material period and that therefore they had not been agreed and settled.”
    47. Mr Harper’s position was therefore deeply unattractive. But the plaintiff was not mistaken about the bargain; her mistake was in believing that Mr Harper and his accountant were giving proper effect to it. As the judge hinted at p 418, the plaintiff might possibly have done better to rely on section 26(b), since although at that time it required fraudulent concealment, that expression was interpreted fairly broadly.
    48. For present purposes the crucial passage is earlier on p 418. It is part of the passage quoted by the Court of Appeal, but it bears repetition:

    “What, then, is the meaning of provision (c)? The right of action is for relief from the consequences of a mistake.   It seems to me that this wording is carefully chosen to indicate a class of actions where a mistake has been made which has had certain consequences and the plaintiff seeks to be relieved from those consequences. Familiar examples are, first, money paid in consequence of a mistake: in such a case the mistake is made, in consequence of the mistake the money is paid, and the action is to recover that money back. Secondly, there may be a contract entered into in consequence of a mistake, and the action is to obtain the rescission or, in some cases, the rectification of such a contract. Thirdly, there may be an account settled in consequence of mistakes; if the mistakes are sufficiently serious there can be a reopening of the account.”

    All these are examples of relief which removes or mitigates the adverse consequences to the claimant of the mistake, while respecting the position of the defendant where justice so requires (for instance by the defence of change of position where money has been paid under a mistake, or the requirement for restitutio in integrum where rescission is granted). It is an important but still relatively narrow category of causes of action, and much narrower than that for which Mr Rabinowitz has contended.

    49. Mr Rabinowitz was critical of the decision of the Court of Appeal as having paid insufficient attention to the statutory language and the traditional equitable rules, and too much attention to the report of the Law Revision Committee. In his reply (day 5, page 136) he invited the court to read the first sentence of para 23 of the report as if it had contained a parenthesis, saying:

    “Where mistake is not an essential part of the claim because we know that’s what the equitable rule is.”

    In support of this he relied on Brooksbank v Smith (1836) 2 Y & C Ex 58 and Denys v Shuckburgh (1840) 4 Y & C Ex 42, another decision of Alderson B sitting in the equity side of the Court of Exchequer.

    50. As to the statutory language, the criticism is in my view misplaced. The Court of Appeal cited and agreed with Pearson J’s view that the wording is carefully chosen to indicate a category of actions with particular characteristics. As to the report of the Law Revision Committee, it showed (as would be expected of its distinguished membership) a full awareness of the historical background. The parenthesis suggested as a gloss by Mr Rabinowitz is not borne out by the example that comes at the end of the first sentence of para 23, that is money or property

    transferred under a mistake, where the mistake is an essential part of the claim, and would have to be pleaded with some particularity.

    51. The authorities cited by Mr Rabinowitz do not support the wide equitable jurisdiction for which he contended. Brooksbank v Smith 2 Y & C Ex 58 was about a will trust. The testatrix died in 1818 leaving a fund in trust, subject to a life interest, for her children in equal shares, with substitutional gifts if any child predeceased her leaving issue. Her daughter Elizabeth did predecease her by two months, but on the death of the life tenant in 1827 the trustees were given incorrect information about the date of Elizabeth’s death and her share (£1,000 nominal of stock) was transferred to her widower instead of to her children. When the mistake was discovered in 1833 the trustees claimed £100 stock (which was all that remained unsold) from Elizabeth’s widower. The bill was issued within six years of discovery of the mistake. Alderson B held that the claim was not statute-barred. He treated it as a proprietary claim based on a mistake of fact.
    52. Denys v Shuckburgh 4 Y & C Ex 42 was similar, though the facts were more complicated. Under a marriage settlement made in 1793 Earl Pomfret settled two quarter shares in some lead mines in Yorkshire on trusts under which he had both an immediate life interest and an ultimate reversion (with intermediate trusts that in due course failed). In 1813 the Earl (whose marriage was childless and ended in judicial separation) sub-settled (but only during his own lifetime) one quarter share on his sister, Lady Caroline, and another on her son William. Lady Caroline owned another quarter share of the mines in her own right. In 1826 the Earl assigned the whole of his reversionary interest to William. On the Earl’s death in 1830 no one adverted to the fact that the 1813 sub-settlement then came to an end, and the right to income from one-quarter share of the mines passed from Lady Caroline to her son William. He went abroad in 1832 and Lady Caroline died in 1835. The mistake was not discovered until 1839, when William brought a bill against his mother’s estate to recover arrears of income.
    53. Alderson B stated the principle at, p 53:

    “The plaintiff contends, that he has established that this receipt has been by mistake of fact, and that this is on the same footing as fraud, and prevents the operation, if made out, of the Statute of Limitations; which in equity is adopted as a guide, but is not at law binding on the court. I agree in that conclusion, if the circumstances of the case warrant it. But here, it seems to me, that the plaintiff had the means, with proper diligence, of removing the misapprehension of fact under which I think he did labour. He had in his power the deed on which the question turns; and, although it is perhaps rather obscurely worded, still I think he has allowed too much time to elapse not to be

    fairly considered as guilty of some negligence; and a Court of Equity, unless the mistake be clear, and the party be without blame or neglect in not having discovered it earlier, ought, in the exercise of a sound discretion, to adopt the rule given by the statute law as its guide.”

    He also referred, during counsel’s argument, to the position at common law. As it happened part of the misapplied income was represented by identifiable lead ore stored at Richmond. When counsel for the plaintiff argued that Lady Caroline became liable to an action for money had and received only when she sold the lead, Alderson B commented, at p 48:

    “If she sold the lead and received the produce, you might have waived the tort, and brought an action for money had and received. But then the Statute of Limitations runs from the conversion, and not from the time of receiving the money.”

    54. These authorities were cited to Warrington J in In Re Robinson [1911] 1 Ch

    502. There the mistake was on a fairly arcane point of law, that an entail created by royal grant as a reward for services cannot be barred: Robinson v Giffard [1903] 1 Ch 865. That decision showed that deeds executed over 40 years before and intended to bar an annuity granted in tail by King Charles II were ineffective. The claim was to recover arrears of the annuity. Warrington J identified, at p 513 three types of case where there is no time-bar for recovery of mistaken payments by trustees: (1) when an estate is being administered by the court; (2) proprietary claims to recover identifiable trust assets or their traceable proceeds; and (3) claims against third parties in knowing receipt of trust property. By contrast the claim before him:

    “is in substance a mere money demand to which a Court of Equity, acting by analogy to the statute, would apply the same period of limitation. I think, therefore, that the plaintiff’s claim is barred by the statute, and that the action fails.”

    55. The analysis in In Re Robinson was followed by Romer J in In Re Mason [1928] Ch 385 and approved by the Court of Appeal on appeal in that case [1929] 1 Ch 1. That was a claim, brought after a very long lapse of time, to recover an estate that had been taken by the Crown as bona vacantia. In the Court of Appeal Lord Hanworth MR distinguished, at p 9, between the discovery of a mistake which was a cause of action and discovery of the evidence needed to prove the cause of action. He said:

    “It is suggested by Miss Mason that it is only when she found proof of the marriage of Maria L’Epine’s parents that she was entitled to bring this claim. A confusion seems to have arisen between the power to prove a claim and the right to bring it. The cause of action on which this claim is founded arose so far back as one of the three dates I have mentioned, 1798, 1801 or 1831, and the last of these dates is nearly 100 years ago. The fact that the useful evidence did not turn up until 1921 does not affect the date when the cause of action arose.”

    56. In re Blake [1932] 1 Ch 54 was another bona vacantia case, though the interest had been assigned by the Crown to third parties. Maugham J stated, p 60:

    “An action in the Chancery Division brought by the next of kin against a person to whom the administrator had wrongly paid part of the personal estate of the intestate under a mistake of fact (not joining the administrator and seeking administration) would be in the nature of a common law action for money had and received, and the Court acting on the analogy of the Statute of James I (21 Jac 1, c 16) would hold the claim to be barred after the lapse of six years from the date of payment: see In Re Robinson [1911] 1 Ch 502, where the law is elaborately explained by Warrington J, and In Re Mason [1928] Ch 385; [1929] 1 Ch 1. A common law action of the same character, assuming that such an action would lie, would also be barred by the same statute after the expiration of six years from the date of payment: Baker v Courage & Co [1910] 1 KB 56, 63. On the other hand there is no doubt that in a proper case the next of kin might bring an action in the Chancery Division to follow the trust property if the defendant to whom the administrator had paid it were still in possession of it.”

    57. The last relevant authority is an obiter passage in the monumental judgment of the Court of Appeal in In Re Diplock [1948] Ch 465. It was concerned with both personal and proprietary claims against numerous charities. The claims arose in consequence of the executors’ calamitous distribution of the testator’s valuable residuary estate in the mistaken belief that it was held on a valid charitable trust. The executors had by then compromised claims against them personally. In relation to a point which was not determinative Lord Greene MR, delivering the judgment of the court, observed at pp 515-516:

    “If [the respondent charities] seek to bring the case, for the purposes of the defence of limitation, within section 2 of the [Limitation Act 1939] and to rely upon the reasoning in In Re Blake [1932] 1 Ch 54,

    they must do so by averring that the cause of action is analogous to the common law action for money had and received. And if they assert the analogy, they must take it with its attributes and consequences. Beyond doubt, it would appear that in the case of an action at common law to recover money paid under a mistake of fact, section 26 would now operate to postpone the running of time. It is true that no such action would lie where the mistake is one of law: but for reasons which we have already given we do not accept the respondents’ contention that the ‘analogous’ claim in equity will also lie only where the mistake was one of fact. In our judgment, therefore, assuming the analogy (as it must be assumed if section 2 is to apply at all) the action is one for the recovery of money paid away by mistake – albeit by the mistake of other persons and by a mistake of law – and in our judgment, on this assumption, is an action for relief from the consequences of mistake no less than would be an action at common law to recover money paid away under a mistake of fact.”

    58. The analogy with the common law action for money paid under a mistake is a recurring feature of these authorities. Indeed, the analogy goes right back to the great case of Moses v Macferlan (1760) 2 Burr 1005, the fountain-head of the English law of unjust enrichment. This has been explained in a recent article by the Hon Justice W M C Gummow of the High Court of Australia, “Moses v Macferlan 250 Years On” (2010) 84 Austl LJ 756, (2011) 68 Washington and Lee Law Review 881, 882-888, citing Moses v Macferlan at 97 E R 676, 679-680 and Clark v Shee and Johnson (1774) 1 Cowp 197, 199-200 for the proposition that the action for money had and received was “a liberal action in the nature of a bill in equity”.
    59. In the old authorities the matter is sometimes treated simply as a case of mistake, without further analysis. But in the cases where the period was or might have been extended the mistake seems to have been an essential ingredient in the cause of action. Dr James Edelman, in “Limitation Periods and the Theory of Unjust Enrichment” (2005) 68 MLR 848, reads Denys v Shuckburgh differently. In this he follows Franks, whose monograph on Limitation of Actions (1959) suggests, at p 206 that the decision in Phillips-Higgins v Harper was too narrow:

    “In particular it seems clear that a beneficiary under a will or trust who claims directly against a person to whom trust property has been wrongfully transferred can rely upon the mistake of the personal representative or trustee to postpone the running of time; although his cause of action rests upon his own title and the defendant’s lack of title to the property – and the action would be just the same if the

    property had been transferred purposely, ie, with knowledge that the recipient was not entitled.”

    But Franks goes on to comment that if Pearson J’s view is rejected the scope of section 26 might be dangerously expanded.

    60. In a footnote to the passage about title to trust property Franks comments that mistake is not an essential allegation and adds:

    “Indeed it may be doubted whether even in a common law action to recover money paid by mistake (ie money had and received to the use of the plaintiff) the mistake is an essential allegation though it would of course in practice be pleaded: see Bullen & Leake, 3^rd ed, 45, 50; 10^th ed, 227-228.”

    This footnote may be thought to anticipate modern controversies about absence of basis in unjust enrichment. In a case like Denys v Shuckburgh 4 Y & C Ex 42 the claimant’s cause of action rests both on his antecedent title and on his mistake. If Lady Caroline’s son had known the true position throughout, but had expressly or impliedly authorised the mine manager to continue paying income to his mother, he would have had difficulty recovering the payments even within the limitation period.

    61. Doubts about Phillips-Higgins v Harper have been expressed not only by Franks and Edelman but also (in a rather more muted way) in Chitty on Contracts, 30^th ed (2008) para 28-088; Goff and Jones, The Law of Restitution, 7^th ed (2007) paras 43-004 to 43-006, and (renamed The Law of Unjust Enrichment) 8^th ed (2011) paras 31-33 to 33-36; H M McLean, “Limitation of Actions in Restitution” [1989] CLJ 472, 493-495. Professor Burrows in a note on DMG in the Court of Appeal is generally supportive of Phillips-Higgins v Harper: (2005) 121 LQR 540,

    544. In DMG in the House of Lords Lord Hoffmann and I expressed some doubts, but Lord Scott of Foscote supported Phillips-Higgins v Harper: [2007] 1 AC 558, paras 22, 91, 147. Lord Hoffmann observed (para 22):

    “The Kleinwort Benson case [1999] 2 AC 349 is recent authority for the proposition that an action for restitution of money paid under a void contract can fall within this description [‘for relief from the consequences of a mistake’]. That does not seem to me inconsistent with the existence of the mistake not being essential to the cause of action but merely one example of a case which falls within a more general principle, just as one could have (say, for the purposes of

    limitation) a category called ‘clinical negligence’ without implying that it is a cause of action different in nature from other kinds of negligence.”

    That is a reminder (and in view of current debates about “absence of basis” a timely reminder) that “cause of action” can bear different meanings, depending on the context.

    62. Having considered the matter with the benefit of much fuller argument than in DMG I have reached the clear conclusion that Phillips-Higgins v Harper was rightly decided, and that we should not seek to develop the law by broadening the interpretation of “an action for relief from the consequences of a mistake.” My reasons are essentially the same as the Court of Appeal’s. In summary, as to the statutory language, I agree with Pearson J’s view that the words have been carefully chosen, and are more precise than some formula such as “based” or “founded” on a mistake. That is an imprecise formula, and legal scholars seem to take different views as to whether it would provide a wider or a narrower test than the words of the statute. As to history, the authorities are rather short on clear exposition of the relevant principles of equity, but on the whole they provide little support for Mr Rabinowitz’s thesis. Their clearest message is the close analogy between the equitable jurisdiction and the common law action to recover money paid under a mistake.
    63. As to policy, departure from Pearson J’s relatively narrow interpretation would bring a real risk (as Franks put it, at pp 206-207) that “the scope of [section 32(1)(c)] might be expanded dangerously close to the basic rule of common law limitation that ignorance of the existence of a cause of action does not prevent time from running.” It would be difficult to find any principled stopping-place for the expansion. The leading case of Cartledge v E Jopling & Sons Ltd [1963] AC 758 (in which this point was not even faintly argued) would be seen to have missed the point. The limits (and indeed the rationale) of sections 11 and 14A of the Limitation Act 1980 would have to be revisited. Further complications would be introduced into claims for pure economic loss for breaches of professional duties of care. Any such developments are a matter for the Law Commission and for Parliament, not for this court.

    Must there be a demand?

    64. At first instance, Henderson J referred to the Woolwich principle in para 245 of his judgment and directed himself in these terms:

    “Conversely, a Woolwich claim must involve, at least in some sense, the making of a demand by the Revenue, whereas there is no need for a demand in cases of [payment under a mistake]”.

    Later in his discussion of the point he referred to the decision of the Court of Appeal in NEC [2006] STC 606, which was decided in the period between the decisions of the Court of Appeal and the House of Lords in DMG. In NEC the Court of Appeal held that since the companies in question had not made a group income election, ACT was lawfully payable, and there had been no unlawful demand (see especially the judgment of Mummery LJ at paras 152 to 162).

    65. In the present case the Court of Appeal addressed this issue at paras 152 to 174 of the judgment of the court delivered by Arden LJ. The court differed from Henderson J. It accorded great respect to the judgment of Mummery LJ in NEC but did not accept that it was a binding precedent. It also pointed out, at para 169, that Mummery LJ’s conclusion (in para 162 of his judgment) tended to elide two distinct issues, that is whether ACT was lawfully due and whether it was demanded.
    66. The Court of Appeal went on to reach a different conclusion. The heart of its reasoning is at paras 157 and 158:

    “In our judgment, the judge was wrong to reject the Revenue’s submission that Woolwich alone provides a sufficient United Kingdom remedy for the San Giorgio claims of the claimants [Amministrazione delle Finanze dello Stato v SpA San Giorgio (Case 199/82) [1983] ECR 3595 – ‘San Giorgio’]. He did so because he considered that he was bound by authority to hold that it is an essential ingredient of the Woolwich cause of action that the tax was paid pursuant to a ‘demand’. We consider that authority does not require a demand, and that it is sufficient that the state has exacted tax, which was not lawfully due, by voluntary compliance by the taxpayer with the legislative imposition of the tax.

    158. As a matter of principle, we do not see why a demand should be a requirement of a Woolwich claim. The underlying principle is that the Revenue should repay tax that has been exacted without legal justification. We can see no reason why the cause of action should be confined to those taxes that are payable on demand as against those, such as VAT, that are payable without a demand. Moreover, it is impossible to see why the citizen who duly accounts for and pays, by way of example, VAT, without waiting for a demand, on the

    assumption that the applicable legislation is valid, should be disadvantaged as against the taxpayer who refuses to account or to pay until a peremptory demand is received.”

    67. Mr Rabinowitz criticised the Court of Appeal’s reasoning and conclusion on the following grounds (in very brief summary): first, that it was contrary to binding authority, that is the decisions of the House of Lords in Woolwich and DMG; second, that it was contrary to what he described as the “conventional understanding” of Woolwich; third, that it would create uncertainty, both as to the boundaries of any extended Woolwich principle and in the general development of the law of unjust enrichment. Mr Rabinowitz also had a further, separate argument based on the Court of Appeal’s conforming interpretation of section 231 of TA 1988 (mentioned in para 33 above). This summary does not do justice to Mr Rabinowitz’s powerful written and oral submissions but it indicates their general scope.
    68. As the matter is now before the Supreme Court, sitting in a constitution of seven, it is unnecessary to embark on a lengthy consideration of the question of precedent. It is clear from paras 108 to 112 of his judgment in NEC [2006] STC 606 that Mummery LJ carefully considered whether it was appropriate for him to express opinions on issues of law that were not necessary to the decision. He reached the conclusion that, in the exceptional circumstances of the group litigation, he should take a course which he would not normally have taken, even though it resulted in judgment being reserved for a longer period. Mummery LJ’s views (with which Sedley and Lloyd LJJ agreed) do not bind this court, but they are entitled to great respect.
    69. Mr Rabinowitz’s strongest point is the frequent and consistent use of the expression “demand”, not only in the speech of Lord Goff in Woolwich, but in the speeches of the other members of the House of Lords majority in that case, and in the speeches of the House of Lords in DMG. Occasional variant uses of “exaction” carry no weight, since the two words have much the same meaning (indeed, arguably “exaction” sounds rather more coercive). Mr Rabinowitz is also right in submitting that most legal scholars have understood Woolwich and DMG as laying down that an official demand is an essential prerequisite for the principle to apply. However legal scholars have also been unanimous, or almost unanimous, in expressing the view that an official demand ought not to be a prerequisite for the application of the principle.
    70. The Law Commission in its report, Restitution: Mistakes of Law and Ultra Vires Public Authority Receipts and Payments (1994) (Law Com No 227) took the view that a demand was not necessary (paras 6.41 to 6.42):

    “Lord Goff’s reasons for the new restitutionary right, described above, also sustain these inferences, as they are based on the special position of the state and other public bodies. They do not focus on the particular requirements of a ‘demand’ or a ‘tax’; but on the manifest injustice of allowing monies unlawfully extracted from the subject by a public authority to be retained by it.

    6.42. Therefore, we believe that the principle may well be held to apply to all taxes, levies, assessments, tolls or charges, whether for the provision of services or not, collected by any person or body under a statutory provision which is the sole source of the authority to charge. We do not think that the Woolwich right is limited to payment of tax or to governmental or quasi-governmental exactions, or to payments made in accordance with a demand. We believe the crucial element is that the payment is collected by any person or body which is operating outside its statutory authority, that is, it is acting ultra vires.”

    71. The editors of Goff and Jones, The Law of Unjust Enrichment 8^th ed [2011], para 22 – 15 comment, after referring to the Court of Appeal’s judgment in NEC:

    “However, provided that a claimant’s money has been paid as tax –ie to discharge a supposed tax liability – it should make no difference in principle whether HMRC demanded the payment. After all, the Woolwich case itself was expressly fought and decided on the basis that the building society’s payment was not made in response to illegitimate pressure exerted by the Revenue, and as Bastarache J has observed in the Supreme Court of Canada ‘The right of [a claimant] to obtain restitution for taxes paid under ultra vires legislation does not depend on the behaviour of each party but on the objective consideration of whether the tax was exacted without proper legal authority.’ [Kingstreet Investments Ltd v New Brunswick (Finance) [2007] 1 SCR 3, para 53]”.

    72. Professor Jack Beatson (as he then was) expressed similar views in an article (written after the Law Commission’s Consultation Paper No 120 on Restitution of payments Made Under a Mistake of Law (1991), para 3.90-3.91 but before its Report), “Restitution of Taxes, Levies and other Imposts: Defining the Extent of the Woolwich Principle” (1993) 109 LQR 401, 405:

    “So, the formulation of the principle indicates that only two of the four features present in the Woolwich case – the demand and its ultra

    vires nature – may be necessary prerequisites. In the case of the demand even this is questionable in view of Lord Goff and Lord Slynn’s view that a payment of tax made under a mistake of law would be recoverable. The Law Commission’s Consultation Paper provisionally recommended that nothing should turn on the existence or otherwise of an actual demand for payment. Quite apart from the difficulties of distinguishing payments made in response to an ‘implied’ demand or an ‘expectation’ of payment generated by the authority (including its literature), which were mentioned, this requirement is wholly inappropriate and may pose difficulties in the context of a system based on self-assessment of tax (and other levies) such as that under consideration by the Revenue at present.”

    73. Similar views have been expressed by Professor Charles Mitchell (English Private Law, ed Burrows, 2^nd ed (2007) para 18-157); Rebecca Williams, Unjust Enrichment in Public Law (2010) pp 40-41; and Professor Burrows, The Law of Restitution 3^rd ed (2011) pp 507-508.
    74. This is a formidable volume of distinguished academic opinion. One of the main themes in the reasoning is the high constitutional importance of the principle that there should be no taxation without Parliament. As Professor Mitchell put it (English Private Law, 2^nd ed para 18.156):

    “One policy justification for the Woolwich entitlement mentioned by Lord Goff is that a general right to recover payments of tax levied without the authority of Parliament is needed to give full effect to the constitutional principle enshrined in article 4 of the Bill of Rights 1689, that the Crown and its ministers may not impose direct or indirect taxes without Parliamentary sanction. Another, latent in their Lordships’ speeches, is the related but wider public law principle of legality, that bodies invested with power by the state must respect the rule of law, and adhere to the limits of the jurisdictions conferred upon them.”

    An earlier footnote refers to two influential articles on the same theme: Professor W R Cornish, “Colour of Office: Restitutionary Redress Against Public Authority” (1987) 14 J Mal & Comp L 41, and Professor Peter Birks, “Restitution from the Executive: a Tercentenary Footnote to the Bill of Rights” in Finn (ed), Essays on Restitution (1990) 164. These were referred to by Lord Goff in Woolwich [1993] AC 70, 166.

