AXA Insurance UK PLC and another v Commissioners of Inland Revenue and another [2026] UKSC 24

[2026] UKSC 24JUDGMENT
Between
AXA Insurance UK PLC and anotherAppellantsbefore Lord Hodge, Deputy President Lord Hamblen Lord Leggatt Lady Rose Lord RichardsHearing JUDGMENT GIVEN ON 27 July 2026 Heard on 18 November 2025Appellants Jonathan Bremner KC (Instructed by Joseph Hage Aaronson LLP)Respondent David Ewart KC Barbara Belgrano Laura Ruxandu Frederick Wilmot-Smith (Instructed by Solicitor’s Office and Legal Services HMRC)Date 2026-07-27
[1]Introduction This appeal raises important issues about the operation of the innovative litigation procedure that was introduced into the Civil Procedure Rules of England and Wales in 2000; the group litigation order or GLO. The rules governing this procedure are primarily found in the Civil Procedure Rules Pt 19, rules 19.21–19.26. The typical situation in which a GLO is made by the court is where there are a large number of different sets of proceedings issued by claimants in the High Court or County Court usually against the same defendants, raising the same or similar issues. If all the sets of proceedings go forward in parallel, there is a risk of different judges arriving at different answers to what are in effect the same legal or factual questions. If a GLO is made grouping the claims together, the common issues of fact or law can be identified and one or more of the claims can be chosen as a test case in which those issues are determined. The other proceedings in the group are then stayed to await the result of the test case.[2]The procedure is intended to ensure consistency of result and to save time and money for all the parties as well as for the court system. If other proceedings are issued by new claimants raising the same issues, they can join the rest of the group by being placed on the group register which is maintained by the court registry. The CPR governing the GLO procedure provide that where a judgment or order is given or made in relation to one or more group issues, that judgment or order is binding on all other claims on the register at the time of the judgment or order “unless the court orders otherwise”: see CPR r 19.23(1)(a). We refer to such other claims as “the follower claims”.[3]Sometimes each of the individual claims grouped together by a GLO is for a relatively small amount of money. In those circumstances the cost of bringing each individual claim would far outweigh the likely compensation to be won if the claim is successful. The GLO procedure can be an important tool for ensuring access to justice. That, however, is not true of the GLO with which this appeal is concerned. The Controlled Foreign Companies and Dividend Group was formed by the making of a GLO in July 2003 (“the CFC and Dividend GLO” or “the GLO”). Most of the proceedings on the register set up for this GLO are claims for many millions of pounds and many of the claimants are sophisticated international businesses. But it is important when addressing the issues raised by this appeal to bear in mind the different kinds of GLOs that have been made in the past and may be made in the future.[4]The claims brought in the proceedings covered by the CFC and Dividend GLO allege that the claimants have paid tax under statutory taxing provisions which, it is now accepted, were incompatible with the law of the European Union (“the EU”). The defendants are the current UK taxing authority, the Commissioners for His Majesty’s Revenue and Customs and their predecessors, the Commissioners of Inland Revenue. The inclusion of the predecessor taxing authority reflects the fact that the claims cover tax paid over many years prior to 18 April 2005 when HMRC replaced the IRC. We shall refer to them both as “the Revenue”.[5]Three test cases from the CFC and Dividend GLO were chosen to go forward dealing with different common issues about the lawfulness of the taxing provisions and the consequences of any incompatibility. The test case most relevant to this appeal is the claim brought by companies in the Prudential group headed by Prudential Assurance Co Ltd plc (“the Prudential test case”). During the course of the Prudential test case a reference was made to the Court of Justice of the European Union (“the CJEU”) We use the term “CJEU” to refer to that Court during the time it was known as the European Court of Justice. and legal issues common to the group were subsequently determined at the level of the High Court, the Court of Appeal and this court. The Prudential claim was finally disposed of by order of the Supreme Court dated 25 July 2018.[6]The present proceedings (“the AXA proceedings”) were commenced in April 2003 by several companies in what was then the AXA Sun Life group. The proceedings were included on the register for the CFC and Dividend GLO. It was not one of the test cases and so was stayed for the long time that it has taken for the test cases finally to be decided. The present appeal is brought by two claimants in the AXA proceedings.[7]The appellants argue that there were two important issues common to the claims included in the GLO which were decided in favour of the taxpayers by the Prudential test case. Both relate to the treatment under the UK tax regime of dividends received from non-UK companies which was less favourable than the treatment of dividends received from UK-resident companies. This discriminatory treatment was unlawful as a matter of EU law.[8]One of those issues relates to the payment of advance corporation tax (“ACT”) on dividends paid by a UK-resident company. ACT was not payable to the extent that the taxpaying company received dividends from UK-resident companies which generally carried a tax credit. No such tax credit was attached to dividends received from non-UK companies, so that the UK-resident company in receipt of such dividends was required to account for ACT without any set off against a tax credit. In due course, the company received a credit for the ACT against its liability to mainstream corporation tax (“MCT”). The issue was whether the Prudential test claimants, and therefore the follower claimants, had a claim at common law which entitled them to interest to compensate them for the loss they had suffered by in effect paying the ACT prematurely. It was a claim for compensation for the loss of the use of the money between payment of the ACT and the due date for the mainstream corporation tax (“the prematurity period”). It was not a claim for the recovery of any tax paid. This is known as the Set Off Issue.[9]The second issue concerns the limitation period for claims for both the recovery of tax paid and compensation by way of interest for the prematurity period. The appellants say this issue (“the Limitation Issue”) was determined in their favour by a decision in the Prudential test case that section 32(1)(c) of the Limitation Act 1980 operated in these circumstances to postpone the start of the applicable limitation period to the date of the judicial ruling that established the unlawfulness of the relevant UK tax provision. This would enable them to pursue claims in relation to periods many years before they issued their proceedings.[10]The Revenue’s response as to the Set Off Issue, is that although it was indeed decided in the Prudential test case, on the basis of the law as it stood at the date of that decision, that there existed a common law restitutionary claim for interest in respect of the prematurity period, this court subsequently held that no such claim existed at common law. They therefore argue that in the circumstances of this case, the court should exercise its discretion under CPR r 19.23(1)(a) and order that the test case decision is not binding in the follower claims.[11]The Revenue’s response as to the Limitation Issue is that a careful analysis of the High Court decision in the Prudential test case shows that in fact there was no determination of that issue. The Revenue say therefore that there is no judgment or order to which CPR r 19.23(1)(a) applies and so no need for the court to order that it does not bind the follower claims.[12]The Court of