    75. These high principles should not depend on the details of the procedure adopted for the levying and payment of any particular tax, especially in an age when (for reasons of economy and efficiency) the trend is towards self-assessment of as many taxes as possible. ACT was self-assessed, as already noted, and so was the tax which HMRC sought to charge under the ultra vires Income Tax (Building Societies) Regulations 1986 in Woolwich.
    76. It is helpful to see how the arguments developed as Woolwich proceeded through the courts. The building society was successful in judicial review proceedings decided by Nolan J on 31 July 1987. The building society had anticipated that decision by issuing a writ on 15 July 1987. Nolan J gave judgment in the action on 12 July 1988, [1989] 1 WLR 137. He felt bound by authority to dismiss the action so far as it claimed interest, holding that there was an implied agreement for repayment of any ultra vires exaction, but without interest. In his judgment Nolan J made detailed findings of fact (at pp 141-142), concluding that “the requirements of the Regulations as amplified in communications from the revenue amounted on their face to lawful demands from the Crown.”
    77. The Court of Appeal [1993] AC 70, 76-142 allowed the building society’s appeal by a majority. The majority (Glidewell and Butler-Sloss LJJ) based their decision on an ultra vires demand and a payment which was not intended to close the transaction. Ralph Gibson LJ, dissenting, held that the payment should be classified as voluntary, with an implied agreement for repayment (without interest) if tax was not due. All three members of the Court of Appeal seem to have accepted, without much discussion, Nolan J’s finding that there had been a demand. The differences between them turned on whether the building society’s response to the demand should be regarded as a voluntary payment.
    78. The matter came before the House of Lords, therefore, on the unchallenged factual basis that there had been a demand. The House was split three-two, with Lord Keith of Kinkel and Lord Jauncey of Tullichettle basing their dissents on the absence of any improper pressure or duress: [1993] AC 70, 160-161, 192-194. There was no difference between the majority and the minority as to the significance of a demand.
    79. In these circumstances it is in my view open to this court (whether or not it was strictly open to the Court of Appeal) to state clearly that where tax is purportedly charged without lawful parliamentary authority, a claim for repayment arises regardless of any official demand (unless the payment was, on the facts, made in order to close the transaction). The same effect would be produced by saying that the statutory text is itself a sufficient demand, but the simpler and more direct course is to put the matter in terms of a perceived obligation to pay, rather

    than an implicit demand. That is how it was put by Wilson J in her well known dissent in Air Canada v British Columbia (1989) 59 DLR (4^th) 161, 169:

    “It is, however, my view that payments made under unconstitutional legislation are not ‘voluntary’ in a sense which should prejudice the taxpayer. The taxpayer, assuming the validity of the statute as I believe it is entitled to do, considers itself obligated to pay. Citizens are expected to be law-abiding. They are expected to pay their taxes. Pay first and object later is the general rule. The payments are made pursuant to a perceived obligation to pay which results from the combined presumption of constitutional validity of duly enacted legislation and the holding out of such validity by the legislature. In such circumstances I consider it quite unrealistic to expect the taxpayer to make its payments ‘under protest’. Any taxpayer paying taxes exigible under a statute which it has no reason to believe or suspect is other than valid should be viewed as having paid pursuant to the statutory obligation to do so.”

    Lord Goff stated in Woolwich that he found this reasoning “most attractive.” The Supreme Court of Canada has in recent years, in a judgment of the Court delivered by Bastarache J, unanimously approved this passage from her dissenting speech: Kingstreet Investments Ltd v New Brunswick (Finance) [2007] 1 SCR 3, para 55. In my view English law should follow the same course. We should restate the Woolwich principle so as to cover all sums paid to a public authority in response to (and sufficiently causally connected with) an apparent statutory requirement to pay tax which (in fact and in law) is not lawfully due.

    80. Mr Rabinowitz argued that to follow that course would introduce uncertainty as to what amounts to a tax. The expression should in my view be generously construed, but there are bound to be borderline cases (the Foreign and Commonwealth Office is said to be engaged in a constant dialogue with foreign embassies in London as to whether the congestion charge is a tax). Borderline cases of that sort will arise whether or not a demand is needed. They would be likely to cause very much less difficulty than deciding, across the whole range of taxes of different sorts, what amounts to an official demand.
    81. Mr Rabinowitz suggested that there would also be uncertainty in the general development of the English law of unjust enrichment. There is vigorous debate among legal scholars on this topic at present, and uncertainty as to the outcome. But to decide that an official demand is not a prerequisite to a claim for the recovery of tax paid when not due ought not to add appreciably to the uncertainty. It would not be a decisive step towards a general “absence of basis” principle in place of the “unjust factors” approach that has prevailed in the past. It would

    merely be creating, in Mr Rabinowitz’s metaphor, a rather larger island of recovery in respect of undue tax.

    82. Finally, under this head, there is the argument based on the Court of Appeal’s conforming interpretation of section 231 of TA 1998. This was the Court of Appeal’s issue 6, addressed at paras 97 to 109 of its judgment. The test claimants’ argument is that section 231, on the interpretation adopted by the Court of Appeal, resulted in dividends from non-resident subsidiaries of a UK-resident company being treated as FII, so that a credit was available in the same way as for dividends received from UK-resident subsidiaries. Therefore, the argument goes, ACT was not unlawfully levied. The appropriate claim was a mistake claim, not a Woolwich claim. This is an ingenious variation on the approach described at para 39 above. The argument looks like another bit of self-inflicted harm for the test claimants, but they seek to turn it to their advantage.
    83. The tactical argument is ingenious but (even if the Court of Appeal was right in its conforming interpretation, a point which may still be revisited if permission is given for a further appeal to this court) it is in my view unsound. It seeks to rewrite history. HMRC stoutly defended its position before the Court of Justice until the judgment of the Grand Chamber at the end of 2006. Until then it consistently contended that there was nothing unlawful about the ACT/FII/FIDs regime, and it performed its statutory functions on that basis. Any suggestion that section 231 of TA 1988 did not mean what it plainly appeared to mean would have been met with incomprehension and disbelief. In short, it did not administer the taxation of UK-resident companies in accordance with any conforming interpretation. The unlawful levying of tax may depend either on the text of the statute (which was on its face discriminatory and contrary to EU law) or on how the tax is administered in practice. In this case HMRC were at fault on at least one, and possibly both of these counts.

    The central issues revisited

    84. For the reasons given in paras 42 to 82 above I consider that the Court of Appeal was correct in its conclusions (i) on section 32(1)(c) of the Limitation Act 1980 and (ii) on an official demand for tax not being a prerequisite of a Woolwich claim. The last 40 paragraphs can therefore be seen as no more than a laborious detour which ultimately leads back to the central issues in the appeal, outlined in paras 38 and 39 above: is a Woolwich claim (on its own) an adequate remedy meeting the principles of effectiveness and equivalence? Or are the test claimants also entitled to regard a claim based on mistake as one which EU law will protect against summary removal by national legislation (with the consequence that the statutory cut-off provisions infringe EU law)?
    85. The Court of Appeal answered the first of these questions in the affirmative, and the second in the negative. The relevant part of the judgment is paras 217 to

    229. The court’s reasoning is quite compressed, the heart of it being in para 225:

    “We have held, in respect of issues 11 and 12, that a demand is not an essential ingredient of the Woolwich cause of action, and that that cause of action provides an effective remedy for all the Claimants’ San Giorgio claims. Thus the cause of action for repayment of monies paid under a mistake is not a cause of action required by Community law. The cause of action for repayment of monies paid under a mistake is a domestic remedy of wide application, which Community law does not require the member states to provide, attended by a limitation period (ie section 32(1)(c) of the Limitation Act 1980) that goes beyond the requirements of Community law: see Marks & Spencer at paragraph [2003] QB 866, para 35, in which the court considered a three-year limitation period to be reasonable. Community law restricts the effectiveness of domestic legislation curtailing a limitation period applicable to a domestic cause of action that protects the Community right. That domestic cause of action is the Woolwich claim, and it is unaffected by sections 320 and 107.”

    86. Mr Aaronson has criticised this reasoning as seriously flawed. The test claimants’ written case sets out an elaborate framework of five reasons, the first and second of which have been the subject of the detour at paras 42 to 82 above. The third, fourth and fifth reasons are considered in the following sections of this judgment.

    Reemtsma

    87. Mr Aaronson relied on the decision of the Court of Justice in Reemtsma Cigarettenfabriken Gmbh v Ministero delle Finanze (Case-35/05) [2007] ECR I- 2425 as authority for the general proposition that EU law requires repayment of tax paid under a mistake (and not unlawfully exacted). In that case an Italian advertising agency had supplied services to a German client and the services were to be treated as supplied in Germany. The Italian supplier erroneously invoiced the client and paid VAT to the Italian tax authorities. Having failed to obtain a refund from the supplier, the German company brought proceedings against the Italian tax authorities. The Court of Justice held that it was not reimbursable under the provisions of the Eighth Directive and should normally be claimed from the supplier. “However”, (para 42) “where reimbursement of the VAT would become impossible or excessively difficult, the member states must provide for the instruments necessary to enable that recipient [of the relevant services] to recover the unduly invoiced tax in order to respect the principle of effectiveness.”
    88. Mr Aaronson submitted that this principle was of general application, and not limited to VAT (as a specifically EU tax). He submitted that this was a mistaken payment which was within the wide San Giorgio principle but not within the Woolwich principle, however much it might be extended. In support of his submission that it was not limited to VAT Mr Aaronson referred to Danfoss AS v Skattministeriet (Case C-94/10), 20 October 2011. Denmark imposed an indirect tax on lubricants and hydraulic oils which failed to give effect to exemptions required by article 8 of Council Directive 92/81 EEC. Danfoss purchased these products in large quantities and the suppliers passed on to Danfoss the amount of unlawfully exacted tax which they had paid. Following the judgment of the Court of Justice in Braathens Sverige AB v Riksskatteverket (Case C-346/97) [1999] ECR I-3419 Danfoss claimed reimbursement direct from the Danish authorities. The Court of Justice referred to the general San Giorgio principle by which a member state is in principle required to pay charges levied in breach of EU law. This is subject to an exception if the wrongly levied charge has been passed on. Where the tax has been passed on the ultimate consumer should normally be able to recover from his supplier, but if that is impossible or unduly difficult there must be a remedy in the form of a direct claim against the tax authorities. Reemtsma was referred to as an authority for this proposition.
    89. Lord Sumption regards this principle as limited to harmonised EU taxes, and I am inclined to agree with that. But in any event it applies to a different and relatively unusual situation, in which it is a third party, and not the original taxpayer, who is seeking to recover tax from the authorities. It does not assist the test claimants in this appeal.

    EU law’s requirements as to national remedies (especially limitation periods)

    90. There is no doubt as to the general principles regulating what EU law requires of national remedies for infringements of EU law. The principles were stated by the Grand Chamber in its judgment on the first reference in these proceedings, Case C-446/04, paras 201 to 203, in terms identical, or almost identical, to those which have been stated many times before by the Court of Justice:

    “It must be stated that it is not for the court to assign a legal classification to the actions brought before the national court by the claimants in the main proceedings. In the circumstances, it is for the latter to specify the nature and basis of their actions (whether they are actions for repayment or actions for compensation for damage), subject to the supervision of the national court (see [Metallgesellschaft (Joined Cases C-397/98 and C-410/98) [2001] ECR I-1727], para 18.

    202. However, the fact remains that, according to established case law, the right to a refund of charges levied in a member state in breach of rules of Community law is the consequence and complement of the rights conferred on individuals by Community provisions as interpreted by the court (see, inter alia San Giorgio (Case C-199/82) [1983] ECR 3595, para 12, and Metallgesellschaft, para 84).   The member state is therefore required in principle to repay charges levied in breach of Community law Comateb (Joined Cases C-192/95 to C-218/95) [1997] ECR I-165, para 20, and Metallgesellschaft, para 84).
    203. In the absence of Community rules on the refund of national charges levied though not due, it is for the domestic legal system of each member state to designate the courts and tribunals having jurisdiction and to lay down the detailed procedural rules governing actions for safeguarding rights which individuals derive from Community law, provided, first, that such rules are not less favourable than those governing similar domestic actions (principle of equivalence) and, secondly, that they do not render virtually impossible or excessively difficult the exercise of rights conferred by Community law (principle of effectiveness) (see, inter alia, Rewe (Case C-33/76) [1976] ECR 1989, para 5, and Comet (Case C-45/76) [1976] ECR 2043, paras 13 and 16; and, more recently, Edis (Case C-231/96) [1998] ECR I-4951, paras 19 and 34; Dilexport (Case C- 343/96) [1999] ECR I-579, para 25; and Metallgesellschaft, para 85).”
    91. This brings us to the fourth and fifth reasons in the test claimants’ written case, which go to the heart of this appeal. They contend that in using the mistake cause of action to vindicate their EU rights they were unquestionably entitled to the protection of EU law. They criticise the Court of Appeal for having asked the wrong question: that is for having asked which domestic remedies give effect to the San Giorgio principle, rather than considering, as they should have done, all national remedies as available for the purpose.
    92. It is not necessary to multiply references to the general principles, which are not in dispute. It is however necessary to look more closely at the attitude of EU law towards limitation of actions under the legal systems of different member states, and towards legislative measures taken by member states to curtail limitation periods, so far as they affect national remedies for breaches of EU law.
    93. It is well established that EU law has no general objection to limitation periods being provided for in the legal systems of member states. On the contrary,

    limitation periods are one manifestation of the principle of legal certainty. As long ago as Rewe I (Rewe-Zentralfinanz eG v Landwirtschaftskammer fur das Saarland (Case C-33/76) [1976] ECR 1989, para 5, the Court of Justice (after referring to the general principle of national courts acting in accordance with national rules) observed:

    “The position would be different only if the conditions and time- limits made it impossible in practice to exercise the rights which the national courts are obliged to protect. This is not the case where reasonable periods of limitation of actions are fixed. The laying down of such time-limits with regard to actions of a fiscal nature is an application of the fundamental principle of legal certainty protecting both the taxpayer and the administration concerned.”

    There is a similar statement, again expressly linked to fiscal proceedings, in Comet BV v Produktschap voor Siergewassen (Case C-45/76) [1967] ECR 2043, para 18.

    94. Limitation periods must be reasonable, but the Court of Justice recognises that national systems vary a good deal, and accepts different approaches so long as there is no infringement of the principles of effectiveness and equivalence, and no disappointment of legitimate expectations. This is made clear in Amministrazione delle Finanze dello Stato v Sas MIRECO (Case C-826/79) [1980] ECR 2559, paras 11 to 13, and other cases of the same vintage involving the Italian tax authorities, including Amministrazione delle Finanze dello Stato v Denkavit Italiana Srl (Case C-61/79) [1980] ECR 1205, paras 23 and 24, and Amministrazione delle Finanze dello Stato v Ariete SpA (Case C-811/79) [1980] ECR 2545, paras 10 and 11.
    95. In line with that approach, in Haahr Petroleum v Abenrå Havn (Case C- 90/94) [1997] ECR I-4085, a five-year period was accepted as reasonable for reimbursement of an unlawful goods duty. Emmott v Minister for Social Welfare (Case C-208/90) [1993] ICR 8 was distinguished ([1997] ECR I-4085, para 52) because in that case the relevant directive had not been properly transposed, and until its proper transposition time was not to start to run. In Edilizia Industriale Siderurgica Srl v Ministero delle Finanze (Case C-23/1996) [1998] ECR I-4951 a three-year period was accepted for recovery of company registration charges levied in breach of article 10 of Council Directive 69/335/EEC despite the fact that the normal limitation period for restitution, under article 2946 of the Italian Civil Code, was ten years.
    96. The principles of effectiveness, equivalence and legitimate expectation also apply if a national legislature enacts a measure to curtail an existing limitation period, especially if the measure appears to be directed at a particular ruling of the

    Court of Justice. The leading authority is the first judgment of the Court of Justice in Marks & Spencer Plc v Customs and Excise Comrs (Case C-62/00) [2003] QB 866 (“M&S”). That litigation was complicated and protracted, involving as it did two distinct claims for repayment of VAT (one concerning gift vouchers, and the other concerning chocolate-covered marshmallow teacakes) which were linked together as a matter of case management. There were two references to the Court of Justice, the first of which attracted criticism from the court because of its restricted scope. The final chapter in the saga is reported at [2009] UKHL 8, [2009] STC 452. For present purposes, however, it is sufficient to note that section 47 of the Finance Act 1997 curtailed the period for a claim for repayment of VAT from six to three years, with retrospective effect, and without any period of grace. Some of the claimant’s claims for VAT on teacakes (which were properly treated as zero-rated) went back to 1973.

    97. The Advocate General (Geelhoed) referred to a summary ([2003] QB 866, para 54) of the EU jurisprudence in Roquette Frères SA v Direction des Services Fiscaux du Pas-de-Calais (Case C-88/99) [2000] ECR I-10465, para 20. He also cited at para 57, Dilexport (Case C-343/96) [1999] ECR I-579, para 43:

    “Community law does not preclude the adoption by a Member State, following judgments of the Court declaring duties or charges to be contrary to Community law, of provisions which render the conditions for repayment applicable to those duties and charges less favourable than those which would otherwise have been applied, provided that the duties and charges in question are not specifically targeted by that amendment and the new provisions do not make it impossible or excessively difficult to exercise the right to repayment.”

    The Advocate General pointed out (para 58) that the retrospective alterations to the Value Added Tax Act 1994 affected “not only taxable persons who expected under the existing rules to have ample time to make their claims but even taxable persons who before the date on which the announcement of a change in the law was made (18 July 1996) or prior to the date on which it was enacted (19 March 1997) had made claims for repayment of unduly levied tax.”

    98. The issue of specific targeting was raised at first instance, but in view of the conclusions which he had already reached Henderson J preferred to express no view on it ([2009] STC 254, paras 428 to 431). His reasons included the difficulty of the constitutional issues which would arise in inquiring into the legislative intention behind the amending legislation. The point was not raised in the Court of Appeal or in this court.
    99. The Court of Justice reached conclusions similar to those of the Advocate General [2003] QB 866, paras 36 to 38:

    “Moreover, it is clear from Aprile [2001] 1 WLR 126, para 28 and Dilexport [1999] ECR I-579 paras 41 and 42 that national legislation curtailing the period within which recovery may be sought of sums charged in breach of Community law is, subject to certain conditions, compatible with Community law. First, it must not be intended specifically to limit the consequences of a judgment of the Court to the effect that national legislation concerning a specific tax is incompatible with Community law. Secondly, the time set for its application must be sufficient to ensure that the right to repayment is effective. In that connection, the court has held that legislation which is not in fact retrospective in scope complies with that condition.

    37. It is plain, however, that that condition is not satisfied by national legislation such as that at issue in the main proceedings which reduces from six to three years the period within which repayment may be sought of VAT wrongly paid, by providing that the new time limit is to apply immediately to all claims made after the date of enactment of that legislation and to claims made between that date and an earlier date, being that of the entry into force of the legislation, as well as to claims for repayment made before the date of entry into force which are still pending on that date.
    38. Whilst national legislation reducing the period within which repayment of sums collected in breach of Community law may be sought is not incompatible with the principle of effectiveness, it is subject to the condition not only that the new limitation period is reasonable but also that the new legislation includes transitional arrangements allowing an adequate period after the enactment of the legislation for lodging the claims for repayment which persons were entitled to submit under the original legislation. Such transitional arrangements are necessary where the immediate application to those claims of a limitation period shorter than that which was previously in force would have the effect of retroactively depriving some individuals of their right to repayment, or of allowing them too short a period for asserting that right.”

    The Court of Justice held the amending legislation incompatible with the principle of effectiveness. It also (paras 45 and 46) held that it was precluded by the principle of the protection of legitimate expectations.

    Legitimate expectations

    100. The principle of protection of legitimate expectations is closely linked to the principle of legality. But in the opinion of the Advocate General (Cosmas) in Duff v Minister for Agriculture and Food, Ireland and Attorney General (Case C- 63/93) [1996] ECR I-569, para 23, the two are not interchangeable.
    101. The Advocate General’s opinion contains (at paras 24 and 25) a passage about timing which is of particular interest (his emphasis):

    “24.     Particularly for the individual the principle of legality would

    in many ways lose its significance as a guarantee of a sphere of freedom, if the temporal succession of legal provisions concerning him was not governed by an elementary consistency and coherence sufficient to enable him to discern the consequences (legal and financial) of his activities.

    25. Thus the principle of legal certainty calls for clarity and accuracy in framing the rules of law, and the individual provisions giving effect to them, which at a given moment in time constitute the legal framework within which the competences of the institutions are exercised and the activities of individuals are carried on. The principle of the protection of legitimate expectations requires the Community legislature and the other Community organs (or the national authorities operating under provisions of Community law) to exercise their powers over a period of time in such a way that situations and relationships lawfully created under Community law are not affected in a manner which could not have been foreseen by a diligent person.”

    This approach was not in terms adopted by the Court of Justice, but para 20 of its judgment appears to be in line with it.

    102. I have quoted this passage at some length because it seems to me to touch on what is, if I may respectfully say so, one of the crucial points in Lord Sumption’s judgment. Lord Sumption ultimately bases his conclusions, on the central issue, on the principle of protection of legitimate expectations (paras 198 to 202). He observes (para 196) that the right of the test claimants to choose from a range of causes of action is a right derived solely from English procedural law and (echoing the Court of Appeal, para 226) that it exists only to the extent that English law so provides. I have considerable difficulty in reconciling that with the

    principles stated by the Advocate General and the Court of Justice in M&S [2003] QB 866. But before addressing that difficulty I should recapitulate the sequence of events in which the statutory cut-off provisions were announced and enacted.

    The enactment of the statutory cut-off provisions

    103. Mr Aaronson provided a useful summary of the key dates. The first two are the decisions of the House of Lords in Woolwich (20 July 1992) and Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349 (29 October 1998). After 1998 English lawyers knew that the recovery of money paid under a mistake of law (perhaps including a mistake of tax law, subject to arguments on exclusive remedies) had become a real possibility, although it was by no means a firmly established cause of action. But until the decision of the Court of Justice in Metallgesellschaft (Joined Cases C-397/98 and C-410/98) [2001] Ch 620 on 8 March 2001 there was no general appreciation that the UK corporation tax regime was seriously open to challenge as infringing the Treaty.   Henderson J did not make any detailed findings about this, since the principle of legitimate expectations does not seem to have been argued as a separate issue before him. But he did (para 267) make a general finding of fact about mistake:

    “The unlawful payments of ACT made from 1973 to 1999, and the unlawful payments of ACT made under the FID regime from 1994 to 1999, were in my view plainly made under a mistake about the lawfulness of the tax regimes under which they were paid. I am satisfied from the evidence, both written and oral, that this was not obvious to anybody within the BAT group at the time, since everybody proceeded on the footing that the tax in question was lawfully due and payable.”

    104. After 8 March 2001 a well-advised multinational group based in the UK would have had good grounds for supposing that it had a valid claim to recover ACT levied contrary to EU law, with at least a reasonable prospect that the running of time could be postponed until then (but not subsequently) by the operation of section 32(1)(c) of the Limitation Act 1980. During 2002 the opinion of the Advocate General and the judgment of the Court of Justice in M&S, while possibly not adding much to the earlier jurisprudence, spelled out very clearly, for UK companies and lawyers, both the capacity and the limits of national legislation in curtailing limitation periods in proceedings for recovery of tax levied in breach of EU law.
    105. The next important date was 18 July 2003, when Park J gave his first- instance judgment in DMG. This was the first judicial decision which positively

    upheld a claim for repayment of unduly levied tax with an extended limitation period under section 32(1)(c). But appeals to the Court of Appeal and the House of Lords were to follow and (as Henderson J observed, para 406), “the outcome of those appeals was, at the time, impossible to predict with any confidence.”

    106. The BAT group started its proceedings on 18 June 2003, a month before Park J’s judgment in DMG. On 8 September 2003 the Paymaster General announced the introduction of retrospective legislation affecting proceedings to recover tax on the ground of mistake if the proceedings were issued on or after that day (the scope of the proposed legislation was later extended to include amendment of existing proceedings). The Aegis group issued its proceedings on that very day, 8 September 2003, and so was one of the very first claimants to be affected by the legislation.
    107. Section 320 of the Finance Act 2004 was enacted on 24 June 2004. Its essential provisions were set out by Henderson J (para 408):

    “Exclusion of extended limitation period in England, Wales and Northern Ireland

    1. Section 32(1)(c) of the Limitation Act 1980 . . . (extended period for bringing an action in case of mistake) does not apply in relation to a mistake of law relating to a taxation matter under the care and management of the Commissioners of Inland Revenue.

    This subsection has effect in relation to actions brought on or after 8 September 2003.

    2. For the purposes of–
    1. section 35(5)(a) of the Limitation Act 1980 . . . (circumstances in which time-barred claim may be brought in course of existing action), and
    2. rules of court . . . having effect for the purposes of those provisions,

    as they apply to claims in respect of mistakes of the kind mentioned in subsection (1), a new claim shall not be regarded as arising out of

    the same facts, or substantially the same facts, if it is brought in respect of a different payment, transaction period or other matter.

    This subsection has effect in relation to claims made on or after 20 November 2003.

    . . .

    6. The provisions of this section apply to any action or claim for relief from the consequences of a mistake of law, whether expressed to be brought on the ground of mistake or on some other ground (such as unlawful demand or ultra vires act).
    7. This section shall be construed as one with the Limitation Act 1980 . . .”
    108. The Court of Appeal gave judgment in DMG, reversing Park J, on 4 February 2005: [2006] Ch 243. Mr Aaronson described the Court of Appeal’s decision as a “bump in the road”, suggesting that it was unforeseen and soon forgotten, but that seems an inappropriate description, even with hindsight. Reference to the judgments (running to nearly 300 paragraphs in all) shows that numerous issues were fiercely contested, including the date of the mistakes discovery (which occurred, on HMRC’s argument, in 1995). The Court of Appeal’s decision was reversed by the House of Lords on 25 October 2006: [2007] 1 AC 558. Shortly afterwards the UK government applied to the Court of Justice for the reopening of the hearing of the first reference in these proceedings so that the United Kingdom could argue for a temporal restriction to the judgment of the Court of Justice. That application was rejected on 6 December 2006, and on the same day HMRC announced the introduction of further retrospective legislation.
    109. This was enacted on 19 July 2007 as section 107 of the Finance Act 2007. The essential terms of the section were set out by Henderson J (para 412):

    “Limitation period in old actions for mistake of law relating to direct tax

    1. Section 32(1)(c) of the Limitation Act 1980 . . . (extended period for bringing action in case of mistake) does not apply in relation to any action brought before 8 September 2003 for relief from the

    consequences of a mistake of law relating to a taxation matter under the care and management of the Commissioners of Inland Revenue.

    2. Subsection (1) has effect regardless of how the grounds on which the action was brought were expressed and of whether it was also brought otherwise than for such relief.
    3. But subsection (1) does not have effect in relation to an action, or so much of an action as relates to a cause of action, if–
    1. the action, or cause of action, has been the subject of a judgment of the House of Lords given before 6 December 2006 as to the application of section 32(1)(c) in relation to such relief, or
    2. the parties to the action are, in accordance with a group litigation order, bound in relation to the action, or cause of action, by a judgment of the House of Lords in another action given before that date as to the application of section 32(1)(c) in relation to such relief.
    4. If the judgment of any court was given on or after 6 December 2006 but before the day on which this Act is passed the judgment is to be taken to have been what it would have been had subsections (1) to (3) been in force at all times since the action was brought (and any defence of limitation which would have been available had been raised).

    . . .

    (6) In this section–

    ‘group litigation order’ means an order of a court providing for the case management of actions which give rise to common or related issues of fact or law . .

    .”

    110. On 30 September 2010 the European Commission announced that it had made a formal request to the UK to change section 107 of the Finance Act 2007.

    On 26 January 2012 there was a further announcement that the European Commission has referred the UK to the Court of Justice because of the absence of proper transitional rules in section 107.

    Discussion of the statutory cut-off provisions

    111. These provisions were challenged in the lower courts primarily on the ground that they infringed the principle of effectiveness. There was little discussion of legitimate expectations. Lord Sumption holds (para 199) that reasonable persons in the position of the test claimants would not, until Park J’s judgment in DMG on 18 July 2003, have counted on being able to recover tax on the ground of mistake of law; and that even after that decision the existence of such a claim was being challenged on serious grounds. He concludes from that proposition that no one in the position of the test claimants could have had a reasonable and realistic expectation of recovering tax on the ground of mistake.
    112. I cannot disagree with that conclusion. The issue of legitimate expectations was not raised before the judge, and he made no findings on it. The issue of reasonable expectations must of course be decided objectively, but it would have been helpful to have had the view of the judge who very carefully considered the whole case. But in any case I do have great difficulty in applying the same reasoning to upholding the validity of section 320 against attack under the principle of effectiveness, in the light of M&S. The judgment of the Court of Justice in that case lays down a clear requirement for transitional provisions, and that requirement is derived at least as much from the principle of effectiveness and the principle of legality as from the more limited principle of protection of legitimate expectations (as Advocate General Cosmas said in Duff (Case C-63/93) [1996] ECR I-569, para 23, they are not interchangeable).
    113. If one asks what the test claimants were entitled to, and what they could expect to continue to be entitled to, in the way of national remedies to recover tax levied and paid contrary to EU law, the answer is plainly not that they were entitled to the indefinite continuation of a range of alternative remedies. The passage from Rewe II on which the test claimants rely (Rewe-Handelsgellschaft Nord mbH v Haupzollamt Kiel (Case C-158/80) [1981] ECR 1805, para 44) is, as Lord Sumption demonstrates, an example of the operation of the principle of equivalence. It is not applicable in this case because both of the statutory cut-off provisions applied to all claims for repayment of direct tax, whether or not the repayment was claimed because of an infringement of EU law.
    114. Nor were the test claimants entitled to a remedy arrived at by some precise formula furnished by EU law. That would be contrary to the basic principles laid

    down in Rewe I (Case C-33/76) [1976] ECR 1989, and repeated in countless cases since then. What they were entitled to was that national law should provide an effective remedy which met the requirements of EU principles of effectiveness and equivalence; and that any curtailment of any relevant limitation period should comply with those principles, as well as with the principle of legitimate expectations. The fact that they could not have complained, in another parallel universe in which section 32 (1)(c) had never existed, is not decisive on the issue of effectiveness.

    115. I would therefore hold that section 320 was contrary to EU law as infringing the principle of effectiveness as explained in M&S, and that section 107 was contrary to EU law both on that ground and (in agreement with Lord Sumption) under the principle of protecting legitimate expectations. Examples can be tendentious, but the drastic way in which section 320 could operate can be illustrated by the example of a UK-resident holding company, part of a multinational group, which paid ACT from 1973 to 1996, building up an ever- increasing surplus of unused ACT, and then (three years before the repeal of ACT) decided that enough was enough, and disposed of its overseas subsidiaries.   In 2001 it would have learned of the possibility of a claim for repayment of tax, and taken advice as to the wisdom of incurring costs by making a claim, which was still doubtful, at some time during the next six years. In 2002 M & S (Case C- 62/00) [2003] QB 866 seemed to confirm that the law would not be changed retrospectively and without reasonable notice. But if the company did not act before 8 September 2003 it would have been deprived, retrospectively and without any notice, of the entirety of its claims for over 20 years’ tax.

    Section 33 of the Taxes Management Act 1970

    116. The last substantive point to be considered is section 33 of the Taxes Management Act 1970, which provided a statutory right to repayment of tax paid by mistake, subject to a number of restrictive conditions. It replaced provisions originally introduced by the Finance Act 1923. It has since been replaced by two different sets of provisions, one applicable to individuals and the other to companies. In the form in which it was in force at the relevant time the conditions were (1) it applied only to excessive tax charged by an assessment (which meant, Lord Goff stated in Woolwich [1993] AC 70, 169, a valid assessment) as a result of an error or mistake in a return; (2) there was a six-year time limit; (3) there was to be no repayment if the erroneous or mistaken return was in accordance with practice generally prevailing at the time; and (4) the repayment was to be such as the Board of Inland Revenue (subject to a possible appeal to the Special Commissioners) considered reasonable and just. The flexibility of the last condition was explained by Mr Ewart by the example of a taxpayer who had paid too much tax six years before, but who ought to have paid more tax on the same income seven or more years before.
    117. The issue on section 33 is whether it is an obstacle to the test claimants and if so, whether it can be given a conforming interpretation under the Marleasing principle², besides reducing the basic limitation period for tax repayment claims from six to three years, in addition eliminated the special advantage for claims in mistake previously introduced by section 24(5) of the Finance Act 1989, delaying the commencement of the limitation period for such claims until the claimants had actually or constructively discovered the mistake – this being the provision held ineffective by the Court of Justice in Marks & Spencer; and, on the other hand, section 320 with which this court is now concerned which (similarly with effect from when government first announced its intention so to legislate) similarly eliminates with regard to tax repayment claims based on a mistake of law the similar special provision enlarging the limitation period to be found in section 32(1)(c) of the Limitation Act 1980.
    123. Now, however, I am inclined to accept Lord Sumption’s view that, by the same token that, on the facts of this case, the appellants can establish no legitimate expectation at any time prior to 8 September 2003 (when government announced its intention to introduce section 320) that the limitation period for mistake of law

    tax repayment claims would not be attenuated by legislation, nor can they make good their argument that section 320 infringes the EU principle of effectiveness. The self-same considerations – essentially of fairness and legal certainty – which underlie the doctrine of legitimate expectation (both domestically and under EU law) to my mind also inform the principle of effectiveness. If, as seems to me plainly to be so, the situation even after Park J’s first-instance decision in Deutsche Morgan-Grenfell Group Plc v Inland Revenue Comrs [2003] 4 All ER 645 (“DMG”) was one of complete uncertainty as to whether tax could be re-claimed on the basis of a mistake of law – there being at least as much room for a mistake of law as to this as for the mistake of law which the majority of the House of Lords in DMG [2007] 1 AC 558 held the taxpayers to remain under until the Court of Justice’s final authoritative decision in the Hoechst case (Metallgesellschaft Ltd v Inland Revenue Comrs (Joined Cases C-397/98 and C-410/98) [2001] Ch 620) – there was to my mind neither unfairness nor any denial of a legitimate expectation from Parliament stepping in to legislate with immediate effect to clarify the situation (albeit to the taxpayers’ obvious disadvantage given that the common law was finally to be developed in their favour).

    124. In short, whereas the position as to limitation with regard to tax recovery claims was crystal clear under section 24 of the 1989 Act – and could not therefore fairly and legitimately be altered without due notice and appropriate transitional provisions – it was entirely unclear under the developing common law when Parliament chose to intervene by the enactment of section 320.   And it is that which provides the principled basis for distinguishing this case from Marks & Spencer.
    125. Section 107, by contrast, is not merely overtly retrospective (eliminating pre-existing claims explicitly preserved by section 320), but was introduced after the House of Lords decision in DMG finally resolved the uncertainty in the law and proclaimed (albeit subject always to lawful legislative change) that mistake of law claims with their extended limitation periods were indeed available to those seeking recoupment of overpaid (or prematurely paid) tax. Small wonder that it is section 107 that the Commission selected for attack.

    LORD CLARKE

    126. In para 9 above Lord Hope has conveniently identified four issues for determination in this appeal. Issues (3) and (4), which raise a question of construction of section 32(1)(c) of the Limitation Act 1980 and the ingredients of the common law Woolwich claim respectively, raise no issue of EC law. I agree with the other members of the court that, for the reasons they give, the decisions of the Court of Appeal on both questions should be upheld and that both questions

    should be answered no. At the end of the argument I was inclined to the view that section 32(1)(c) should be given the wider meaning contended for by the Test Claimants, but I have been persuaded by the reasoning of Lord Walker and Lord Sumption that it should not.

    127. I also agree with the other members of the court that the restitution and damages remedies sought by the Test Claimants are not excluded by section 33 of the Taxes Management Act 1970 and that it follows that question (2) must be answered no and that the Test Claimants’ appeal on this issue must succeed. This seems to me to be essentially a matter of construction of section 33. In so far as it involves an issue of EU law, I would hold that it is acte clair, and would not refer it to the Court of Justice.
    128. By contrast, the questions posed by issue (1) raise difficult questions of EU law. This is evident from the differences of opinion between members of the court. A comparison between the judgments in this case shows that the members of the court are divided, not only as to the question whether EU law protects the mistake claims and, in particular, whether section 320 of the Finance Act 2004 infringes the EU law principles of effectiveness, legal certainty and legitimate expectation, but also as to the correct reasoning for the conclusions reached. I too would refer the section 320 issues to the Court of Justice.
    129. If there is to be a reference, any further analysis of the position by me will be largely, if not entirely, redundant, since all will depend upon the conclusions ultimately reached by the Court of Justice. I will therefore only add this. I agree that section 107 infringes EC law for the reasons given by Lord Sumption. As to section 320, in agreement with Lord Hope, Lord Walker, Lord Dyson, and Lord Reed my provisional view is that the appeal should be allowed.
    130. The problem (or potential problem) facing the Test Claimants is that English law provides two remedies for their claim that tax has been exacted from them contrary to EU law. If the only available remedy were the mistake claim, the position would be clear. It would fall within the principle in Marks & Spencer Plc v Customs and Excise Comrs (Case C-62/00) [2003] QB 866 (“M&S”), which is discussed in some detail by Lord Walker at paras 96 to 99. The principle is summarised both by the Advocate General and by the Court of Justice at paras 36 to 38 (quoted at para 99 above). It applies in respect of national legislation curtailing the period within which recovery may be sought of sums charged in breach of EU law and may be summarised as follows: (1) such legislation must not be intended specifically to limit the consequences of a judgment of the Court of Justice to the effect that national legislation concerning a specific tax is incompatible with EU law; (2) the time set for its application must be sufficient to ensure that the right to repayment is effective; and (3) where a new limitation

    period limits the previously permitted period, the new period must be reasonable and the new legislation must include transitional arrangements allowing an adequate period for lodging claims which were available under the previous legislation.

    131. As Lord Walker explains at para 104, after 8 March 2001, when the Court of Justice decided Metallgesellschaft Ltd v Inland Revenue Comrs (Joined Cases C-397/98 and C-410/98) [2001] Ch 620, the Test Claimants would have had good grounds for supposing that they had a good claim to recover ACT levied contrary to EU law, with at least a reasonable prospect that the running of time could be postponed until then by section 32(1)(c) of the Limitation Act 1980. In so far as proceedings had not been issued, their claims were therefore in time as at 8 September 2003 when HMRC announced the introduction of what became section 320 of the Finance Act 2004.   The effect of section 320, which is set out at para 107 above and was enacted on 24 June 2004, was to deprive those Test Claimants of rights which were available to them by reason of section 32(1)(c) without any transitional provisions to allow them to issue proceedings within a reasonable time.
    132. But for the availability of the Woolwich claim, section 320 would therefore be contrary to the principles clearly set out in M&S. It made it impossible for those Test Claimants to proceed with their mistake claim because of the absence of the introduction of a reasonable period of limitation as from then and because of the complete absence of transitional provisions. The facts are very similar to those relating to the chocolate covered marshmallow teacakes in M&S.
    133. Does the existence of the Woolwich remedy make all the difference? I agree with Lord Hope, Lord Walker, Lord Reed and Lord Dyson that it does not. To my mind it would be remarkable if it did. In this regard, I agree in particular with the reasoning of Lord Hope at paras 16 to 19 above. As Lord Hope shows, the Test Claimants had every prospect of success. It is plain from the fact that section 320 was enacted that HMRC shared that view, since (at any rate as it seems to me) the whole point of the section was to ensure that such a claim would not succeed. In any event, in the period before section 320 came into force the Test Claimants were entitled to have their mistake claim adjudicated upon by the English courts. In my opinion they had a legitimate expectation that, as Lord Hope puts it at para 19 and Lord Reed puts it at para 243, that entitlement would not be removed from them by the introduction without notice of a limitation period that was not fixed in advance.
    134. Before the decision in Deutsche Morgan Grenfell Group Plc v Inland Revenue Comrs [2007] 1 AC 558 (“DMG”) the Test Claimants knew that there was a reasonable prospect that they had a good mistake of law claim against the Revenue. I agree with Lord Sumption (at para 201) that it must be relevant to ask

    on what basis the Test Claimants must be taken to have made their plans and that the issue is whether there is an assumption reasonably to be attributed to them about how long they had to bring their claims, which was then retrospectively falsified by Parliament. It seems to me that they can reasonably be taken to have made their plans on the basis of an expectation that the State would not remove their rights without warning or transitional provisions. That expectation was then retrospectively falsified by section 320.

    135. In all these circumstances, I prefer the reasoning of Lord Hope and Lord Reed to that of Lord Sumption. It follows that in my opinion section 320 infringes their rights under EC law on the ground that it infringes the principle of legitimate expectation.
    136. In addition I agree with Lord Hope, Lord Walker, Lord Dyson and Lord Reed that an application of the principle of effectiveness also leads to the conclusion that section 320 infringes their rights under EC law. This part of the case has been analysed in some detail by Lord Hope, Lord Walker and Lord Reed. In particular, Lord Reed’s analysis is considerably more extensive than that of Lord Walker. As I read Lord Reed’s judgment, a critical part of his reasoning is his reliance upon his view of the principles of equivalence, which he then deploys in reaching his conclusion that section 320 infringes the principle of effectiveness. His reasoning is to my mind convincing and, for the reasons he gives, I too would so hold.
    137. I have a slight concern that so to hold is to determine the issue on a basis which was not advanced in argument on behalf of the Test Claimants, which (to put it no higher) is surprising given the many years they have been considering these issues. However, if the Court of Justice were to prefer the approach to equivalence adopted by Lord Sumption to that adopted by Lord Reed, I would nevertheless hold that section 320 infringes the principle of effectiveness. Although there is, so far as I am aware, no decision of the Court of Justice which directly addresses the point, this conclusion seems to me to receive some support from the opinion of Advocate General Sharpston in Unibet (London) Ltd v Justitiekanslern (Case C-432/05) [2008] All ER (EC) 453, where she said this at para 32 of her opinion:

    “The starting point to my mind must be the principle, first laid down in Rewe I [(Case 33/76) [1976] ECR 1989, para 5], that it is for the domestic legal system of each member state to determine the procedural conditions governing actions at law intended to ensure the protection of Community law rights, provided that those conditions are not less favourable than those relating to similar actions of a domestic nature (principle of equivalence) and do not

    make it impossible in practice to exercise those rights (principle of effectiveness). That approach was confirmed in Rewe II [(Case 158/80] [1981] ECR 1805, para 44], where the court stated that the Treaty was not intended to create new remedies in the national courts to ensure the observance of Community law other than those already laid down by national law and that the system of legal protection established by the Treaty implies that it must be possible for every type of action provided for by national law to be available for the purpose of ensuring observance of Community provisions having direct effect.” (Original emphasis)

    138. I recognise that, as Lord Sumption observes at para 194, Rewe- Handelsgellschaft Nord mbH v Haupzollamt Kiel [1981] ECR 1805 was an equivalence case and that the Court of Justice did not expressly comment upon this passage, but it nevertheless seems to me that in her para 32 the Advocate General was putting the point more generally in the context of effectiveness and that, in that context it provides some support for the Test Claimants’ case.
    139. I appreciate that the views that I (and others) have expressed on the section 320 point can only be provisional and that it will ultimately be resolved in the light of the answers to the questions referred to the Court of Justice. I nevertheless hope that these views will be of some assistance in the formulation of those questions.

    LORD DYSON

    140. I too agree with the judgments of Lord Walker and Lord Sumption on all the issues on which they agree. Like Lord Hope and Lord Reed, I agree with Lord Walker on the DMG/section 320 issue. Nevertheless, I acknowledge the force of Lord Sumption’s reasoning on this issue. For that reason I have concluded that the question cannot be regarded as acte clair and that a reference to the European Court of Justice is necessary.

    LORD SUMPTION

    Introduction

    141. It is not in dispute that under EU Law, the United Kingdom is bound to provide an effective means under its national law of recovering tax charged contrary to the EU Treaty. It is common ground that it is open to member states to

    impose reasonable periods of limitation, even on actions to enforce directly effective EU law rights. It is also common ground that six years is a reasonable period of limitation for an action to recover tax charged contrary to EU law, and that if English law had always provided for the period to run from the date of payment in cases of mistake, then that too would have been reasonable. Broadly stated, the issue on this appeal is whether the United Kingdom was entitled to change the law relating to the running of the limitation period, without notice or transitional provisions for actions which were pending or in the pipeline. The commissioners say that the change related only to actions to recover tax paid under a mistake of law and that there are other causes of action unaffected by the change which satisfy the United Kingdom’s obligation to provide an effective means of recovering the tax. The Test Claimants say, in bald summary, (i) that every cause of action available to them for   common law restitution is, on analysis, an action for relief against the consequences of a mistake and therefore affected by the change, (ii) that so far as there are other causes of action available to them which are not affected by the change, they are subject to legal limitations which make it impossible to regard them as an effective means of recovery, and (iii) that irrespective of the fate of points (i) and (ii) the United Kingdom was not entitled to curtail, without notice or transitional provisions, the availability of any cause of action which might serve their purpose.

    142. In my judgment, the Test Claimants and other companies in their position have an effective means of recovering the overpaid tax under the principle stated by the House of Lords in the landmark decision in Woolwich Equitable Building Society v Inland Revenue Commrs [1993] AC 70. The availability of that cause of action entirely satisfies the obligations of the United Kingdom under the EU Treaty, notwithstanding that it is subject to a limitation period which runs from the date of payment. Neither section 320 of the Finance Act 2004 nor section 107 of the Finance Act 2007 had any impact on a claim made on that basis, because both enactments were concerned only with actions for the recovery of tax paid under a mistake of law. Mistake of law is a more limited cause of action, which is neither necessary nor sufficient to satisfy the obligations of the United Kingdom under the EU Treaty. In those circumstances, I consider that the validity of those enactments depends entirely on whether they defeated the legitimate expectations of taxpayers as that concept is understood in EU law. I do not think that section 320 of the Finance Act 2004 can be criticised on that ground. Its effect was that the limitation period for an action to recover tax paid under a mistake of law was to run from the date of payment in the same way as the limitation period for an action to recover tax on any other ground. It was announced almost as soon as the existence of a right to recover tax paid under a mistake of law had been judicially recognised. It follows that taxpayers in the position of these claimants cannot at the relevant time have had any reasonable expectation that a cause of action to recover tax paid under a mistake of law would be available to them. For that reason, I think that they would suffer no injustice if section 320 of the 2004 Act were to be given effect according to its terms, whereas a significant injustice would be suffered by

    the general body of taxpayers if it were not. Different considerations apply to section 107 of the 2007 Act, which was retrospective in an altogether more radical and objectionable sense. It does not surprise me that the European Commission has referred the enactment of 2007 to the European Court of Justice, but has taken no comparable step in the case of the enactment of 2004.

    143. I propose in this judgment to deal first with the general principles of EU law which are relevant, and on which I believe that there is substantial agreement among the members of the court. I shall then address the argument that a claim to recover overpaid corporation tax on the principle in Woolwich Equitable is not enough to satisfy those principles. I shall then, finally, return to EU law to consider the main question which has divided this court, namely whether, even if English law did not need to make available a right to recover the tax on the footing of mistake, having done so it could lawfully curtail the limitation period for that right retrospectively and without warning or transitional provisions.

    EU law

    144. Unlike Value Added Tax and certain other taxes and duties which are required and directly regulated by EU law, corporation tax is a creature of the domestic law of the United Kingdom. Apart from the limited requirements of Directive 90/435/EEC relating to withholding tax and double taxation relief, it is not subject to any EU scheme of harmonisation. Like other national tax systems, however, corporation tax is affected by EU law because it must be assessed and collected on a basis consistent with the Treaty. In particular, it must comply with the requirements of the single market, including the freedom of establishment and the free movement of capital guaranteed by what are now articles 49 and 63 of the Treaty: Commission v France (Case C-270/83) [1986] ECR 273; Staatssecretaris van Financien v Verkooijen (Case C-35/98) [2000] ECR I-7321.
    145. The internal market is a domain in which competence is shared between the institutions of the EU and those of member states under article 4 of the Treaty. It follows that even in cases where EU law confers direct rights on private parties, it is for national courts applying national law to determine what rights of action are available against member states to vindicate those rights, and subject to what procedural or other conditions. In Rewe-Zentralfinanz eG v Landwirtschaftskammer fur das Saarland Case 33/76 [1976] ECR 1989 (Rewe I), the principle was stated at para. 5 in terms which have been repeated or paraphrased in many cases decided since:

    “it is the national courts which are entrusted with ensuring the legal protection which citizens derive from the direct effect of the

    provisions of Community law. Accordingly, in the absence of Community rules on this subject, it is for the domestic legal system of each member state to designate the courts having jurisdiction and to determine the procedural conditions governing actions at law intended to ensure the protection of the rights which citizens have from the direct effect of Community law, it being understood that such conditions cannot be less favourable than those relating to similar actions of a domestic nature. Where necessary, articles 100 to 102 and 235 of the Treaty enable appropriate measures to be taken to remedy differences between the provisions laid down by law, regulation or administrative action in member states if they are likely to distort or harm the functioning of the Common Market. In the absence of such measures of harmonization the right conferred by Community law must be exercised before the national courts in accordance with the conditions laid down by national rules. The position would be different only if the conditions and time-limits made it impossible in practice to exercise the rights which the national courts are obliged to protect.”

    One consequence of this, as the court pointed out in Metallgesellschaft Ltd v Inland Revenue Comrs (Joined Cases C‐397/98 and C‐410/98) [2001] Ch 620, para 81, is that the nature, basis and legal classification of rights of action available for this purpose in the national court is a matter for national courts:

    “It must be stressed that it is not for the court to assign a legal classification to the actions brought by the plaintiffs before the national court. In the circumstances, it is for the claimants [Metallgesellschaft Ltd. and others and Hoechst AG] to specify the nature and basis of their actions (whether they are actions for restitution or actions for compensation for damage), subject to the supervision of the national court.”

    146. This is, however, subject to the overriding requirement derived from the Treaty and referred to in the passage which I have quoted from Rewe I, that national legal systems should provide a minimum standard of protection for EU law rights. In the case law of the Court of Justice, the standard of protection required is embodied in two principles which are restated in almost every decision on the point. First, the substantive and procedural provisions of national law must be effective to protect EU law rights (the “principle of effectiveness”). Their enforcement in national law must not be subject to onerous collateral conditions or disproportionate procedural requirements. They must not render “virtually impossible or excessively difficult” the exercise of rights conferred by EU law. Secondly, the relevant provisions of national law must not discriminate between the rules and procedures applying to the enforcement of EU law rights, and those

    applying to the enforcement of comparable national law rights (the “principle of equivalence”). There is a third principle which features less prominently in the case law on this subject but is of considerable importance because it informs the approach of the Court of Justice to the first two. This is the principle of legal certainty, which lies at the heart of the EU legal order and entails (among other things) that those subject to EU law should be able clearly to ascertain their rights and obligations. One aspect of that principle is that within limits EU law will protect within its own domain legitimate expectations adversely affected by a change in the law.

    147. The leading case on the principle of effectiveness is Amministrazione delle Finanze dello Stato v SpA San Giorgio (Case 199/82) [1983] ECR 3595. This concerned charges levied for frontier health inspections of imported animals or animal products under Italian legislation but contrary to EU law. Italian law provided for the recovery of the charges on conditions that were in themselves perfectly acceptable, but which were in practice almost impossible to satisfy because of the exacting rules of evidence applicable to such claims. The court held (para 12):

    “In that connection it must be pointed out in the first place that entitlement to the repayment of charges levied by a member state contrary to the rules of Community law is a consequence of, and an adjunct to, the rights conferred on individuals by the Community provisions prohibiting charges having an effect equivalent to customs duties or, as the case may be, the discriminatory application of internal taxes. Whilst it is true that repayment may be sought only within the framework of the conditions as to both substance and form, laid down by the various national laws applicable thereto, the fact nevertheless remains, as the court has consistently held, that those conditions may not be less favourable than those relating to similar claims regarding national charges and they may not be so framed as to render virtually impossible the exercise of rights conferred by Community law.”

    These principles were restated in the judgments of the European Court of Justice in Metallgesellschaft [2001] Ch 620, paras 84-86 and in the first reference in this litigation: FII Group Litigation v Inland Revenue Comrs (Case C-446/04) [2007] STC 326, paras 201-208. It follows that a member state is “in principle required to repay charges levied in breach of Community law”: Société Comateb v Directeur Général des Douanes et Droits Indirects (Joined Cases C-192/95 to 218/95) [1997] ECR I‐165, para 20. Subsequent case law has emphasized the absolute character of this obligation. The only exception which has been recognized to date is the case where the charge has been passed on by the party who paid it, with the result that he would be unjustly enriched were he to recover it for his own benefit:

    see Weber’s Wine World Handels GmbH v Abgabenberufenskommission Wien (Case C-147/01) [2003] ECR I-11365, para 94. So, although national courts and legislatures are the masters of their own law and procedure, in so far as the legal system of a member state fails to give adequate effect to directly effective EU law rights, it is incumbent on national courts to give effect to those rights by filling the gap between existing causes of action or if necessary to create a new one: see Unibet (London) Ltd v Justitiekanslern (Case C‐432/05) [2008] All ER (EC) 453, paras 40-1.

    148. The combined effect of (i) the requirement of EU law that there must be an effective right of recovery of tax charged contrary to that law and (ii) the primacy of national law as the source of that right, is that EU law does not, indeed cannot, require that national law should recognise or create any particular cause of action or any particular remedy. It simply requires that whatever causes of action or remedies exist in national law must, taken as a whole, be effective and non- discriminatory.
    149. The implications of these principles for the operation of rules of limitation in national systems of law is the subject of a considerable body of case law in the Court of Justice. Not only is limitation a feature of every national legal system of the EU, but the recognition of national rules of limitation as both necessary and desirable is treated as part of the principle of legal certainty in EU law. In Rewe I [1976] ECR 1989, one of the first cases to come before the Court of Justice about the application of limitation periods to claims to enforce directly effective rights in the area of tax, the court observed (para 5) that “the laying down of such time- limits with regard to actions of a fiscal nature is an application of the fundamental principle of legal certainty protecting both the tax-payer and the administration concerned.” This is so, notwithstanding that “the effect of that rule is to prevent in whole or in part the repayment of those charges”: Haahr Petroleum Ltd v Åbenrå Havn (Case C-90/94) [1997] ECR I-4085, para 45. Subject to the overriding principles of effectiveness and equivalence, EU law recognizes the public interest in orderly national budgeting and equity between generations of taxpayers, which will generally require rules for establishing clear limits beyond which tax accounts may not be reopened.
    150. In the present appeals it has not been argued that section 320 of the Finance Act 2004 or section 107 of the Finance Act 2007 are inconsistent with the principle of equivalence. I do not find that surprising. The two enactments with which we are concerned apply in precisely the same way to claims to recover taxes charged contrary to domestic and EU law. So far as they create practical limitations on a claimant’s choice of legal route to recovery, they have precisely the same effect whether the charging of the tax was contrary to EU or domestic law. It is not suggested in these appeals that either enactment offended against the principle in Deville v Administration des Impôts (Case 240/87) [1988] ECR 3513 on the

    ground that they were specifically targeted at the assertion of rights under EU law. We are therefore concerned on these appeals only with the principle of effectiveness and the principle of the protection of legitimate expectations.

    151. The fundamental requirement of the principle of effectiveness is that limitation periods should be reasonable, ie not so short as to make recovery by action “impossible or excessively difficult”: see Rewe I, [1976] ECR 1989, and Comet v Produktschap voor Siergewassen (Case 45/76) [1976] ECR 2043, paras 16-18. But the assessment of what is reasonable allows for considerable variation between different national systems. There is abundant case-law concerning limitation periods much shorter than six years, which have been held to be reasonable. Moreover, it is not inconsistent with the principle of effectiveness that under national law the limitation period for the recovery of unlawful charges should run from the time of payment: see Edilizia Industriale Siderurgica Srl (Edis) v Ministero delle Finanze (Case C-231/96) [1998] ECR I-4951, para 35, Ministero delle Finanze v SPAC (Case C–260/96) [1998] ECR I-4997, para 32. Nor is there any rule of EU law requiring the running of a limitation period to be deferred until the existence of a right to recover the payment has been judicially established. It is not uncommon for a claim to repayment to have become time- barred in national law while proceedings are still in progress to determine whether the member state was in breach of EU law. This was, for example, the position in Rewe I. It was also the position in many of the decisions about the retrospective curtailment of limitation periods, which I shall consider next.
    152. The curtailment of an existing limitation period gives rise to special considerations. There are two objections that might in principle be taken to it. First, even if the change applies only to future claims, it is likely to operate retrospectively to some extent. It will usually extinguish the possibility of enforcing existing rights to recover sums which have already been paid and could in due course have been reclaimed and recovered under the previous law, but are time-barred under the new one. This necessarily engages the principle of effectiveness. Of course, the legislation may also be retrospective in the more radical sense of abrogating claims that have already been properly made under the old law. The second potential objection is that to the extent that the change is retrospective, it may offend against the principle of legal certainty. People must be taken to appreciate that the law may be changed. But until it is, they are entitled to organise their affairs on the basis of the law as it stands and to assume a sufficient measure of predictability in its future development to enable them to exercise their EU law rights. This means that if they have already paid money which is in principle recoverable, they are entitled to be guided by the existing law when deciding how long they have left in which to claim. This objection is commonly analysed as depending on the principle of the protection of legitimate expectations. But this is not really a distinct principle. It has been described as “the corollary of the principle of legal certainty, which requires that legal rules be clear and precise,

    and aims to ensure that situations and legal relationships governed by Community law remain foreseeable”: Duff v Minister of Agriculture, Ireland (Case C-63/93) [1996] ECR I-569, para 20. It is one manifestation of the broader principle that those subject to the law should be able to ascertain their rights and obligations at the time that they are called on to decide what to do about them.

    153. EU law might have taken an absolute line on national legislation retrospectively extinguishing the possibility of enforcing existing rights to recover money charged contrary to EU law. In fact, it has taken a more flexible and nuanced position. It follows from the liberty given to member states to devise their own domestic law means of giving effect to EU rights, that national legislatures are in principle entitled to change their laws. Because they are not obliged to provide more than the minimum level of protection for EU rights necessary to make them effective, the changes may adversely affect claims to assert EU rights, provided that the new law still provides an effective means of doing so. The compromise which EU law has adopted between these conflicting considerations is to allow the retrospective curtailment of limitation periods within limits set by the principle of the protection of legitimate expectations. Legislation curtailing limitation periods is in principle consistent with the principle of effectiveness provided that a period of grace, which may be quite short, is allowed, either by giving sufficient advance notice of the change or by including transitional provisions in the legislation. These propositions are derived from the four leading decisions of the Court of Justice on this question, namely Aprile Srl v Amministrazione delle Finanze dello Stato (Case C-228/96) [2000] 1 WLR 126 (Aprile II), Dilexport Srl v Amministrazione delle Finanze dello Stato (Case C- 343/96) [1999] ECR I-579, [2000] All ER (EC) 600, Grundig Italiana SpA v Ministero delle Finanze (Case C-255/00) [2003] All ER (EC) 176, and Marks & Spencer v Customs & Excise Comrs (Case C-62/00) [2003] QB 866.
    154. The first two cases had a similar legal background. Italy had unlawfully levied charges equivalent to customs duties, which the claimant sought to recover under Italian law. Italian law conferred a general right to recover payments made without legal basis (“pagamento non dovuto”) under article 2033 of the Civil Code, which was subject to the general limitation period of ten years provided for by the article 2946 of the Code. In addition, there was a specific right to a refund under the Consolidated Customs Code in cases of “calculation errors in the assessment or the application of a duty other than that laid down in the tariff”, which was subject to its own limitation period of five years. The latter right had no application to a claim for a refund of tax charged contrary to Community law. These provisions were amended by legislation so that the limitation period in the Customs Code applied to actions under article 2033 whenever the claim was for a “refund of sums paid in connection with customs operations.” In other words, the limitation period for the only relevant right of recovery, under article 2033, was reduced from ten years to five. It is clear that the decisive considerations which led

    the Court of Justice to conclude that the amendment was consistent with the EU law were (i) that the reduced limitation period was still long enough to satisfy the principle of effectiveness and (ii) that the Italian courts had treated the amending legislation as allowing claimants a period of grace of three years from the time the amending legislation came into force, which meant that the legislation “cannot be regarded as having retrospective effect”; see Aprile II, para 28 and Dilexport, para

    42. This was not enough to help the claimants, for the period of grace had already expired by the time that they succeeded in obtaining a judicial decision that the charges were unlawful. But it was held to be enough to satisfy the principle of effectiveness.

    155. In Grundig Italiana [2003] All ER (EC) 176, the Court of Justice had to consider the minimum transitional period which would enable a retrospective curtailment of the limitation period to satisfy EU law. The case concerned the same amending legislation which had featured in Aprile II and Dilexport, but a different aspect of it, namely the reduction of the special limitation period from five years to three, which took effect ninety days after the legislation came into force. This transitional period was held to be too short. The court considered that a period of grace must be “sufficient to allow taxpayers who initially thought that the old period for bringing proceedings was available to them a reasonable period of time to assert their right of recovery in the event that under the new rules they would already be out of time” (para 38). In the context of an original limitation period of five years, six months was the least that should have been allowed (para 42). It is accordingly clear that a reasonable period of grace may be considerably shorter than the amount by which the limitation period has been abridged. It is a period long enough to allow potential claimants to consider their position, not a period long enough to save every existing right of recovery.
    156. In none of the Italian cases was separate consideration given by the Court of Justice to the principle of the protection of legitimate expectations. But that principle must necessarily have informed the court’s analysis of what was required by the principle of effectiveness. The point was made in terms by Advocate- General Ruiz-Jarabo in his opinion in Grundig Italiana, where he observed that a retrospective reduction in the limitation period without a period of grace would be contrary to the principle of effectiveness “on the grounds that the reduction is unexpected and contrary to the principle of the protection of legitimate expectations and to the principle of legal certainty” (para 30). The court must have agreed with that. It was critical to its view that legislation retrospectively curtailing an existing limitation period so as to bar some existing rights, would nevertheless be consistent with the principle of effectiveness if it allowed a sufficient period of grace for taxpayers to adjust their plans to the new order of things.
    157. In Marks & Spencer [2003] QB 866, the facts were more complex. It was a reference from the Court of Appeal in England about a claim to recover VAT

    unlawfully charged by the Commissioners of Customs and Excise. By statute, the only right to obtain a refund from the Commissioners was by way of a claim under section 24 of the Finance Act 1989 (subsequently section 80 of the Value Added Tax Act 1994). Subsections (4) and (5) of section 24 provided for a six year limitation period, which was to run from the date of payment save in cases of mistake, when it was to run from the time when the mistake was or could with reasonable diligence have been discovered. On 18 July 1996, the government announced its intention of introducing what later became section 47(1) of the Finance Act 1997. The effect of this enactment was to reduce the limitation period for the statutory claim from six years to three, and to provide that it was to run in all cases from the time of payment. Section 47(2) provided that subsection (1) should be deemed to have come into effect on 18 July 1996 and should apply to all claims unsatisfied at that date whether made before or afterwards. There were no relevant transitional provisions. The reference was concerned with a claim to recover VAT overpaid on sales of gift vouchers. This claim was affected by the reduction of the limitation period to three years. It was not affected by the removal of the extended period of limitation in cases of mistake, because the relevant payments had all occurred within six years before the claim was made. But the facts are complicated by the existence of another claim, to recover VAT paid in respect of sales of teacakes going back to 1973, which was significantly affected by the removal of the extended limitation period. The teacakes claim was not part of the reference: see the Advocate-General at para 27. But before us a submission was based on it by Mr Aaronson QC (for the Test Claimants) because of the analogy with the removal of the extended period of limitation in the present case. It is therefore right to point out that it arose only in the context of a “preliminary observation” of the Advocate-General about the way in which the Court of Appeal had framed the reference. The Court of Appeal had limited it to (i) the gift vouchers claim, (ii) the reduction of the limitation period from six years to three, and (iii) the period before August 1996 when the Sixth VAT Directive 77/388/EEC had been in force but not properly transposed into the law of the United Kingdom. The Advocate-General, while acknowledging that the court was bound by the terms of the reference, pointed out that it had been framed on the assumption that the Directive had no further relevance as a source of rights once it had been properly transposed into English law in August 1996. This assumption was in his opinion wrong: paras 32-34. He thought that the Court of Appeal’s error about the period in which the Directive was relevant had led it to treat the whole of the teacakes claim and the later part of the gift vouchers claim as depending only on national law: see paras 30 and 44, and his citations from the judgments of the High Court and the Court of Appeal at para 32. None of this had anything to do with the compatibility of section 47 of the Finance Act 1997 with EU law. The Court of Justice, in its judgment, agreed that the Court of Appeal’s assumption about the Directive was mistaken (paras 22-28), but dealt only with the application of the 1997 Act to the gift vouchers claim: see para 13.

    158. The Court of Justice had no difficulty in concluding that section 47 was contrary to the principle of effectiveness. There was only one means allowed by English law of recovering overpaid VAT, and the effect of the amendment was to extinguish without notice any possibility of using that method to recover overpayments between three and six years old. Indeed, it extinguished it even when there was already a pending claim at the date of the announcement. The court took the opportunity to restate the effect of previous case law in the following terms:

    “35. As regards the latter principle, the court has held that in the interests of legal certainty, which protects both the taxpayer and the administration, it is compatible with Community law to lay down reasonable time-limits for bringing proceedings: Aprile, paragraph 19, and the case-law cited therein). Such time-limits are not liable to render virtually impossible or excessively difficult the exercise of the rights conferred by Community law. In that context, a national limitation period of three years which runs from the date of the contested payment appears to be reasonable (see, in particular, Aprile, paragraph 19, and Dilexport, paragraph 26).

    36. Moreover, it is clear from the judgments in Aprile [2000] 1 WLR 126, para 28, and Dilexport [1999] ECR I-579, paras 41 and 42, that national legislation curtailing the period within which recovery may be sought of sums charged in breach of Community law is, subject to certain conditions, compatible with Community law. First, it must not be intended specifically to limit the consequences of a judgment of the court to the effect that national legislation concerning a specific tax is incompatible with Community law. Secondly, the time set for its application must be sufficient to ensure that the right to repayment is effective. In that connection, the court has held that legislation which is not in fact retrospective in scope complies with that condition.
    37. It is plain, however, that that condition is not satisfied by national legislation such as that at issue in the main proceedings which reduces from six to three years the period within which repayment may be sought of VAT wrongly paid, by providing that the new time-limit is to apply immediately to all claims made after the date of enactment of that legislation and to claims made between that date and an earlier date, being that of the entry into force of the legislation, as well as to claims for repayment made before the date of entry into force which are still pending on that date.
    38. Whilst national legislation reducing the period within which repayment of sums collected in breach of Community law may be sought is not incompatible with the principle of effectiveness, it is subject to the condition not only that the new limitation period is reasonable but also that the new legislation includes transitional arrangements allowing an adequate period after the enactment of the legislation for lodging the claims for repayment which persons were entitled to submit under the original legislation. Such transitional arrangements are necessary where the immediate application to those claims of a limitation period shorter than that which was previously in force would have the effect of retroactively depriving some individuals of their right to repayment, or of allowing them too short a period for asserting that right.”
    159. The decision is also notable as being the one case in this area in which the court has given separate and explicit consideration to the principle of the protection of legitimate expectations. It did so because it was expressly invited to deal with both effectiveness and legitimate expectations by the terms of the Court of Appeal’s reference. But it dealt with the point under both heads. In dealing with the principle of effectiveness, it observed (para 38) that the principle of effectiveness required that potential claimants should be given time to assert existing rights under the old law. This was because (para 39) the right of member states to impose reasonable limitation periods was an exception to the rule that member states must repay taxes charged in breach of Community law, and that exception was founded on the principle of legal certainty. “However, in order to serve their purpose in ensuring legal certainty, limitation periods must be fixed in advance.” In other words, the curtailment of limitation periods is consistent with the principle of effectiveness if it is subject to provisions protecting legitimate expectations. As the Advocate General had pointed out in his opinion (para 68), the principle of protecting legitimate expectations is based on the need for legal certainty. Addressing the same point, the court held:

    “44. In that connection, the court has consistently held that the principle of the protection of legitimate expectations forms part of the Community legal order and must be observed by the member states when they exercise the powers conferred on them by Community directives: see, to that effect, Krücken (Case 316/86) [1988] ECR 2213, para 22, Alois Lageder SpA v Amministrazione delle Finanze dello Stato (Joined Cases C-31 to C-44/91) [1993] ECR I-1761, para 33, Belgocodex v Belgian State (Case C-381/97) [1998] ECR I-8153, para 26, and Grundstückgemeinschaft Schlossstrasse GbR v Finanzamt Paderborn (Case C-396/98) [2000] ECR I-4279, para 44).

    45. The court has held, in particular, that a legislative amendment retroactively depriving a taxable person of a right to deduction he has derived from the Sixth Directive is incompatible with the principle of the protection of legitimate expectations (Schlossstrasse, cited above, paragraph 47).
    46. Likewise, in a situation such as that in the main proceedings, the principle of the protection of legitimate expectations applies so as to preclude a national legislative amendment which retroactively deprives a taxable person of the right enjoyed prior to that amendment to obtain repayment of taxes collected in breach of provisions of the Sixth Directive with direct effect.”
    160. Whether it is put on the basis of the principle of effectiveness or the protection of legitimate expectations or on a combination of the two, the rule of EU law which requires a reasonable period of grace before a retrospective curtailment of the limitation period can be lawful, assumes that claimants generally can legitimately count on having the whole of the old limitation period in which to bring whatever claims may be available to them as a matter of domestic law, unless they have a reasonable warning that the position is about to change. Thus far, I do not think that there is any fundamental difference in principle between my views and those of other members of the court.
    161. The assumption that a claimant can legitimately count on having the whole of the old limitation period in which to bring whatever claims are available to him is one which would normally be made as a matter of course. But this is not an ordinary case. The position is complicated by the highly unusual way in which the right to recover unlawfully charged tax has developed in England over the last two decades. It is a problem which could only have arisen in a common law country such as England, where the law of restitution has been the piecemeal creation of judges while limitation is exclusively the creature of statute. To these peculiarly English developments, I now turn.

    Rights of recovery in English law

    162. Until surprisingly recently, English law afforded only very limited possibilities of recovering overpaid tax. As Lord Goff of Chieveley observed in Woolwich Equitable Building Society v Inland Revenue Comrs [1993] AC 70, 172, English law had not recognised a condictio indebiti allowing an action for the recovery of payments on the simple ground that they were not due. It has still not done so. It is necessary, as the law presently stands, to bring the facts within one of the categories of case in which the law recognises that the recipient’s retention of

    the money would be unjust. The relevant categories as they had stood for a considerable time up to 1992 were described by Lord Goff in his speech in Woolwich Equitable at pp 164-166. Money was recoverable if it was paid under a mistake of fact, but not if it was paid under a mistake of law (as it generally would be if taxes were paid which were not duly authorised by law). It was recoverable if it was exacted by actual or threatened duress to the person or to the person’s goods, but not on a mere threat to assert a claim by a method provided for by law (for example, by legal proceedings). It was recoverable if it was demanded by a public official or a person charged with a statutory duty as a condition of his performing his duty. None of these situations was likely to cover the case where a taxpayer paid money which was not in fact due under the relevant legislation, because it had been misconstrued or was contrary to EU law, or because (being secondary legislation) it was ultra vires the enabling Act.

    163. A limited statutory right to claim repayment from the Commissioners had been introduced in 1923 by section 24 of the Finance Act of that year. Substantially the same provision has remained in force in successive statutory iterations ever since. It is currently to be found in section 33 of the Taxes Management Act 1970. In that form, its effect is that overpaid tax may be reclaimed if (i) it was charged by an assessment, (ii) the assessment was excessive because of a mistake in the taxpayer’s return, (iii) in the case of a mistake about the basis on which the taxpayer’s liability should be computed, the return was not in the relevant respect made in accordance with the “practice generally prevailing at the time”, and (iv) having examined all the relevant circumstances of the case, the Board of Inland Revenue or the Special Commissioners on appeal from them considered that repayment would be “reasonable and just”. It will be apparent that if tax was paid under invalid or unlawful legislation the claim will almost inevitably fail on the ground that the return having been made in accordance with the statute it accorded with the practice generally prevailing at the time. Even if the taxpayer can demonstrate that his return was not in accordance with that practice, the fate of his claim will depend on the exercise of a discretion by the Commissioners. His only right is to have his claim fairly considered in the light of all relevant circumstances. As Lord Goff pointed out in Woolwich Equitable (at p 170B), historically this provision presupposed that there was no right of recovery at common law.
    164. The first major change in this state of affairs occurred with the judgment of the House of Lords in the Woolwich Equitable case, which was delivered on 20 July 1992. The Woolwich Equitable Building Society paid the composite rate tax charged on building societies under statutory regulations which it considered to be ultra vires the enabling primary legislation, and which it then successfully challenged in proceedings for judicial review. It took this course because it was concerned about the reputational damage that it might suffer if it was seen to withhold tax which other building societies were paying, at a time when there had

    been no definitive decision on the status of the regulations. The Commissioners, having failed to justify the charge in the judicial review proceedings, repaid the tax, but declined to recognise that they were bound to do so and therefore felt entitled to reject a claim to interest. The question at issue was whether the Commissioners had been bound to repay the principal and were therefore amenable to an order for the payment of interest as well. Woolwich was unable to bring itself within any of the established categories of restitution. In particular, it could not claim repayment on the ground of mistake, because it had not been mistaken. It had always believed that the regulations were void. Nor could it claim under section 33 of the Taxes Management Act 1970, because there had been no assessment. It had pre-empted an assessment by paying. It followed that under the law as it had previously stood, the claim for interest was bound to fail. The question, as Lord Goff put it at p 171, was whether the House in its judicial capacity should “reformulate the law so as to establish that the subject who makes a payment in response to an unlawful demand of tax acquires forthwith a prima facie right in restitution to the repayment of the money”. The claim failed in the High Court, but it succeeded, by a majority, first in the Court of Appeal and then, on somewhat different grounds, in the House of Lords. In summary, the House of Lords fashioned a cause of action which was (i) acknowledged to be new, (ii) specific to the case of money charged by a public authority in the absence of a valid statutory power to do so, and (iii) available irrespective of whether the payer was mistaken or whether, if he was mistaken, his mistake was one of fact or law.

    165. It was not necessary in Woolwich Equitable to consider the rule that money paid under a mistake of law was irrecoverable. That question came before the House of Lords in 1998 in Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349, one of the last cases to be decided in the great tide of litigation arising out of ultra vires interest rate swap agreements with local authorities. Kleinwort Benson had made net payments to the local authorities under the terms of these agreements, which they claimed had been made under a mistake of law, namely that they were valid. In the High Court, Langley J dismissed the claims on the ground that the law did not recognise a right to recover in these circumstances. The case was leapfrogged to the House of Lords on the ground that the Court of Appeal would be bound by authority to reach the same conclusion. In the House of Lords, the bank acknowledged that the existing law did not allow the recovery of money paid under a mistake of law. For their part, the local authorities made no attempt to defend that state of the law in principle, in the face of sustained criticism by academic writers and the Law Commission, its total or partial abandonment in many common law jurisdictions, and the recognition of a wider basis of recovery, independent of mistake, in major civil law systems, notably those of Germany, France and Italy. The fundamental issue before the appellate committee was whether the law should be changed by judicial decision, or the task left to Parliament. The House allowed the appeal and recognised a right in principle to recover money paid under a mistake of law, while acknowledging that this

    represented “a departure, even a major departure, from what has previously been considered to be established principle”: see Lord Goff at p 378.

    166. For a number of years it remained uncertain whether the new cause of action to recover money paid under a mistake of law extended to mistaken payments of tax. Kleinwort Benson was a case about private law transactions. In his speech Lord Goff (with whom on this point the rest of the appellate committee agreed) expressed the view at pp 381-382 that there was a distinction between claims to recover payments made in private law transactions and claims to recover payments of taxes and other charges levied by public authorities. In the latter category, payments were recoverable as of right under the principle laid down in Woolwich Equitable without the need to invoke a mistake of law, or under section 33 of the Taxes Management Act in cases of mistake to which that provision applied. Lord Goff continued at p. 382:

    “Two observations may be made about the present situation… The first observation is that, in our law of restitution, we now find two separate and distinct regimes in respect of the repayment of money paid under a mistake of law. These are (1) cases concerned with repayment of taxes and other similar charges which, when exacted ultra vires, are recoverable as of right at common law on the principle in Woolwich, and otherwise are the subject of statutory regimes regulating recovery; and (2) other cases, which may broadly be described as concerned with repayment of money paid under private transactions, and which are governed by the common law. The second observation is that in cases concerned with overpaid taxes, a case can be made in favour of a principle that payments made in accordance with a prevailing practice, or indeed under a settled understanding of the law, should be irrecoverable. If such a situation should arise with regard to overpayment of tax, it is possible that a large number of taxpayers may be affected; there is an element of public interest which may militate against repayment of tax paid in such circumstances; and, since ex hypothesi all citizens will have been treated alike, exclusion of recovery on public policy grounds may be more readily justifiable.”

    The Commissioners, relying mainly on this passage, subsequently contended that tax was subject to a special legal regime, and that the only cause of action at common law for the recovery of overpaid tax was a cause of action on the principle stated in Woolwich Equitable. The recognition of this basis of claim, it was said, impliedly excluded all other bases of claim apart from the statutory procedure under section 33 of the Taxes Management Act 1970.

    167. This proposition was tested in Deutsche Morgan Grenfell Group Plc v Inland Revenue Comrs [2007] 1 AC 558. The case foreshadowed some of the issues on the present appeals, and was the genesis of section 320 of the Finance Act 2004. It concerned claims for the recovery of interest on corporation tax which the European Court of Justice had held to have been prematurely charged in Hoechst/Metallgesellschaft. The taxpayer company wanted to claim interest for the period when it was out of pocket, on the footing that the tax itself had been paid under a mistake of law. It took this course because a claim on that basis would benefit from the extended limitation period under section 32(1)(c) of the Limitation Act 1980, whereas claims based on Woolwich Equitable or section 33 of the Taxes Management Act ran from the time of payment and would have been time-barred. There were three main issues: (i) whether, in a case covered by the principle in Woolwich Equitable, a common law claim based on mistake was also available to the taxpayer; (ii) if so, what was the mistake, bearing in mind that the tax had been paid in accordance with the correct construction of the taxing Acts, which was only later shown to be inconsistent with EU law by the decision of the Court of Justice in Metallgesellschaft Ltd v Inland Revenue Comrs (Joined Cases C-397/98 and 410/98) [2001] Ch 620; and (iii) at what stage, for the purpose of section 32(1)(c) of the Limitation Act 1980, could it be said that the taxpayer “discovered… or could with reasonable diligence have discovered” that mistake, so as to start the limitation period running. Park J gave judgment on 18 July 2003. He decided all three questions in favour of the taxpayer, and held that accordingly a claim on the basis of mistake was available to it. In February 2005, the Court of Appeal unanimously overruled him and held that it was not. The House of Lords restored the judgment of Park J on 25 October 2006. On the first issue, the House held that the claimant could choose between concurrent causes of action on the principle in Woolwich Equitable and on the basis of mistake of law. On the second issue, there were some differences of reasoning within the majority of the appellate committee, but all of them agreed that by virtue of the theory that judicial decisions are deemed to declare the existing law, the taxpayer company had made a “retrospective” or “deemed” mistake. The mistake consisted in its failing to appreciate that it was entitled to make a group income election and defer the payment of tax, notwithstanding that the statute said that it did not have this right. On the third issue, the House of Lords held that under section 32(1)(c) of the Limitation Act 1980 the limitation period for a claim in respect of that mistake did not start to run simply because the claimant was aware of a worthwhile claim or of doubts about the lawfulness of the legislation.   It started to run only when the Court of Justice definitively held that the relevant features of the United Kingdom corporation tax regime were contrary to EU law. The combined effect of the decisions on these three points was in one respect extremely remarkable. If tax was overpaid under a mistake of law, then provided that a claim to recover it was brought before six years had elapsed from the judgment establishing the correct legal position, there was no limit upon how far back the claim could go. In the present cases, it goes back to the accession of the United Kingdom to the Common Market in 1973. If it had arisen from a mistake of purely domestic law, it might have gone back to the inception of corporation tax in 1965. In other cases where

    the unlawfulness of the charge depended wholly on English law, it could in principle go back indefinitely.

    168. It has been suggested in argument before us that once the House of Lords in Kleinwort Benson [1999] 2 AC 349 had accepted the right to recover money paid under a mistake of law, the Commissioners’ case in Deutsche Morgan Grenfell was never likely to be accepted. Its acceptance by the Court of Appeal was an aberration, a “bump in the road” to borrow Mr Aaronson’s arresting phrase. Such arguments often sound plausible in hindsight, after the highest court has laid down the law, and ultimately of course the Commissioners’ argument was not accepted. But it was nevertheless a formidable argument, to which the observations of Lord Goff appeared to lend substantial support. In Kingstreet Investments v New Brunswick (Finance) [2007] 1 SCR 3, considerations rather similar to those raised by Lord Goff had led the Supreme Court of Canada to treat claims to recover unlawfully charged tax as governed by a distinct body of constitutional principle relating to tax charged without legislative authority, and not by the general law of unjust enrichment. At least part of the Canadian court’s reasoning was that the concurrent availability of both causes of action was liable to have unacceptable collateral consequences: see paras 32-42 (Bastarache J). Indeed, the decision of the House of Lords in Deutsche Morgan Grenfell is even now not beyond academic controversy. The decision on issue (ii) is criticised by the current editors of Goff & Jones, The Law of Unjust Enrichment, 8^th ed (2011), paras 22.29 – 22.31 on grounds closely related to the observations which I have quoted from Lord Goff in Kleinwort Benson. I do not intend by making these points to reopen a debate which has been settled for more than five years. My point is more straightforward: no reasonable and well-advised person could have counted on the decision in Deutsche Morgan Grenfell going the way it did on all three points, until the House of Lords delivered its judgment.
    169. Section 320 of the Finance Act 2004 was a direct response to the decision of Park J in Deutsche Morgan Grenfell. It altered not the limitation period itself but the statutory rule postponing its commencement in cases of mistake until the taxpayer had discovered or could with reasonable diligence have discovered the mistake. It had the effect of barring older claims for repayment of tax paid under a mistake which might otherwise have succeeded. But the mischief to which section 320 was addressed was not the existence of a right to repayment, whether arising from EU or domestic law, but the problem created by Park J’s decision that section 32(1)(c) of the Limitation Act 1980 might now enable past tax accounts to be reopened without limit of time.

    Is the right to bring a claim based on Woolwich Equitable an effective remedy?

    170. Logically, the first question to be decided is whether a cause of action based on the Woolwich Equitable principle is an effective means of asserting the right to repayment required by EU law. The Test Claimants say that it is not. Therefore, the argument runs, their only effective means of recovery was by way of an action to recover on the ground of mistake, and their right to bring such an action has been unlawfully curtailed by section 320 of the Finance Act 2004. They make three points. First, they say that a claim based on Woolwich Equitable requires an unlawful demand by a public authority and is not therefore available to recover taxes such as advance corporation tax which are paid with the return, not upon an assessment or other demand by the Commissioners. Second, the Court of Appeal has held, applying the principle in Marleasing SA v La Comercial Internacional de Alimentación SA (Case C‐106/89) [1990] ECR I-4135 that section 231 of the Income and Corporation Taxes Act 1988 (which provides for a tax credit in respect of distributions paid by UK resident companies) can be given a construction which, however strained, applies it to distributions by companies resident anywhere in the EU. It follows, they say, that there is nothing unlawful about section 231 which can engage the principle in Woolwich Equitable. If either of these points is right, then a claim based on Woolwich Equitable is not an effective remedy in this case. Third, the Test Claimants submit that section 320 of the Finance Act 2004 and section 107 of the Finance Act 2007 curtail the limitation period for a claim based on Woolwich Equitable, because although such a claim does not legally depend on mistake, they were in fact mistaken. Their action is therefore an action for “relief from the consequences of a mistake” for the purposes of section 32(1)(c) of the Limitation Act 1980. If this point is right, then Parliament has without warning curtailed the limitation period for all available methods of obtaining restitution, apart from a claim under section 33 of the Taxes Management Act for a small part of the overpayment and a somewhat problematical claim for damages founded on the principle of state liability stated by the European Court of Justice in Francovich v Italian Republic (Joined Cases C-6/90 and 9/90) [1999] ICR 722, [1991] ECR I-5357. The Test Claimants need only be right on one of these three points, but in my view they are wrong on all of them.

    The demand point

    171. In spite of the importance attached to this point in argument, it can I think be dealt with quite shortly. The speeches of the majority in Woolwich Equitable [1993] AC 70 are full of expressions which, read literally and out of their analytical context, might support the suggestion that the cause of action recognised in that case was dependent on the making of an “unlawful demand”: see Lord Goff at 171F-G, 172B-C, 174C-D, 177F, Lord Browne-Wilkinson at 196G-H, 197C-H, 198B-C, G-H, and Lord Slynn of Hadley at 199B-D, 200B-C, 201D-E, 202G- 203A, 204F-H, 205A-B. None of the majority in Woolwich Equitable discusses what they meant by a “demand”. But both the facts of the case and the reasoning

    of the majority show that they cannot have had in mind a formal demand by the Inland Revenue triggering a payment or an apparent obligation to pay.

    172. The facts, which are succinctly set out by Ralph Gibson LJ in the Court of Appeal (pp 104-105) show that Woolwich Equitable did not pay the composite rate tax in response to a formal demand. The inspector had simply invited it to agree figures and the collector had sent it a return form. The society sent in a return computed according to the Regulations, under cover of a letter informing them that they proposed to challenge their validity in legal proceedings. They then paid without prejudice to the outcome. As Lord Goff pointed out (at p 169) no assessment was ever made, because Woolwich pre-empted it by paying.
    173. It is fair to look for the reasoning of the House of Lords mainly in the classic analysis of Lord Goff, although similar points were made by Lord Browne- Wilkinson, who agreed with Lord Goff in terms and by Lord Slynn, who agreed with him in substance. It is apparent that the mischief which justified in Lord Goff’s eyes a special rule for unlawful charges by public authorities was (i) that no tax should be collected without Parliamentary authority, and (ii) that citizens did not deal on equal terms with the state, and could not be expected to withhold payment when faced with the coercive powers of the Revenue, whether those powers were actually exercised or merely held in reserve: see pp. 172. At pp. 175- 176, Lord Goff adopted the dissenting judgment of Wilson J in the Supreme Court of Canada in Air Canada v British Columbia (1989) 59 DLR (4^th) 161. In her judgment, Wilson J had expressed the view that there was a general right to recover money paid under unconstitutional legislation, and deprecated any suggestion that it must have been paid under protest. The reason, as she pointed out at p 169, was that the legislature holds out its legislation as valid and that any loss resulting from payment under it “should not fall on the totally innocent taxpayer whose only fault is that it paid what the legislature improperly said was due”. The emphasis in this reasoning was on the unlawful character of the legislation, with which in practice the citizen was bound to comply even if it might subsequently be shown to be void. This approach has subsequently been adopted by the Supreme Court of Canada in Kingstreet Investments Ltd v New Brunswick (Finance) [2007] 1 SCR 3, to which I have already referred in another context. Lord Goff not only found the reasoning of Wilson J “most attractive” (p 176D), but expressed his own conclusions in very similar terms. “In the end,” he said (p 173), “logic appears to demand that the right of recovery should require neither mistake nor compulsion, and that the simple fact that the tax was exacted unlawfully should prima facie be enough to require its repayment”. The “exaction” of which he is speaking here is not confined to demands by any particular administrative agency of the state. It includes exaction by the state by enacting void legislation, which taxpayers are likely to pay because they know that the state will act on the footing that it is valid. It is not a condition of the taxpayer’s right of recovery that it should have put the matter to the test by waiting until the Inland

    Revenue insisted. In a passage at p 177 which strikingly foreshadows some of the issues in the present appeals, Lord Goff assimilated the rule of English law as he had formulated it to the absolute right of recovery recognized by the European Court of Justice in San Giorgio (Case 199/82) [1983] ECR 3595 in cases where tax was charged contrary to EU law. Although the majority of the appellate committee stopped well short of adopting a concept of “absence of legal basis” as a general ground of recovery even in cases of taxation without lawful authority, Lord Browne-Wilkinson’s analysis of the legal basis of recovery in such cases was also very similar to that of the case law of the Court of Justice. Money unlawfully “demanded” was recoverable because it was paid for no consideration: see p 198.

    174. The word “demand” as it was used in the speeches in Woolwich Equitable referred in my view simply to a situation in which payment was being required of the taxpayer without lawful authority. Nothing in the principle underlying the decision turned on the mechanism by which that requirement was communicated to the taxpayer. It is therefore a matter of supreme indifference whether it was communicated by assessment, or by some other formal mode of demand, or by proceedings for enforcement, or by the terms of the legislation itself coupled with the knowledge that the Inland Revenue would be likely to enforce it in accordance with those terms.

    The Marleasing point

    175. The Court of Appeal [2010] STC 1251, para 107 held that on the principle of conforming construction stated in Marleasing, section 231 of the Income and Corporation Taxes Act 1988 should be construed so as to remove the discriminatory features of the United Kingdom’s advance corporation tax regime. For present purposes we must assume that they were right about this. An appeal on that issue is not before us. The right to apply for permission to appeal on it has been deferred pending the outcome of the second reference to the Court of Justice and its application by the courts below. The argument of the Test Claimants is that on the assumption that the Court of Appeal’s construction is correct the legislation conformed to EU law. Therefore, it is said, the principle in Woolwich Equitable is not engaged.
    176. Marleasing (Case C-106/89) [1990] ECR I-4135, at any rate as it has been applied in England, is authority for a highly muscular approach to the construction of national legislation so as to bring it into conformity with the directly effective Treaty obligations of the United Kingdom. It is no doubt correct that, however strained a conforming construction may be, and however unlikely it is to have occurred to a reasonable person reading the statute at the time, a later judicial decision to adopt a conforming construction will be deemed to declare the law retrospectively in the same way as any other judicial decision. But it does not

    follow that there was not, at the time, an unlawful requirement to pay the tax. It simply means that the unlawfulness consists in the exaction of the tax by the Inland Revenue, in accordance with a non-conforming interpretation of what must (on this hypothesis) be deemed to be a conforming statute. This is so, notwithstanding that the tax may have been paid without anything in the nature of a formal demand by the Inland Revenue. The rule as the House of Lords formulated it in Woolwich Equitable is in large measure a response to realities of the relationship between the state and the citizen in the area of tax. The fact that as a matter of strict legal doctrine a statute turns out always to have meant something different from what it appeared to say is irrelevant to the realities of power if it was plain at the relevant time that the tax authorities would enforce the law as it then appeared to be. Strictly speaking, in Woolwich Equitable itself there were no unlawful regulations, because, being ultra vires the enabling Act, they were and always had been a nullity. But that did not stop the Woolwich from recovering.

    The section 32(1)(c) point

    177. Section 32(1) is (so far as relevant) in the following terms:

    “Postponement of limitation period in case of fraud, concealment or mistake

    (1) …where in the case of any action for which a period of limitation is prescribed by this Act, either-

    1. the action is based upon the fraud of the defendant; or
    2. any fact relevant to the plaintiff’s right of action has been deliberately concealed from him by the defendant; or
    3. the action is for relief from the consequences of a mistake;

    the period of limitation shall not begin to run until the plaintiff has discovered the fraud, concealment or mistake (as the case may be) or could with reasonable diligence have discovered it.”

    178. The argument for the Test Claimants on these appeals is that in section 32(1)(c) actions “for relief from the consequences of a mistake” are not confined to actions where the mistake is part of the legal foundation of the claim. They

    extend to at least some actions where it was merely part of the history. Mr Rabinowitz QC (who argued this point for the Test Claimants) accepted some limitations of the range of relevant mistakes. He said that there had to be a sufficient causal nexus between the mistake and the claim, in the sense that the facts constituting the cause of action have come to pass because of the mistake. It followed that although the Woolwich Equitable cause of action was available to claimants in the position of his clients regardless of whether they were mistaken or not, those who were in fact mistaken in some historically relevant respect would have benefitted from the extended limitation period until the law was changed by section 320 of the Finance Act 2004. They have been deprived without notice of that right. Section 320(6) removes any doubt about this by providing that it applies “to any action or c1aim for relief from the consequences of a mistake of law, whether expressed to be brought on the ground of mistake or on some other ground (such as unlawful demand or ultra vires act)”.

    179. Section 32(1) of the Limitation Act 1980 substantially re-enacts section 26 of the Limitation Act 1939, with one minor change to paragraph (b) (from concealment “by… fraud” to “deliberate” concealment). The Act of 1939 was a notable monument of law reform, replacing an incoherent series of statutes and equitable rules by a coherent statutory scheme. It was enacted on the recommendation of the Law Revision Committee in its Fifth Interim Report (Cmd 5334), which was prepared in 1936 under the auspices of Lord Wright, then Master of the Rolls. Section 26 substantially followed the language of the report. It is clear from paragraphs 22 and 23 of the Committee’s report that the intention was to replicate certain features of the rules applied by courts of equity in the absence of any statutory limitation period. The equitable rules on this subject had originally been developed in the context of cases involving fraud. The doctrine of laches was applied by analogy with statutory limitation at law, save that in cases of fraud time ran from the point when the fraud was discovered or could with reasonable diligence have been discovered, and not from the accrual of the right as it did at law. It is clear that fraud was relevant in equity in two circumstances, (i) that the right to equitable relief was itself based on fraud, in the sense that fraud was a legally essential element of it, and (ii) that whether or not the right to relief was based on fraud, its existence had been concealed from the plaintiff by the fraud of the defendant. The Law Revision Committee summarised the position at paragraph 22 of their report as follows:

    “Either the cause of action may spring from the fraud of the defendant or else the existence of a cause of action untainted in its origin by fraud may have been concealed from the plaintiff by the fraudulent conduct of the defendant.”

    180. In 1936, when the Committee was considering these matters, there was inconsistent authority on the question whether since the fusion of law and equity

    the equitable rule about the running of time in cases of fraud applied to causes of action at law. They recommended that it should. The result was section 26(a) and

    (b) of the 1939 Act, corresponding to section 32(1)(a) and (b) of the 1980 Act. These two paragraphs dealt with the two circumstances in which fraud was relevant to postpone the running of time in equity, as summarised in the Committee’s report. As applied to fraud neither paragraph admits of the construction now proposed by the Test Claimants. Paragraph (a) is concerned with cases where the action is “based upon” fraud, ie where it is part of the legal foundation of the claim. Paragraph (b) is concerned with cases where fraud by the defendant is not necessarily part of the legal basis of the claim, but it has concealed the relevant facts from the claimant and thereby delayed his taking action to enforce his right. The fact that fraud, although no part of the legal basis of the claim, may have brought about the factual situation which is the legal basis of the claim, does not engage either paragraph.

    181. The reason for enacting section 26(c) of the Limitation Act 1939 (now section 32(1)(c) of the Act of 1980) was that courts of equity had previously applied the equitable rule relating to fraud by analogy to cases of mistake. As Baron Alderson put it in Brooksbank v Smith (1836) 2 Y & C Ex 58, “mistake is… within the same rule as fraud.” The Law Revision Committee considered that in this respect the rule for mistake should be the same at law, and at paragraph 23 of their report they recommended the statutory reversal of the decision in Baker v Courage [1910] 1 KB 56, which had held that it was not. Section 26(c) of the 1939 Act was the result. On the face of it, therefore, the intention behind paragraph (c) was to replicate the rule of equity by providing that mistake should give rise to an extended limitation period in the same circumstances in which fraud had that effect under paragraph (a), namely where it was the legal basis of the claim. The use of a different phraseology in (a) and (c) (an “action for relief from…” instead of “based upon”) simply reflects the phraseology used in the Committee’s discussion, which was lifted verbatim from the report by the Parliamentary draftsman. There is no indication in the report itself that the difference was thought to be significant.
    182. It is fair to say that there are cases decided in equity before the Limitation Act 1939 where the court does not seem to have asked itself whether the mistake was the foundation of the cause of action. Brooksbank v Smith itself was one of them. Denys v Shuckburgh (1840) 4 Y& CEx 42, also decided by Baron Alderson, was another. In both cases, the reason for this appears to have been that Baron Alderson was trying to apply the equitable rule about fraudulent concealment (corresponding to section 31(1)(b)) by analogy to cases of mistake, by holding that a mistake on the part of the plaintiff which concealed from him his right was equivalent to the dishonest or deliberate concealment of his right by the defendant. If so, the idea was still-born. Lord Wright’s committee may well have had these cases in mind when it went out of its way in paragraph 23 of its report to say that they

    “desire[d] to make it clear, however, that the mere fact that a plaintiff is ignorant of his rights is not to be a ground for the extension of time. Our recommendation only extends to cases when there is a right to relief from the consequences, of a mistake.”

    This reservation was adopted by the draftsman of section 26 of the Limitation Act 1939 and the corresponding provision of the 1980 Act, both of which exclude from the ambit of paragraph (b) cases where the claimant was mistaken about the existence of his right. There are clearly obscurities about how the old rule in equity operated before statute intervened, attributable at least in part to the absence of analysis in the few reported cases. But there is, as it seems to me, no difficulty in ascertaining what rule the Law Revision Committee thought that it was proposing to Parliament.

    183. Nor, in my view, is there any real difficulty in understanding what Parliament must have intended by accepting that proposal when it enacted section 26(c) of the 1939 Act. The point has been directly considered only once, by Pearson J in Phillips-Higgins v Harper [1954] 1 QB 411. That was an action by an assistant solicitor to enforce a term of her contract of employment which entitled her to a share of the profits of the firm for which she worked. She claimed to have been underpaid under the profits agreement for the whole 13 years of her employment. In response to a plea of limitation in respect of the early years, she contended that she had been mistaken in failing to realise that she was being underpaid, and relied on section 26(c) of the Limitation Act 1939. Pearson J rejected her argument. In his view the wording of the provision was “carefully chosen to indicate a class of actions where a mistake has been made which has had certain consequences, and the plaintiff seeks to be relieved from those consequences” (p 418). He gave as examples an action for the restitution of money paid in consequence of a mistake; or for the rescission or rectification of a contract on the grounds of mistake; or an action to reopen accounts settled in consequence of a mistake. Mrs. Phillips-Higgins’s alleged mistake had no consequences relevant to her cause of action. Its only consequence was that because she was unaware that she had a cause of action she missed the limitation period. “But that is not sufficient”, said Pearson J; “Probably provision (c) applies only where the mistake is an essential ingredient of the cause of action, so that the statement of claim sets out, or should set out the mistake and its consequences and pray for relief from those consequences” (p 419). It is fair to say about this reasoning that Mrs. Phillips-Higgins would have failed even on Mr Rabinowitz’s construction of the Act, because the mistake that she alleged was not the cause of the factual situation which she relied on for her claim. It only explained why she had allowed so long to pass before bringing her action. But what matters for present purposes is that her argument failed because her action was an action for relief from a breach of contract, to which the fact that she was mistaken was legally irrelevant. As Pearson J went on to point out, “No doubt it was intended to be a narrow

    provision, because any wider provision would have opened too wide a door of escape from the general principle of limitation.”

    184. I think that it is difficult to fault Pearson J’s succinct and principled analysis of the point. Section 32(1)(c) refers to a type of action and a type of relief. They are assumed to be organically related to the relevant mistake. But if the Test Claimants are right, there is no organic connection, but only an adventitious one. The result would be a state of the law that would operate quite arbitrarily. Some Woolwich Equitable claims would benefit from the extended limitation period while others would not, depending on whether the underlying facts arose from a mistake. I can see no principled ground for making such a distinction in a context where the mistake has no bearing on the nature of the action or the relief claimed.
    185. It has been suggested by academic commentators that this result may be anomalous, in that the extended period of limitation applies to a claim to recover a mistaken overpayment of a debt but not to a claim to recover a mistaken underpayment. Pearson J himself drew attention to this in his judgment in Phillips- Higgins at p 419. But for my part, I do not see the anomaly. The difference simply arises from the fact that if the claimant is underpaid and sues for the balance, he is enforcing the obligation that gave rise to the debt, whereas if he is overpaid then that obligation will have been discharged, so that he needs some other legal basis for getting it back. By comparison, there are far graver anomalies associated with the wider construction proposed by the Test Claimants. Once one departs from a construction of the subsection which requires the cause of action to be founded on the mistake, it is difficult to discern any principled limit to the reach of this provision. Mr Rabinovitz distinguishes between cases where the mistake, albeit legally irrelevant, was an effective cause of the facts giving rise to the claim and cases where it was merely a background fact. I find this distinction conceptually difficult to grasp and almost impossible to apply. Questions of causation are notoriously difficult and highly sensitive to the legal context in which they fall to be answered. Where parties have fallen out, there is very likely to be mistake on the part of the claimant somewhere in the chain of events that led to his losing money or property. If at some stage he could have done something to save himself from loss, in what circumstances is that to be a sufficient causal nexus between the legally irrelevant mistake and the legally relevant facts which give rise to the claim? The question will often be incapable of a clear answer. Moreover, if the test is not to depend on whether the claimant is asserting one of the established grounds of relief from the consequences of his mistake, and depends on the mere fact that a mistake has brought about the situation in which he has a claim, then there is nothing in the language or purpose of the provision which would limit it to his own mistakes. It could be the defendant’s mistake against whose consequences the plaintiff is seeking to be relieved, for example by an action for damages. This would mean that section 26(c) of the Limitation Act 1939 unwittingly covered at least part of the ground which Parliament later covered by providing an extended

    limitation period for actions for damages for negligence or in respect of personal injuries and certain categories of property damage: see sections 11 to 14B of the Limitation Act 1980. Mr Rabinowitz disclaimed any suggestion that the extended limitation period would apply to a claim for damages, with the possible exception of damages for misrepresentation or negligent mis-statement. This was no doubt tactically wise. But it is hard to see how such a restriction can be justified if his basic submission is accepted. The difficulties associated with the claimants’ construction of section 32(1)(c) persuade me that Lord Wright is unlikely to have proposed such an indefinite rule without any discussion of these problems, and that Parliament is unlikely to have intended to enact it. In an ideal world, all rules of law would be clear, but there are few areas where clarity is as important as it is in the law of limitation, whose whole object is to foreclose argument on what ought to be well-defined categories of ancient dispute.

    Mistake

    186. It follows that the principle in Woolwich Equitable applies generally in all cases where tax has been charged unlawfully, whether by the legislature or by the tax authorities, whether by overt threats or demands or simply by the taxpayer’s appreciation of the consequences of not paying, and whether the taxpayer was mistaken or not. By comparison, an action for restitution on the ground of mistake is a more limited remedy, for the obvious reason that it is necessary to prove the mistake. That will not always be easy, as the facts of Woolwich Equitable itself demonstrate. On the face of it, the only case where the Woolwich Equitable cause of action is probably not available and where a claimant may therefore need a right of restitution for mistake, is the case where there is no unlawful exaction of tax but the taxpayer has simply paid in error: e.g. he has miscalculated his liability under a self-assessed tax or accidentally paid twice. But that has no bearing on the position of the present claimants.
    187. Does this mean that that the existence of the Woolwich Equitable cause of action in English law is enough to satisfy the obligations of the United Kingdom in EU law? The Test Claimants submit that it does not. Their case is that notwithstanding the sufficiency of a Woolwich Equitable claim as a means of recovering unlawfully charged tax, at least in the circumstances of the present case, EU law requires that English law should also maintain a fully effective cause of action to recover tax paid by mistake. Two quite different arguments are advanced in support of this proposition. The first is that EU law specifically requires that national legal systems should provide for the recovery of overpaid taxes in all cases where they were not due, including the one case where the principle in Woolwich Equitable probably has no application, viz where there is no breach of EU law by the state but the taxpayer has simply overpaid by mistake. I shall call this the “absence of basis point”. The second argument is that even if EU law does not specifically require national law to confer a right to recover taxes

    overpaid on the ground of mistake, if national law allows a choice between two causes of action to recover the tax, each of them must be independently effective. I shall call this the “choice of remedies point”.

    The absence of basis point

    188. The Test Claimants’ argument is that the obligation of a member state to provide an effective means of recovering overpaid taxes is not limited to cases where the state was in breach of EU law. It also applies in cases where the national law entirely conformed with EU law but the claimant paid more than the law required of him. This, they submitted, reflected the principle of restitution applied in EU law and in most civil law jurisdictions (but not England) that a payment is recoverable merely on account of the absence of a legal basis for making it: see Masdar (UK) Ltd v Commission of the European Communities (Case C-47/07) [2008] ECR I-9761, paras 44-46, 49.
    189. In Reemtsma Cigarettenfabriken GmbH v Ministero delle Finanze (Case C‐35/05) [2007] ECR I-2425 a German company purchased services from an Italian advertising agency and paid VAT to them which was not due. There was nothing wrong with the relevant provisions of Italian law for charging and collecting the tax, which in the relevant respects entirely conformed with the Directives. The Italian tax authorities had charged no tax unlawfully. All that happened was that the German purchaser received an invoice from the Italian supplier for the VAT and paid it, not appreciating that the relevant services were by law deemed to have been supplied in Germany. The supplier then accounted for the tax to the Italian tax administration. There was no provision of the two relevant VAT Directives requiring a refund to be made in these circumstances, but it was held that the principle of effectiveness required Italy to make available an effective means of recovering sums paid but not due, either from the Italian supplier or from the state. Mr Aaronson QC argued that the juridical basis for the obligation to repay overpaid tax in these circumstances was the mere absence of a legal basis for the original payment. I think that he may well be right about that. But the reason for the decision was that VAT is an EU tax whose incidence and administration is governed by mandatory requirements of EU law. The purpose of the VAT Directives is to produce a harmonized system operating according to uniform rules across the EU. The payment of VAT otherwise than in accordance with that scheme distorts its uniform operation. The point was made in Danfoss AS and Sauer-Danfoss ApS v Skatteministeriet (Case C-94/10), 20 October 2011, where a similar result was arrived at in the context of the common EU scheme for taxing mineral oils. In its judgment in that case, the court observed (para 23) that the purpose of a right of recovery in a harmonized tax scheme is not only compensatory but economic. “The right to the recovery of sums unduly paid helps to offset the consequences of the duty’s incompatibility with EU law by neutralising the economic burden which that duty has unduly imposed on the

    operator who, in the final analysis, has actually borne it.” In those circumstances, a right of action to recover money paid but not due is required so “that the economic burden of the duty unduly paid can be neutralised” (para 25).

    190. If this issue were to arise in England in the context of an EU tax, the case would be classified in English law as one of mistake and recovery could probably be had on no other basis. But where the relevant tax is wholly a creature of national law, and no tax has been charged in breach of EU law, EU law is not engaged at all.

    The choice of remedies point

    191. This point is at the heart of the division of opinion within this court. The Test Claimants argue, and the majority agrees, that the principle of effectiveness in EU law requires that all remedies which are available to recover the tax should be independently effective for that purpose. Therefore, so the argument goes, it was not open to the United Kingdom to compromise the effectiveness of the right to recover on the ground of mistake by curtailing the limitation period for that right without a period of grace.
    192. In argument, this point was founded mainly on the decision of the European Court of Justice in Rewe Handelsgesellschaft Nord mbH v Hauptzollamt Kiel (Case 158/80) [1982] 1 CMLR 449 (Rewe II). This was another case about VAT and excise duty chargeable under the terms of a Directive. It concerned not an unlawful charging of tax, but an unlawful exemption from tax. The claimants were companies operating supermarkets in German coastal towns, who were adversely affected by tax-free sales made in international waters during shopping cruises in the Baltic which began and ended in Germany. Under the terms of the Directives, a limited exemption was allowed for goods coming from member states in the personal luggage of travellers, but German law allowed an exemption of its own which was in some respects wider. The Court of Justice held that the exemption in the Directive was not available for sales made on shopping cruises beginning and ending in the same member state, that the tax ought to have been charged, that the incidence of VAT and excise duty was an occupied field governed exclusively by Community law, and that Germany had accordingly had no power to grant further exemptions of its own. The relevant question for present purposes concerned the remedies available to rival traders against the cruise operators. German law allowed a right of action to those adversely affected by breaches of national laws regulating economic activity. At para 40 of its judgment, the Court of Justice referred to this German right of action and then summarised the question at issue as follows:

    “Placed in that context, the questions raised by the national court are intended in substance to establish whether that right of action may be exercised in similar conditions within the framework of the Community legal system in particular in the sense that if the economic interests of a person to whom Community law applies are adversely affected by the non-application of a Community provision to a third party, either through the action of a member state or of the Community authorities, that person may institute proceedings before the courts of a member state in order to compel the national authorities to apply the provisions in question or to refrain from infringing them.”

    The court’s answer to that question appears at para 44 of the judgment:

    “it must be remarked first of all that, although the Treaty has made it possible in a number of instances for private persons to bring a direct action, where appropriate, before the Court of Justice, it was not intended to create new remedies in the national courts to ensure the observance of Community law other than those already laid down by national law. On the other hand, the system of legal protection established by the Treaty, as set out in article 177 in particular, implies that it must be possible for every type of action provided for by national law to be available for the purpose of ensuring observance of Community provisions having direct effect, on the same conditions concerning the admissibility and procedure as would apply were it a question of ensuring observance of national law.”

    193. In their printed case (paragraph 67) the Test Claimants rely on this statement of principle, and in particular the passage which I have italicised, as authority for the proposition that EU law requires a “right to choose from the range of national remedies.” Of course the Test Claimants do have a right to choose either or both of a Woolwich Equitable claim or a claim based on mistake. Neither of the Acts of 2004 and 2007 took it away from them. Their real complaint is not that the right to claim on the basis of mistake of law has been withdrawn, but that the law has been changed to make it subject to a period of limitation running from the date of payment in the same way as the limitation period for a Woolwich Equitable claim. The argument, as it was developed at the hearing, was that even on the footing that a Woolwich Equitable cause of action was enough and that the United Kingdom was not obliged to confer an additional right to recover tax paid by mistake subject to an extended limitation period, since it has chosen to do so, the principle of effectiveness requires that that right with its extended limitation period should remain available for the purpose of recovering tax charged contrary

    to EU law. This submission is accepted by the majority on the present appeal. I regret that I am unable to accept it for three reasons.

    194. First, the argument is not supported by either the decision or the reasoning in Rewe II, nor by the many subsequent cases in which the relevant statement has been cited. Rewe II was concerned with the principle of equivalence, as the language and the legal context show. The issue was whether Germany was bound to make a right of action derived from economic regulation under its national law available to litigants who wanted to enforce comparable rights derived from economic regulation under Community law. What the court was saying was that any cause of action available to enforce a national law right must be equally available to enforce a corresponding Community law right. Provided that there remains an effective remedy, it does not follow from this that national law is bound to maintain that cause of action subject to unchanged incidents or conditions. Nothing was said in Rewe II about protecting the choice of litigants between concurrent national law rights or remedies. The question did not arise because the Court of Justice was considering the only German law remedy which appeared to exist.
    195. Second, the Test Claimants’ submission is inconsistent with the established case law of the Court of Justice. In Edilizia Industriale Siderurgica Srl v Ministero delle Finanze (Case C-231/96) [1998] ECR I-4951 and Ministero delle Finanze v SPAC (Case C-260/96) [1998] ECR I-4997, para 32, the facts were that in breach of a Directive which prohibited taxes on the raising of capital, Italy had charged fees for registering companies. The general limitation period under the Civil Code was ten years, but the decree-law authorizing the registration fees provided (and always had provided) for their repayment within three years if they had been wrongly charged. The Italian courts had held that as a matter of domestic law, the effect of the creation of a specific right to repayment within three years under the decree-law was to displace the general right conferred by the Civil Code to claim restitution on the ground of absence of basis within ten years. One of the questions referred was whether Italy was bound to make available the cause of action with the more generous limitation period for the purpose of giving effect to EU law rights. The court held that it was not. Provided that the right of action carrying the more restrictive limitation period was effective and applied without discrimination whether the claim to repayment was based on EU or national law, there was no obligation to provide in addition a right of action under the Civil Code with a more generous limitation period. In Edilizia Industriale Siderurgica Srl v Ministero delle Finanze, the court said:

    “36. Observance of the principle of equivalence implies, for its part, that the procedural rule at issue applies without distinction to actions alleging infringements of Community law and to those alleging infringements of national law, with respect to the same kind

    of charges or dues (see, to that effect, Amministrazione delle Finanze dello Stato v Salumi (Joined Cases 66/79, 127/79 and 128/79) [1980] ECR 1237, para 21. That principle cannot, however, be interpreted as obliging a member state to extend its most favourable rules governing recovery under national law to all actions for repayment of charges or dues levied in breach of Community law.

    37. Thus, Community law does not preclude the legislation of a member state from laying down, alongside a limitation period applicable under the ordinary law to actions between private individuals for the recovery of sums paid but not due, special detailed rules, which are less favourable, governing claims and legal proceedings to challenge the imposition of charges and other levies. The position would be different only if those detailed rules applied solely to actions based on Community law for the repayment of such charges or levies.”

    The same observations were made in Ministero delle Finanze v SPAC SpA, at paras 20 and 21. They were later repeated and applied in Aprile II and Dilexport, where the facts were very similar (see paras 151-152 above) but the question arose from a change in the law.

    196. Third, the Test Claimants’ argument is contrary to principle. The starting point for any analysis of the law in this area is that, subject to the principles of effectiveness and equivalence, it is for national law to determine what remedies are available to enforce a directly effective EU right and on what procedural or other conditions. I have made this point already: see paragraph 145 above. The right of the claimants to choose from the range of causes of action recognised by English law is a right derived solely from English procedural law and it exists only to the extent that English law so provides. So long as the principles of effectiveness and equivalence are respected, a choice between concurrent national law remedies need not exist, and in some member states does not exist, at any rate to the same extent. Thus English law allows a claimant to choose between concurrent rights of action in contract and tort, a principle which was applied by analogy in Deutsche Morgan Grenfell [2007] 1 AC 558 to allow a choice between concurrent rights to recover under the Woolwich Equitable principle and on the ground of mistake. French law, by comparison, is more prescriptive. The principle of non-cumul des responsabilités, which excludes delictual claims which fall naturally within the scope of a contract is generally thought to reflect a more general juristic preference for keeping legal categories distinct and allowing claims to be brought in the category to which their subject matter is appropriate. The same approach appears to lie behind the restriction of claims under the general doctrine of unjust enrichment (enrichissmement sans cause légitime) to cases where no other action is available: Flour, Aubert et Savaux, Droit civil, Les obligations, 2 Le fait

    juridique, 11^th ed. (2011), 57-64. I can see no principled reason why EU law should wish to control these divergent features of national legal systems, provided that the choice which the relevant law mandates and the conditions on which it does so are non-discriminatory and effective to vindicate EU rights.

    The protection of legitimate expectations: Finance Act 2004, section 320

    197. I have already analysed the case law of the Court of Justice on the retrospective curtailment of limitation periods for the exercise of directly effective EU law rights. It establishes, first, that the retrospective curtailment of a limitation period is not necessarily inconsistent with the principle of effectiveness; and, secondly, that the combined effect of the principle of effectiveness and the principle of the protection of legitimate expectations is to preclude national legislatures from retrospectively curtailing the limitation period applicable to the recovery of overcharged tax, unless there is a sufficient period of grace to enable actual and potential claimants to safeguard their existing rights. However, it is important to note that in every case in which these principles have been considered by the Court of Justice, the amending legislation curtailed the limitation period for the only right available in national law for recovering the tax. In none of them was there an effective right of recovery on another legal basis, unaffected by the amendment. The observations of the Advocate-General and the court, especially those made in Marks & Spencer must be read in that light.
    198. The primary case put forward on behalf of the Commissioners is that because (i) English law would be compatible with EU law if the only means of recovering the overpaid tax was a claim on a Woolwich Equitable basis, and (ii) the Finance Act 2004 did not affect a claim on that basis, it follows that the principles of effectiveness and the protection of legitimate expectations are not engaged at all. In common with every other member of the court, I reject that submission. The reason is that if, as I have sought to demonstrate (i) a right to claim on the principle in Woolwich Equitable with a normal limitation period is an effective means of asserting the Test Claimants’ EU law right, and (ii) there is no obligation on the United Kingdom in EU law to maintain a concurrent right to claim on the basis of mistake with an extended limitation period, then logically there still remains one complaint that might arguably be made about section 320 of the Finance Act 2004. That complaint is that before the intention to legislate was announced potential claimants were entitled to make their plans on the assumption that they could recover the overpaid tax on the ground of mistake with the benefit of an extended limitation period, but their right to do so was then curtailed without notice or transitional provisions. I think that this complaint depends on the principle of the protection of legitimate expectations, whereas Lord Walker and Lord Reed consider that it can be justified on the basis of the principle of effectiveness alone. I doubt whether this difference matters. In either case, the

    force of the complaint depends entirely on the proposition that reasonable persons in their position could have made their plans on that assumption.

    199. Could they? I think not. If English law had never recognised a right to recover tax on the ground of mistake of law, but only on the basis of the principle in Woolwich Equitable, it is not disputed that that state of affairs would have satisfied the requirements of EU law. If Parliament had retrospectively created a concurrent right to recover tax on the ground of mistake of law, but in the same enactment made it subject it to a limitation period of six years to run from the time of payment, it is not disputed that that state of affairs would also have satisfied the requirements of EU law. The question whether the right to recover money paid under a mistake of law extended to mistaken payments of tax was a difficult question. There were powerful voices raised in favour, such as that of Professor Birks, but also strong and principled arguments against. I have dealt with this matter at paragraphs 166-168 above. Before Park J gave judgment in Deutsche Morgan Grenfell [2003] 4 All ER 645 on 18 July 2003, no one could reasonably have counted on being able to recover tax on the ground of mistake of law. They might have thought that there were strong arguments to that effect, but I do not believe that they could reasonably have assumed when deciding how long they had in which to bring their claims that those arguments would prevail. Even after Park J’s judgment, the right to recover tax on the ground of mistake of law was being challenged on appeal on serious grounds. The existence of such a right was rejected by the Court of Appeal [2006] Ch 243 and was not definitively established until the judgment of the House of Lords [2007] 1 AC 558 on 25 October 2006.
    200. In a common law system, it is open to the courts to create new causes of action, but limitation is necessarily a matter for the legislature. On 8 September 2003, just seven weeks after the decision of Park J, the government announced its intention to introduce what became section 320 of the Finance Act 2004, with its provision that the limitation period for the newly recognised claim to recover tax on the ground of mistake of law should run from the date of payment and not from the date of discovery. I find it impossible to regard that sequence of events as any different in substance from the situation that would have existed if Parliament had simultaneously created a right to recover tax for mistake of law and subjected it to a limitation period running from the date of payment. If potential claimants in the position of the present appellants claim to have been entitled to count on being able to recover on the ground of mistake of law with an extended limitation period, then the highest that they can put their case is that they were entitled to do so in the seven week interval between 18 July and 8 September 2003. Bearing in mind the brevity of the interval, the virtual certainty of an appeal and the uncertainty about its outcome, the argument that they had a legitimate expectation of the kind suggested seems to me to be unrealistic. In my judgment, section 320 of the Finance Act 2004 was not inconsistent with the protection of legitimate expectations. All that Parliament did was to provide for the limitation period

    applicable to a cause of action which English law had only just recognised. This was a lawful exercise by Parliament of the discretion allowed to member states as to the conditions regarding limitation on which any national law right is be available.

    201. The contrary view of the majority depends on the declaratory theory of judgments. It proceeds upon the basis that when Park J and then the House of Lords held in Deutsche Morgan Grenfell that there was a right to recover tax on the basis of mistake, they were declaring the law as it had always been. At a purely formal level, this proposition is undoubtedly correct. Judgments of the courts about the common law are deemed to be declaratory and not legislative. But we are, I think, in danger of allowing the form to overlay the substance. In Deutsche Morgan Grenfell, at p 570, Lord Hoffmann distinguished between two questions raised by the declaratory theory of judgments:

    “One is whether judges change the law or merely declare what it has always been. The answer to this question is clear enough. To say that they never change the law is a fiction and to base any practical decision upon such a fiction would indeed be abstract juridical correctitude. But the other question is whether a judicial decision changes the law retrospectively and here the answer is equally clear. It does. It has the immediate practical consequence that the unsuccessful party loses, notwithstanding that, in the nature of things, the relevant events occurred before the court had changed the law: see In re Spectrum Plus Ltd [2005] 2 AC 680. There is nothing abstract about this rule.”

    In my judgment, it is the first of Lord Hoffmann’s propositions which is relevant for present purposes. The question is not whether the law must be treated as always having been as Park J and the House of Lords declared it to be. It is whether before those judgments were delivered a litigant could reasonably count on being able to recover the overpaid tax on the ground of mistake (with an extended period of limitation), as opposed to being limited to the already established remedy under the Woolwich Equitable principle (with a normal period of limitation). The question must in my judgment be put in this way, because the issue is whether there is an assumption reasonably to be imputed to litigants about how long they had in which to bring their claim, which was then retrospectively falsified by Parliament. The answer to the question cannot depend on any legal fiction. It must depend on the position as it appeared to stand, before those judgments were given. This must in particular be true when one is seeking to apply to the relevant English law principles of EU law which have always depended on substance rather than form. The reality is that the Test Claimants never were in a position to make their plans on the footing that they had a right of action for

    mistake until at the very earliest the judgment of Park J, but more realistically until the matter was definitively settled by the House of Lords in 2006.

    202. It is right to point out that this is substantially the same principle as that on which the Test Claimants themselves rely when they say (with the support of the House of Lords in Deutsche Morgan Grenfell) that they cannot be taken to have discovered their mistake about the lawfulness of the United Kingdom’s corporation tax regime until the European Court of Justice definitively decided the point. By the same token, the Test Claimants cannot be taken to have assumed that they had a right to recover the tax on the ground of mistake at a stage when they had arguments and hopes but no definitive decision.

    The protection of legitimate expectations: Finance Act 2007, section 107

    203. As I have already indicated, I regard this provision as more problematic. It was announced on 6 December 2006, more than three years after the announcement which preceded section 320 of the Act of 2004. It went a great deal further than the earlier enactment, since it applied retrospectively without limit of time to any action brought before the first announcement had been made on 8 September 2003. It might be said that the announcement of 2006 was a response to the decision of the House of Lords in Deutsche Morgan Grenfell and that the interval between judgment and announcement was no greater than it had been in 2003. But the circumstances were different. Companies in the position of the British American Tobacco group who had already brought their actions before the announcement of 8 September 2003 had been expressly excluded from the operation of the legislation proposed on that date. That exclusion was duly contained in section 320 of the Finance Act 2004. The British American Tobacco group and other companies in the same position had been pursuing their claims through the English courts and the Court of Justice on that basis since 2003, when their right to the fruits of those proceedings was removed in 2006. In my view, while they had had no legitimate expectation of being able to bring an action to recover on the ground of mistake of law in 2003, they had acquired such an expectation by 2006, not least as a result of the terms of the announcement of September 2003 and the 2004 Act. It was therefore contrary to the principle of the protection of legitimate expectations, for that expectation to be defeated without notice of transitional provisions.

    Section 33 of the Taxes Management Act 1970

    204. This provision applies only to assessed taxes, and therefore only to a very small part of the present claims. It confers a right subject to highly restrictive conditions to invoke what is essentially a discretionary power of the

    Commissioners to grant a refund of overpaid tax. No one suggests on this appeal that such a limited remedy could possibly be enough in itself to satisfy the virtually unqualified obligation of the United Kingdom to provide an effective means of recovering tax overcharged contrary to EU law. This does not of course matter if it is an additional remedy as opposed to an exclusive one. There is certainly nothing in the provision which expressly excludes the availability of other causes of action at common law. If that is its effect, it must be by implication. In the ordinary way, such an exclusion might be implied, on the ground that where Parliament confers a restricted right of recovery, that must impliedly displace a corresponding right at common law which would be unrestricted. However, it is axiomatic that the courts cannot imply an exclusion of unrestricted rights of action at common law where that would be inconsistent with an overriding rule of EU law that an unrestricted right must be available. Section 33 cannot therefore be an exclusive right to recover tax overcharged contrary to EU law. Whether it is an exclusive right in other circumstances, is not a point which needs to be considered on this appeal.

    205. The Court of Appeal held that section 33 did impliedly exclude a right of action at common law, even in relation to claims for tax overcharged contrary to EU law. They then dealt with the resulting inconsistency with EU law by reinterpreting the section so as remove the offending restrictions and the element of discretion. I think that this was wrong in principle. I very much doubt whether such radical surgery can be justified even under the extended principles of construction authorised in Marleasing. Its effect would be fundamentally to alter the scheme of the provision. But, however that may be, it seems, with respect, eccentric to imply an ambit for section 33 which is inconsistent with EU law and then to torture the express provisions so as to deal with anomalies that but for the implication would never have arisen.

    The damages claims

    206. In addition to their claims in restitution, the claimants have claims against the Commissioners in damages on the principle of state liability adopted by the European Court of Justice in Francovich v Italian Republic (Cases C-6 and 9/90) [1995] ICR 722. This cause of action is subject to a number of conditions, one of which is that the breach should be sufficiently serious, ie should involve a “grave and manifest” disregard of the limits of the member state’s discretion: see Brasserie du Pecheur SA v Federal Republic of Germany (Joined Cases C-46/93 and C-48/93) [1996] QB 404. Both courts below have dismissed the claim for damages on the ground that that condition is not satisfied. That may explain why, although the issues before us were formulated so as to cover their implications for the damages claim also, the argument focused exclusively on the claim for restitution. In fact, the damages claims do not call for separate consideration because neither section 320 of the Finance Act 2004 nor section 107 of the Finance Act 2007 applied to those claims unless they fall within section 32(1)(c) of the

    Limitation Act 1980. It follows from the construction that I would give to that provision that they do not fall within it. It is not suggested that section 33 of the Taxes Management Act 1970 has any bearing on a claim for damages on the principle of state liability.

    Conclusion

    207. In the result, I would
    1. affirm the decision of the Court of Appeal on the requirements of the cause of action based on Woolwich Equitable and the absence of any requirement for an additional remedy in mistake (Issue 12 in their numbering);
    2. affirm their decision on the effect of section 32(1)(c) of the Limitation Act 1980) (Issue 22);
    3. allow the appeal on section 107 of the Finance Act 2007 (Issues 20 and 21); and
    4. allow the appeal on section 33 of the Taxes Management Act 1970 (Issue 23).
    208. The question whether section 320 of the Finance Act 2004 is compatible with EU law cannot be decided without a reference to the Court of Justice. It is plain from the novelty of the circumstances in which it arises, and from the differences of opinion within the court that it is not acte clair. I would, however, limit the reference to section 320 of the 2004 Act.

    LORD REED

    209. Lord Walker and Lord Sumption have expressed different views about the way in which EU law applies to the grounds of action available to the test claimants for the recovery of taxes which were levied contrary to EU law, and in particular about the way in which EU law applies to legislation which shortened, retroactively and without transitional provisions, the limitation period applicable to one of those grounds of action. In my opinion, Lord Walker’s analysis of the compatibility of section 320 of the Finance Act 2004 and section 107 of the

    Finance Act 2007 with the principle of effectiveness, and of the compatibility of the latter provision with the principle of the protection of legitimate expectations, is consistent with the relevant case law of the Court of Justice of the European Union. I agree with his reasoning and conclusions in relation to those issues, as well as in relation to the issues of domestic law before the court. For my part, in agreement with Lord Hope and Lord Clarke, I am inclined to the view that section 320 of the 2004 Act also infringes the principle of the protection of legitimate expectations. I add some observations of my own in relation to the issues of EU law only because of the importance of those issues and the division of opinion in the court. It is perhaps unusual to discuss EU law in such detail when the matter is to be referred to the Court of Justice, but in the present case the issues of EU law and domestic law are closely inter-related.

    210. The difficulties in this case arise partly from the fact that the relevant principles of English law have been in the course of development during much of the relevant period of time. The principal milestones along the road are three decisions of the House of Lords. First, in 1992 the House of Lords held that a taxpayer was entitled to recover taxes paid in response to an unlawful demand: Woolwich Equitable Building Society v Inland Revenue Comrs [1993] AC 70. Secondly, in 1998 the House of Lords held that money paid under a mistake of law was recoverable: Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349. Thirdly, in 2006 the House of Lords held that the latter principle applied to taxes paid under a mistake of law, including taxes paid in ignorance of the fact that the legislation under which they were levied was incompatible with EU law: Deutsche Morgan Grenfell Group plc v Inland Revenue Comrs [2007] 1 AC 558 (“DMG”). Two other important matters were also decided in that case. The first concerned the limitation period applicable to the claim. In terms of section 32(1)(c) of the Limitation Act 1980, that period would not begin to run until the mistake was discovered, or could with reasonable diligence have been discovered. The House of Lords held that, in the circumstances of the case, the mistake could not be discovered until the incompatibility of the tax with EU law had been established by a judgment of the Court of Justice. The second matter was that the fact that the taxpayer might have a concurrent ground of action under the Woolwich principle, which was subject to a limitation period running from the date of the payment, did not prevent it from pursuing its claim on the ground of mistake if the extended limitation period best suited its interests. Finally, in its present decision this court has held that a taxpayer who pays taxes in compliance with legislation which is incompatible with EU law has a ground of action under the Woolwich principle, in addition to any ground of action which may be available on the basis of mistake.
    211. The legislative provisions with which we are now concerned alter the limitation period applicable to claims for the repayment of taxes on the ground of mistake, so that it runs from the date when the payment was made, rather than the date when the mistake was discovered or could reasonably have been discovered.

    The first provision with which we are concerned, section 320 of the Finance Act 2004, applies to claims which were made on or after 8 September 2003. The second provision, section 107 of the Finance Act 2007, applies to claims made before that date. The claims with which we are concerned were made on 18 June 2003, in the case of the BAT group claimants, and on 8 September 2003, in the case of the Aegis group claimants. They were based on both grounds of action. The principal issue we have to determine is whether the application of the legislation to the claims is compatible with EU law.

    212. In considering that issue, there appear to me to be three central questions, which can at this stage be broadly stated as follows. The first is whether the ground of action enabling taxes levied in breach of EU law to be recovered on the basis of mistake falls within the ambit of the EU principle of effectiveness. It is argued that it does not, since the ground of action based on an unlawful demand in itself fully satisfies the requirement of EU law that there should be an effective remedy. Since no additional remedy is required by the principle of effectiveness, it follows, so the argument runs, that the additional ground of action which English law provides, based on mistake, falls outside the scope of that principle. I disagree. As I shall explain, it appears to me that the EU principle of equivalence, which is the complement of the principle of effectiveness, applies to the grounds of action available for the recovery of taxes in domestic law. Where an action for the recovery of taxes under domestic law can be based either on the ground of mistake or on the ground of unlawful demand (or, as in the present case, on both grounds), it follows from the principle of equivalence that both grounds of action should also be available in similar circumstances to enforce an analogous right under EU law. So long as they must both be available, they must also both be effective. The principle of effectiveness therefore applies to both grounds of action.
    213. The second question, which arises only if the first question is answered in the affirmative, is whether the application of section 320 of the 2004 Act to the Aegis claims, and of section 107 of the 2007 Act to the BAT claims, is compatible with the principle of effectiveness. As I shall explain, I consider that it is not compatible in either case, since the retroactive curtailment of the limitation period and the absence of any transitional provisions rendered impossible in practice the exercise of rights derived from EU law. If that is correct, it follows that the legislation cannot be relied upon against the claimants, whatever the answer to the third question may be.
    214. The third question is whether the application of the legislation to these claims is compatible with the EU principle of the protection of legitimate expectations. That is a question which arises even if the first question is answered in the negative, since the procedural rules laid down by domestic law for the enforcement of rights derived from EU law must be in conformity with the general principles of EU law, including the general principle requiring the protection of

    legitimate expectations. The answer to the third question is however of no practical significance if the first two questions are answered as I would answer them. In the event, we are all agreed that the application of section 107 of the 2007 Act to the BAT claims is incompatible with the protection of the BAT claimants’ legitimate expectations. In agreement with Lord Hope and Lord Clarke, I have also reached the same provisional conclusion in respect of the application of section 320 of the 2004 Act to the Aegis claims, for reasons which I shall explain.

    215. It might be argued that a fourth question also arises on the facts of these cases: namely, whether the application of the legislation in issue to these claims would be compatible with the rights recognised in the Charter of Fundamental Rights of the European Union (OJ 2000 C 364, p 1) (notably in article 47), to which effect is given by article 6(1) of the Treaty on European Union (“TEU”), or with the fundamental rights recognised by article 6(3) TEU, including in particular the right of access to a court, guaranteed by article 6(1) of the European Convention for the Protection of Human Rights and Fundamental Freedoms, and the right to the peaceful enjoyment of possessions, guaranteed by Article 1 of the First Protocol to the Convention. That question however goes beyond the ambit of the dispute as defined by the parties, and it raises issues on which the court has not been addressed. In those circumstances it would not be appropriate for the court to consider that question of its own motion. My answers to the first three questions in any event produce a result which is not incompatible with the fundamental rights just mentioned.
    216. I turn now to consider in greater detail the three questions which I have identified.

    The mistake ground of action and the principles of equivalence and effectiveness

    217. Under the principle of cooperation laid down in article 4(3) TEU, it is for the member states to ensure the effective judicial protection of an individual’s rights under EU law: see, for example, Unibet (London) Ltd v Justitiekanslern (Case C-432/05) [2008] All ER (EC) 453, paras 37-44. In particular, in the absence of EU rules governing the matter, it is for the domestic legal system of each member state to lay down the procedural rules governing actions for safeguarding rights which individuals derive from EU law. In a case such as the present, it may seem idiosyncratic to describe the grounds of action available under domestic law as procedural rules, but that description reflects the distinction drawn in the case law of the Court of Justice between the right derived from EU law and the national law by means of which effect is given to that right, which may govern such matters as the procedure to be followed, the period within which claims must be made, and the proof of such claims.
    218. That general approach applies to the right to recover the taxes in issue in the present case, to the extent that they were levied in breach of EU law: see the judgment of the Grand Chamber on the first reference in these proceedings, Test Claimants in the FII Group Litigation v Inland Revenue Comrs (Case C-446/04) [2006] ECR I-11814, para 203. As the Grand Chamber stated, the procedural rules laid down by domestic law must comply with two conditions. First, they must not be less favourable than those governing similar domestic actions. That is the principle of equivalence. Secondly, they must not render virtually impossible or excessively difficult the exercise of rights conferred by EU law. That is the principle of effectiveness. Equivalence and effectiveness are complementary requirements.
    219. For the purpose of applying the principle of equivalence, a claim for the recovery of taxes levied by a member state in breach of EU law is similar to a claim for the restitution of taxes unlawfully levied under domestic law. In England and Wales, the rules laid down by domestic law governing such claims are in large part rules of common law. The procedure laid down by section 33 of the Taxes Management Act 1970 is an exception. For the reasons given by Lord Walker and Lord Sumption, however, that statutory procedure is not applicable in the circumstances of this case. The relevant rules of common law include those laid down by the House of Lords in the three cases which I have mentioned – Woolwich, Kleinwort Benson and DMG – and by this court in the present case. In particular, as I have explained, it was held in the DMG case that a person who had mistakenly paid taxes which had been levied in breach of EU law had a ground of action based upon the fact that the payment had been made under a mistake: that is to say, the ground of action whose general nature was established in Kleinwort Benson. The present decision holds that such a person also has a ground of action based upon the fact that the payment was made in compliance with legislation which was incompatible with EU law: that is to say, the ground of action whose general nature was established in Woolwich. The two grounds of action are in some respects subject to different rules, and in consequence one or the other may be more suitable to a claimant, depending upon the circumstances. For example, apart from the legislation in issue in the present case, the two grounds of action are subject to different limitation periods. There may be other differences. In the present case, as I have explained, the claims are based upon both the mistake ground of action and the unlawful demand ground of action.
    220. Where both these grounds of action are available for the recovery of taxes which have been levied in breach of domestic law, and a person seeking to recover such taxes can choose to base his claim upon whichever ground of action best suits his interests, it follows from the principle of equivalence that the same grounds of action, and the same freedom of choice, must equally be available in analogous circumstances to a person seeking to recover taxes which have been levied in breach of EU law: otherwise, claims based on EU law would be less favourably

    treated than similar claims based on domestic law. As the Court of Justice stated in Rewe-Handelsgesellschaft Nord mbH v Hauptzollamt Kiel (Case 158/80) [1981] ECR 1805, para 44, the system of legal protection established by the Treaties implies that “it must be possible for every type of action provided for by national law to be available for the purpose of ensuring observance of Community provisions having direct effect” (emphasis added).

    221. It might however be argued that a complication arises from the fact that it had not been definitively decided at the time when the claims were made, or at the time when the legislation was enacted, that those grounds of action were available for the bringing of claims such as those with which the present proceedings are concerned. Does that make a difference to the way in which the principle of equivalence applies? In my view it does not. The decision of the House of Lords in DMG, confirming the soundness of a claim to the repayment of unlawfully levied tax on the basis of a mistake in law, was in no sense prospective only. The decision of this court in the present case, confirming that claims to the repayment of unlawfully levied tax can be made on the basis of the Woolwich principle even in the absence of a formal demand, has similarly determined what the law was at the time when the claims were made. Although each of those decisions determined a question of law which was previously contestable, and can therefore be said to have involved a development of the law, they cannot be equiparated to legislation: such decisions actually, and not merely formally, declare the law that is applicable to the case before the court and all other comparable cases. As Lord Goff of Chieveley explained in Kleinwort Benson at pp 378-379, the declaratory theory of judicial decision is not an aberration of the common law, but reflects the nature of judicial decision-making (an aspect which is also reflected in the temporal effects of the judgment of the Grand Chamber on the first reference in these proceedings). It follows that these claims, although made in proceedings which commenced prior to the decisions of the House of Lords in DMG and of this court in the present case, are based on grounds of action which were available under English law at the time when the claims were made, as a means of recovering taxes which had been unlawfully levied, even if that could not have been known with certainty until the matter had been finally determined by the highest courts.
    222. It accordingly appears to me that the grounds of action based on mistake and on an unlawful demand were both available at all material times, in the circumstances laid down in the relevant case law, for the recovery of taxes which had been levied contrary to domestic law. It follows from the principle of equivalence that both grounds of action must also have been available in analogous circumstances for the recovery of taxes levied contrary to EU law. That is not, of course, to say that English law was bound to maintain both grounds of action subject to unchanged incidents or conditions; but any changes would have to comply with the requirements of EU law, including the requirement of effective judicial protection.
    223. That conclusion is challenged on the basis that the mistake ground of action is neither necessary nor sufficient to meet the requirements of EU law, as laid down in such cases as Amministrazione delle Finanze dello Stato v SpA San Giorgio (Case 199/82) [1983] ECR 3595: it is not necessary, since the unlawful demand ground of action is in itself adequate; and it is not sufficient, since it requires the presence of an additional element besides the levying of the taxes in breach of EU law, namely that they must have been paid under a mistake as to the lawfulness of the domestic legislation.
    224. The first of these contentions appears to me to be off the point.   The fact that the ground of action based on an unlawful demand satisfies the San Giorgio principle does not exclude the possibility that the ground of action based on mistake also satisfies that principle. Indeed, the ground of action based on mistake is of considerable practical importance as a means of enforcing rights to repayment derived from EU law, as the present case demonstrates, since it enables claims relating to taxes levied in breach of EU law to be brought outside the six year limitation period, reckoned from the date of the payment, which applies to claims based upon the Woolwich principle: a period which may have expired before the mistake as to the validity of the tax legislation is discovered. Admittedly, if English law had evolved differently, and the ground of action based on mistake had not been available, then the ground of action based on an unlawful demand might well have met the requirements of EU law. The fact of the matter, however, is that English law provides two grounds of action which are capable of satisfying the San Giorgio principle, and the principle of equivalence therefore requires that both grounds of action should be available for the enforcement of rights derived from EU law.
    225. The second contention also appears to me to be mistaken. The two grounds of action are not identical: in particular, subject to the legislation at issue in the present case, they are subject to different limitation periods. The mistake ground of action admittedly includes an additional element, namely that the taxes were paid under a mistake; but it is the presence of that additional element which enables the claimant to benefit from an extended limitation period which begins when the mistake is discovered or could with reasonable diligence have been discovered, rather than beginning when the payment was made. The mistake ground of action is therefore a valuable remedy for the recovery of taxes levied contrary to EU law. If it were not available for that purpose, then the person who had paid taxes levied contrary to EU law would be in a less favourable position than the person who had a similar claim under domestic law.
    226. The principle of equivalence does not of course oblige a member state to extend its most favourable rules governing recovery under national law to all actions for repayment of charges or dues levied contrary to EU law (Edilizia Industriale Siderurgica Srl v Ministero delle Finanze (Case C-231/96) [1998] ECR

    I-4951, para 36 (“Edis”)). It was therefore open to the United Kingdom to curtail the limitation period applicable to the ground of action based on mistake without offending against the principle of equivalence, so long as it did so not only for claims based on a breach of EU law but also for similar claims based on a breach of domestic law. That had not however been done by the time the present actions were commenced. Whether the retroactive manner in which the limitation period was subsequently curtailed was compatible with EU law raises issues not in relation to the principle of equivalence but in relation to the principle of effectiveness.

    227. If, then, the principle of equivalence required that the mistake ground of action should be available to the claimants at the time when they made their claims, then it follows under EU law that the principle of effectiveness also applied to that ground of action, and continues to apply until the claims are determined. The question which arises, and to which I turn next, is whether the application of section 320 of the 2004 Act to the Aegis claims, and of section 107 of the 2007 Act to the BAT claims, would be compatible with that principle.

    The application of the principle of effectiveness

    228. The principle of effectiveness requires that the national procedural rules required by the principle of equivalence must provide effective judicial protection in conformity with EU law. Taken in conjunction with the principle of equivalence, it is a principle which has far-reaching implications for domestic law.
    229. The principle of effectiveness may in particular impinge upon domestic laws relating to limitation periods. There is of course no objection in principle to limitation periods under EU law: on the contrary, it is recognised that reasonable periods of limitation are necessary in the interests of legal certainty (Rewe- Zentralfinanz eG v Landwirtschaftskammer für das Saarland (Case 33/76) [1976] ECR 1989, para 5 and Comet BV v Produktschap voor Siergewassen (Case 45/76) [1976] ECR 2043, paras 17-18). Equally, there is no requirement that rights derived from EU law should be subject to the most favourable limitation period available under domestic law, provided the principle of equivalence is respected (Edis).
    230. National legislation curtailing the period within which recovery may be sought of sums which have been levied in breach of EU law is not in principle incompatible with EU law. The Court of Justice has however laid down certain requirements with which such legislation must comply. It must for example not be intended specifically to limit the consequences of a judgment of the Court of Justice (see eg Deville v Administration des Impôts (Case 240/87) [1988] ECR

    3513). In that regard, I note that the Government announced its intention to introduce the provision which became section 107 of the Finance Act 2007 on 6 December 2006, which was the day on which the Court of Justice had rejected the Government’s application to re-open the hearing in the first reference in this case so as to allow it to seek a temporal restriction to the effect of the judgment. The effect of section 107 is not however confined to the taxes with which the court’s judgment was concerned, and it is not contended that the provision offended against the Deville requirement. In the circumstances, I proceed on that basis.

    231. A further requirement of legislation curtailing a limitation period is that the arrangements for its entry into force must be consistent with effective judicial protection of the rights derived from EU law. In particular, such legislation must ensure that it remains possible in practice to enforce the right to repayment derived from EU law. In order to understand how that principle applies in the present case, it is helpful to consider some of the judgments of the Court of Justice.
    232. First, Aprile v Amministrazione delle Finanze dello Stato (No 2) (Case C- 228/96) [2000] 1 WLR 126 concerned a claim for repayment of charges wrongfully levied in 1990, when such claims were subject to the general limitation period of ten years. On 27 January 1991 legislation was enacted which brought such claims within the scope of a shorter limitation period prescribed by customs legislation, which was then a period of five years, and in addition reduced that limitation period to three years as from 27 April 1991. The action was begun on 30 March 1994. It was accepted by the national authorities that the legislation could not be applied to claims which had been lodged prior to 27 April 1991. In that regard, the Advocate General observed at para 41 of his opinion that the legislation would be clearly incompatible with Community law if it applied to claims which had been lodged before that date: the Community principle of legal certainty did not allow such claims to be affected by a later provision not existing at the time of lodgement which detracted from the legal situation of the claimants. The issue concerned claims lodged after 27 April 1991 in respect of payments which had been made at a time when the longer limitation period applied. As the Court of Justice noted, the national courts interpreted the legislation as not having any retroactive effect: it was construed as meaning that persons whose claims had arisen before the date when the legislation came into force had three years from that date within which to commence proceedings: a period which was sufficient to guarantee the effectiveness of the right to reimbursement (para 28). On that basis, the legislation was compatible with Community law. The same conclusion was also reached, on similar facts, in Dilexport Srl v Amministrazione delle Finanze dello Stato (Case C-343/96) [2000] All ER (EC) 600.
    233. Secondly, Marks & Spencer plc v Customs and Excise Comrs (Case C- 62/00) [2003] QB 866 concerned a claim for repayment of VAT unduly paid between May 1991 and August 1996, when the relevant limitation period was six

    years. On 19 March 1997 legislation was enacted which reduced the limitation period to three years. The legislation was deemed to have come into force on 18 July 1996. The action was begun on 15 April 1997. The Court of Justice considered the legislation both in relation to the principle of effectiveness and in relation to the principle of the protection of legitimate expectations. I shall consider the second of those aspects below. In relation to the principle of effectiveness, the court derived from its judgments in Aprile and Dilexport the proposition that, in order for national legislation curtailing the period within which recovery may be sought of sums charged in breach of Community law to be compatible with Community law, “the time set for its application must be sufficient to ensure that the right to repayment is effective” (para 36). The Court continued:

    “37. It is plain, however, that that condition is not satisfied by national legislation such as that at issue in the main proceedings which reduces from six to three years the period within which repayment may be sought of VAT wrongly paid, by providing that the new time limit is to apply immediately to all claims made after the date of enactment of that legislation and to claims made between that date and an earlier date, being that of the entry into force of the legislation, as well as to claims for repayment made before the date of entry into force which are still pending on that date.

    38. Whilst national legislation reducing the period within which repayment of sums collected in breach of Community law may be sought is not incompatible with the principle of effectiveness, it is subject to the condition not only that the new limitation period is reasonable but also that the new legislation includes transitional arrangements allowing an adequate period after the enactment of the legislation for lodging the claims for repayment which persons were entitled to submit under the original legislation. Such transitional arrangements are necessary where the immediate application to those claims of a limitation period shorter than that which was previously in force would have the effect of retroactively depriving some individuals of their right to repayment, or of allowing them too short a period for asserting that right.
    39. In that connection it should be noted that member states are required as a matter of principle to repay taxes collected in breach of Community law (Société Comateb v Directeur Général des Douanes et Droits Indirects (Joined Cases C-192 to 218/95) [1997] ECR I- 165, para 20, and Dilexport [1999] ECR I-579, 610-611, para 23), and whilst the court has acknowledged that, by way of exception to that principle, fixing a reasonable period for claiming repayment is

    compatible with Community law, that is in the interests of legal certainty, as was noted in paragraph 35 hereof. However, in order to serve their purpose of ensuring legal certainty limitation periods must be fixed in advance (ACF Chemiefarma v Commission of the European Communities (Case 41/69) [1970] ECR 661, para 19).”

    234. As the court made clear at para 38, the legislation in issue in Marks & Spencer was objectionable not only because it applied retroactively to persons who had already made claims for repayment which were within the limitation period then in force, but also because it precluded claims by persons who could otherwise have made claims within that period, without any transitional provisions to protect the rights of such persons. A similar conclusion was also reached in Grundig Italiana SpA v Ministero delle Finanze (Case C-255/00) [2003] All ER (EC) 176, where a limitation period of five years was replaced by one of three years, and a transitional period of 90 days was held to be insufficient to ensure that the right of recovery was not rendered excessively difficult.
    235. It follows from cases such as Aprile, Dilexport, Marks & Spencer and Grundig that a taxpayer who has paid taxes levied contrary to EU law is not vested with a right to repayment in accordance with the domestic provisions which were in force at the time when the payment was made. It is permissible to alter the applicable rules of domestic law, including rules as to limitation, provided the legislation effecting the alteration does not in practice deprive the persons affected of their right to seek reimbursement. In order for that proviso to be met, however, the legislation must not apply the new limitation period retroactively so as to bar claims which were made timeously according to the law then in force, and the arrangements for its entry into force must also allow persons who have not yet made claims an adequate period of time to ensure that their right to repayment remains effective.
    236. In the present case, the claims are for the repayment of taxes unduly paid between 1973 and 1999, when the relevant limitation period was six years. That period generally ran from the date of the payment, but in an action for relief from the consequences of a mistake the period was extended: it did not begin to run until the claimant discovered the mistake or could with reasonable diligence have discovered it (section 32(1)(c) of the Limitation Act 1980, re-enacting a provision previously contained in section 12 of the Limitation Act 1939). As Lord Walker has explained at paras 103-104, it has been established in this case that the payments were made under a mistake about the lawfulness of the tax regimes under which they were paid; and it was only after the Court of Justice issued its judgment in Metallgesellschaft Ltd v Inland Revenue Comrs (Joined Cases C- 397/98 and C-410/98) [2001] Ch 620 that it was generally appreciated that the UK corporation tax regime was open to challenge as infringing Community law. A well-advised company in the position of the claimants would then have had

    grounds for considering that it was entitled to the repayment of tax which had been levied contrary to Community law, and that there was at least a reasonable prospect that it could rely upon the extended limitation period provided by section 32(1)(c) of the 1980 Act in order to recover any taxes paid more than six years before the proceedings were begun. In order to do so, it would of course have to base its claim upon the mistake ground of action. The BAT action was begun in June 2003, and the Aegis action on 8 September 2003. In each action, the claim was based upon the mistake ground of action (as well as the unlawful demand ground of action), and reliance was placed on section 32(1)(c). Section 320 of the 2004 Act, enacted in July 2004, excluded the application of section 32(1)(c) of the 1980 Act in relation to taxation matters where the action was brought on or after 8 September 2003. Section 107 of the 2007 Act, enacted in July 2007, excluded the application of section 32(1)(c) where the action was brought prior to 8 September 2003.

    237. It is apparent from that summary that the claims, so far as they relate to payments made more than six years before the proceedings were commenced, have always been dependent on the application of section 32(1)(c) of the 1980 Act. The effect of the legislation of 2004 and 2007 is thus to deprive the claimants, retrospectively, of the ability to pursue their claims so far as they relate to those payments. Since the legislation was retroactive in its effect, there was nothing the claimants could do to avoid its operation: that, of course, was the point of making the legislation retroactive.
    238. Since the legislation retroactively restricts the possibility of repayment to claimants who brought an action within six years of the date of the payment, rather than six years of the date when their mistake was discovered or could with reasonable diligence have been discovered, it deprives persons who do not satisfy that condition of any possibility of exercising the right to repayment derived from EU law, which they previously enjoyed. In the circumstances of this case, it retroactively renders the taxes unduly paid by the BAT group prior to June 1997, and by the Aegis group prior to September 1997, irrecoverable: taxes whose reimbursement had been timeously sought under the law then in force. It therefore renders impossible in practice the exercise of rights derived from the EU treaties which national courts are bound to protect. That is the first reason why I have reached the provisional conclusion that it is contrary to EU law and cannot be relied on in these proceedings.
    239. That conclusion does not appear to me to be affected by the argument that the legislation serves the legitimate purpose of avoiding the disruption of public finances which the present claims, and other similar claims, would otherwise cause. As the Court of Justice observed in its Marks & Spencer judgment at para 39, member states are required as a matter of principle to repay taxes collected in breach of EU law. Legal certainty, which protects both taxpayers and the

    administration, can justify fixing reasonable limitation periods for bringing claims for repayment, but it cannot in my view justify applying them in such a way that the rights conferred by EU law are no longer safeguarded.

    240. Nor in my view can the present case be distinguished from such cases as Marks & Spencer on the ground that those cases concerned situations where there was only one basis on which repayment could be sought, whereas the present case concerns a situation where two grounds of action exist, with differently calculated limitation periods, and the effect of the legislation in issue is merely to apply the same method of calculating the limitation period to both grounds of action. I accept that the present case differs in that respect from the cases which have come before the Court of Justice, but the difference is in my view of no consequence. Since both grounds of action are available as means of enforcing EU rights in accordance with the principle of equivalence, it follows that the principle of effectiveness must also be respected in relation to both. The vice of the legislation in issue is not that it seeks to apply a common limitation period to the two grounds of action, but that it does so retroactively and without transitional provisions, and so fails to conform to the principle of effective judicial protection.

    The principle of the protection of legitimate expectations

    241. A further reason for my provisional conclusion that the legislation is incompatible with EU law is that it is in my view incompatible with the principle of the protection of legitimate expectations. As a general principle of EU law, this principle binds member states when implementing EU law at national level. In particular, it applies to national rules governing the protection of EU rights in national courts. The point is illustrated by Marks & Spencer (Case C-62/00) [2003] QB 866, where the Court of Justice rejected the Government’s contention that the procedural rules governing the recovery of overpayments of VAT were entirely a matter of domestic law, subject only to the Community principles of equivalence and effectiveness. As the Court held (para 44), the principle of the protection of legitimate expectations forms part of the Community legal order; and, on the facts of that case, legislation retroactively curtailing the period within which repayment might be sought of taxes collected in breach of Community law was incompatible with that principle.
    242. It is in my opinion an even clearer breach of that principle for legislation which has the effect of reducing the limitation period applicable to actions for the enforcement of rights derived from EU law to be applied to actions which were already pending before the courts when the legislation was enacted. Although persons cannot legitimately expect that the legal rules applicable to them will not be altered, they may legitimately expect that rights which they possess will not be retroactively abridged. They are therefore entitled to expect that a claim which was

    not time-barred when it was made will not subsequently become time-barred as a result of retroactive legislation.

    243. My conclusion on this point does not depend on an assumption that the claimants knew, at the time when they commenced proceedings, that their claims could validly be based upon the mistake ground of action, and could therefore benefit from the extended limitation period provided by section 32(1)(c) of the 1980 Act. Although the validity of claims to the repayment of unlawfully levied tax on the basis of mistake was strongly arguable at that time, and was of course ultimately established, I accept that it was only some years later that the point was definitively resolved by the decision of the House of Lords in DMG [2007] 1 AC

    558. Although there was therefore an arguable question in 2003 as to whether the claims which they had submitted to the court were time-barred, the claimants could legitimately expect that that question would be decided by the court in accordance with a proper understanding of the law in force at the time when the claims were made. They could legitimately expect that the court’s decision of that question would not be pre-empted by retroactive legislation subsequently enacted by Parliament.

    244. Nor does it appear to me to be material that the legislation in issue left untouched the limitation period which applied to the ground of action based on an unlawful demand. The claimants had based their claims upon both grounds of action, as they were entitled to do. The fact that their claims in respect of payments made during the six years prior to the commencement of the proceedings, so far as based on the unlawful demand ground of action, were not affected by the legislation in issue does not diminish the significance of the fact that their right to pursue claims in respect of earlier periods, on the basis of mistake, was taken away from them after proceedings relying upon that right had been commenced.
    245. The protection of legitimate expectations is not of course an absolute principle, and even retroactive measures interfering with the administration of justice may sometimes be justified by compelling considerations relating to the public interest; and, in any assessment of whether such a justification existed, a lack of certainty as to the law at the material time might be a relevant consideration. In the present case, however, for the reasons explained in para 239, there appear to me to be no other considerations capable of outweighing the breach of legitimate expectations which resulted from the legislation in issue.

    Conclusion

    246. In view of the division of opinion on the court in relation to the compatibility of section 320 of the 2004 Act with EU law, I agree that that issue

    will require to be the subject of a reference to the Court of Justice in accordance with the directions proposed by Lord Hope. The other issues should in my view be dealt with as proposed by Lord Walker.

    1. Case C- 446/04) [2007] STC 326), all the issues as to infringement have been considered by Henderson J [2008] EWHC 2893 (Ch), [2009] STC 254 and by the Court of Appeal [2010] EWCA Civ 103, [2010] STC 1251. Some have been decided and are no longer in dispute. In particular, it is now common ground that corporation tax measures relating to advance corporation tax (“ACT”) and foreign income dividends (“FIDs”) infringed former article 43 (freedom of establishment) and former article 56 (free movement of capital) of the EC Treaty, now articles 49 and 63 of the Treaty on the Functioning of the European Union. Other points have been made the subject of a second reference to the Court of Justice. On yet further points this court has extended time for an application for permission to appeal. One of these is the concurrent finding of the courts below that the infringements which have been established did not amount to grave and manifest breaches of EU law so as to give rise to a claim for damages on the principles in Brasserie du Pecheur SA v Federal Republic of Germany (Joined Cases C-46/93 and C-48/93) [1996] QB 404.
    26. It is now clear that, apart from any possible claim for damages, the claims to be met by HM Revenue and Customs (“HMRC”, so as to include its predecessors) are restitutionary in nature. Some are straightforward claims for recovery of tax which the claimants paid when it was not due. Other claims are for less direct losses which the claimants say they sustained in consequence of the non-compliance of the corporation tax system with EU law. In relation to restitutionary relief for both the direct and the indirect losses there are important differences between the parties as to the characterisation of the remedies available to the claimants as a matter of English law. There are also important differences as to how far EU law requires the full range of domestic remedies to be made available for the recovery of unduly paid tax, despite parliamentary intervention (in the form of section 320 of the Finance Act 2004 and section 107 of the Finance Act 2007) to curtail those remedies drastically and with retroactive effect. Those two provisions (“the statutory cut-off provisions”) are challenged as infringing EU law.
    27. That is a brief sketchy overview of the significance of this appeal in the context of the larger campaign of the FII group litigation. Except in relation to the statutory cut-off provisions the Supreme Court does not on this appeal have to revisit any issue as to infringement of EU law. But it is appropriate to give a brief explanation of the ACT system, now abolished, that gave rise to the substantive infringements. A much fuller explanation can be found in the first instance

    judgment of Henderson J [2009] STC 254, paras 12 to 28. This draws on the first order for reference to the Court of Justice made by Park J on 13 October 1994. Since his retirement Sir Andrew Park has himself given an objective account of the progress of several of the associated sets of group litigation in “A Judge’s Tale: Corporation Tax and Community Law” [2006] BTR 322.

    The ACT system

    28. Corporation tax was introduced in the UK in 1965. At first the system was a “classical” system, with full double taxation of company profits and non-corporate shareholders’ dividends. In 1973 the system changed to one of “partial imputation”. When a UK-resident company paid a dividend it was required (by way of self-assessment) to pay an amount of ACT equal to the mainstream corporation tax (“MCT”) payable on the part of its profits distributed as dividend. A non-corporate shareholder became entitled to a tax credit equal to the ACT paid in respect of his dividend. A UK-resident corporate shareholder receiving a dividend from another UK-resident company received it as “franked investment income” (“FII”), and if it both received and paid dividends, ACT was payable only on the excess of its outgoing “franked payments” over its FII.
    29. The position was different if a UK-resident company received a dividend from a non-resident company in which it was a shareholder. That was so whether or not the two companies were part of a group, but this group litigation, and the parallel ACT group litigation, have both been concerned with groups of companies. Most of the test claimants in this litigation are members of the British American Tobacco (“BAT”) group. In para 2 of his judgment Henderson J gave a concise explanation of this group litigation as compared with the ACT group litigation:

    “Whereas the focus of the ACT Group Litigation was on the UK domestic legislation which prevented UK-resident subsidiaries of foreign parents from making group income elections, thereby obliging them to pay ACT when paying dividends to their foreign parents, the focus of the FII Group Litigation has been on UK- parented groups with foreign subsidiaries, and on the tax treatment of dividends coming into the UK from abroad. At the simplest level, therefore, the present litigation is concerned with factual situations which are the opposite of those which gave rise to the questions considered in Hoechst [Metallgesellschaft Ltd v Inland Revenue Comrs, (Joined Cases C-397/98 and C-410/98) [2001] Ch 620] and the ACT Group Litigation.”

    30. Since 1973 the BAT group has gone through various structural changes (summarised in paras 1.8 to 1.21 of an agreed statement of facts set out in para 29 of the judge’s judgment) but it has always had as its ultimate holding company a UK-resident company whose shares are listed and whose thousands of shareholders expect to receive regular dividends. After 1973 the BAT group (in common with many large multinational groups) faced a difficulty in that when it received dividends from overseas subsidiaries it did not receive a tax credit that could be used to eliminate or reduce ACT payable in respect of its dividends to its shareholders. The overseas dividends were not FII. Although the UK-resident company was entitled to double taxation relief against MCT (in the form of a credit against foreign taxes paid by the subsidiary), it still had to pay ACT. If relatively little MCT was payable (because of double taxation relief) the ACT became surplus and of little or no utility to the holding company. A UK-resident company with overseas subsidiaries (whether resident within or outside the EU) was therefore at a disadvantage, and articles 43 and 56 of the Treaty were infringed.
    31. The other test claimants are members of the Aegis group, another multinational group whose holding company is based in the UK. These claimants have been included because they are (and claimants in the BAT group are not) affected by section 320 of the Finance Act 2004.
    32. The ACT regime was in force from 1973 to 1999. Its disadvantages for multinational groups were to some extent mitigated by provisions as to foreign income dividends (FIDs) which were in force from 1994 to 1999. A UK-resident company receiving dividends from non-resident companies could elect that dividends paid to its shareholders should be treated as FIDs. The effect was that ACT was still payable, but would in some circumstances be repaid after an interval, normally of a duration of between eight and a half months and seventeen and a half months. There is a fuller explanation of the law in paras 23 to 25 of the judge’s judgment, and of the facts as to FID “enhancements” in paras 277 to 302.
    33. The principal statutory provision giving a tax credit on qualifying distributions between UK-resident companies was section 231 of the Income and Corporation Taxes Act 1988 (“TA 1988”). Issue 6 before the Court of Appeal was whether section 231 could be interpreted, under the Marleasing principle (Marleasing SA v La Comercial Internacional de Alimentación SA (Case C- 106/89) [1990] ECR I-4135) so as to be compatible with EU law. The Court of Appeal held that it could be interpreted in that way. That is however an issue on which this court has deferred a decision on permitting a further appeal. The uncertainty as to section 231 is a further complication in clarifying the issues that are before the court on this appeal.

    The issues

    34. The Supreme Court gave permission to appeal on four of the 23 issues identified by the Court of Appeal (and set out in the index to its judgment, [2010] STC 1251). This permission was later extended to cover a fifth issue, numbered 22 in the Court of Appeal’s judgment, that is the correct construction and scope of section 32(1)(c) of the Limitation Act 1980. The other four issues covered by the formal order granting permission to appeal are wholly or largely questions of EU law, and the impact of EU law on domestic rights and remedies: that is (issue 12) remedies in English law; (issues 20 and 21) the compatibility with EU law of the statutory cut-off provisions; and (issue 23) whether section 33 of the Taxes Management Act 1970 (as amended) provides an exclusive code for recovery of tax mistakenly paid under an assessment, and the impact on that section of EU law.
    35. However, as the parties’ written and oral submissions have developed it has become apparent that there is another wholly domestic issue of central importance to the appeal. The Court of Appeal differed from Henderson J as to whether the principle in Woolwich Equitable Building Society v Inland Revenue Comrs [1993] AC 70 (“Woolwich”) applies only to tax that is demanded by revenue authorities (and if so, what amounts to a demand).
    36. For the appellants Mr Aaronson QC took the lead in making submissions on issues of EU law, followed by Mr Rabinowitz QC on issues of English law. This sequence of argument may have been unavoidable, but it produced the result that the court heard submissions about the attitude of EU law towards national procedures and remedies─which is an important part of this appeal─before hearing submissions about the English remedies themselves. It is more helpful to start with the issues of English law, and then assess the impact that EU law has on them. So this judgment proceeds to consider (i) the scope of section 32(1)(c) of the Limitation Act 1980 and (ii) the scope of the decision in Woolwich, before addressing the effect of EU law.
    37. It may not be immediately apparent why these two domestic issues have assumed such significance, so a brief explanation is called for. The reason is certainly not the disinterested and scholarly interest of the parties, or either of them, in the development and clarification of English private law. That is apparent from another of the group litigation proceedings, NEC Semi-Conductors Ltd and Other Test Claimants v Inland Revenue Comrs [2006] EWCA Civ 25, [2006] STC 606 (“NEC”), in which (at paras 140 to 147) the claimant companies and HMRC (through counsel, most of whom have appeared on this appeal) made submissions on the Woolwich issue to the contrary effect, in each case, to those they have made on this appeal. These tactical shifts have occurred because, naturally enough, each

    side wants to win, by any proper line of argument, because of the very large sums of money at stake.

    38. The main issue of EU law to be decided can be put, in a very simplified (but not, it is to be hoped, tendentious) form, as follows. When in any member state tax has been paid which was not due because the national taxing measure infringed the Treaty, must the member state make available to its aggrieved taxpayer (i) an adequate remedy which meets the principles of effectiveness and equivalence; or

    (ii) every available national remedy, including any that offers the taxpayer special advantages as regards limitation of actions?

    39. At first glance the Woolwich principle provides an adequate remedy, subject to a six-year limitation period unaffected by the statutory cut-off provisions. Similarly at first glance mistake of law, following the decision of the House of Lords in Deutsche Morgan Grenfell Group Plc v Inland Revenue Comrs [2007] 1 AC 558 (“DMG”), provides a specially advantageous basis of claim because of the possibility of an extended limitation period under section 32(1)(c) of the Limitation Act 1980, but subject to the statutory cut-off provisions (if and so far as valid under EU law). But if the test claimants have no Woolwich claim, because as a matter of law such a claim requires an unlawful demand, and there was no such demand, mistake of law would be promoted, as it were, to being the only remedy available under national law, and so to being more surely entitled to protection under EU law. So it is expedient for the test claimants in this appeal to reverse the stance taken by the test claimants in NEC and argue that the Woolwich principle does not extend to self-assessed taxes, for which there is no official demand.
    40. The issue on section 32(1)(c) of the Limitation Act 1980 is part of an alternative line of argument by which the test claimants seek to promote the mistake of law claim and so ensure its protection under EU law. They submit that section 32(1)(c) should be widely construed, contrary to the authority of Phillips- Higgins v Harper [1954] 1 QB 411, a first-instance decision which has however stood and been followed for over half a century. They submit that section 32(1)(c) is applicable, regardless of the cause of action, wherever there is a causally relevant mistake. In the words of Mr Rabinowitz (day 2, page 80), “The mistake element does not have to be a necessary part of the cause of action, so long as the mistake is materially causal or causally material in producing the circumstances from which relief is sought.” So this is an alternative method by which the test claimants seek to saw off the apparent support of the Woolwich branch in order to rely on mistake of law alone.
    41. It seems very doubtful, even if their argument on section 32(1)(c) is sound, whether the claimants’ aim would be achieved. In other, more mainstream parts of their argument they rely heavily on the principle (reasserted in this context by the

    House of Lords in DMG [2007] 1 AC 558) that English law permits litigants to choose, as between concurrent causes of action, the cause or causes of action most advantageous to their interests. The test claimants have done so. In the amended particulars of claim of the BAT group, paras 15 and 15A, they have clearly and distinctly relied on two separate causes of action in unjust enrichment, that is (para 15) payment of tax unduly levied and (para 15A) payment under a mistake. Section 32(1)(c) is relied on in relation to “mistake claims” only (paras 18, 18A and 18B). The position is the same on the Aegis group’s pleadings. The statutory cut-off provisions (the essential text of which is set out at paras 107 and 109 below) do contain (in section 320(6) and section 107(2 [↩]

26. Case C-106/89) [1990] ECR I-4135). In terms of the amount of tax at stake, this issue is relatively minor in the context of the litigation as a whole, as it extends only to tax charged under Schedule D, Case V, pursuant to section 18 of TA 1988. But it is still a point of some general importance. Before Henderson J HMRC argued, but only it seems quite briefly, that the decision of the Court of Appeal in Monro v Revenue and Customs Comrs [2009] Ch 69 established that section 33 was an exclusive remedy which left no room for any common law claim in unjust enrichment. The judge [2009] STC 254, paras 438-439 rejected that on two grounds: first that section 33 did not extend to tax levied otherwise than by an assessment; secondly that in any event the national legislation must, in a San Giorgio claim, yield to the principle of effectiveness. It now seems to be common ground that the first of these reasons does not hold good for tax under Schedule D Case V.

118. The Court of Appeal took a different approach. It concluded ([2010] STC 1251, paras 261 and 264) that a conforming interpretation was possible, and did sufficiently “go with the grain of the legislation” (the expression used in relation to section 3(1) of the Human Rights Act 1998 by Lord Rodger of Earlsferry in Ghaidan v Godin-Mendoza [2004] 2 AC 557, para 121, also adopted by Lord Nicholls of Birkenhead at para 33). The conforming interpretation adopted was (para 261) that the restrictive condition about prevailing practice in section 33(2A) “is to be read as subject to the limitation that it applies only if and to the extent that the United Kingdom can consistently with its Treaty obligations impose such a restriction”.

119. I have grave doubts as to whether that interpretation does not go against the grain of the legislation, since the “practice generally prevailing” condition is of long standing and has always been regarded as an important safeguard for the public revenue. I am inclined to think that Mr Aaronson was right (Day 2, pp 25- 26) to call it a “cardinal feature” of the legislation. In my view the Marleasing principle can be applied in a simpler and more natural way by not construing section 33 as impliedly setting itself up as an exclusive provision (which it did not do expressly, unlike section 80 of the Value Added Tax Act 1994). The test claimants submit that the application of Marleasing cannot rework section 33 in a way that serves any relevant purpose. But to read it as non-exclusive does not go against its grain. It would merely exclude an implication which is itself no more than a process of statutory construction.    In practical terms the effect is the same as that which Henderson J reached by the second limb of his reasoning. I would therefore allow the appeal on this point (although it may not, in the end, make much practical difference).

120. In summary, therefore, my provisional view is that we should –
121. uphold the Court of Appeal as to (i) the scope of section 32(1)(c) of the Limitation Act 1980 and (ii) the scope of the Woolwich principle;

2. allow the appeal on section 320 and section 107; and

3. allow the appeal on section 33 of the Taxes Management Act 1970.

But in view of the difference of opinion in the court I consider (in common with Lord Hope, Lord Dyson and Lord Reed) that it is necessary for the court to make a further reference to the Court of Justice of the European Union in accordance with directions in para 23 of Lord Hope’s judgment.

LORD BROWN

121. I have had the great advantage of reading in draft the judgments of Lord Walker and Lord Sumption and am in full agreement with them both on the several issues upon which they each agree. What, then, of the single issue upon which they disagree: was section 320 of the Finance Act 2004 contrary to EU law as infringing the principle of effectiveness as explained by the Court of Justice in Marks & Spencer Plc v Customs & Excise Comrs (Case C-62/00) [2003] QB 866?

122. During the hearing I confess to having found difficulty in recognising any principled basis for distinguishing between on the one hand section 47(1) of the Finance Act 1997 which (with effect from when government had earlier announced its intention so to legislate: section 47(2 [↩]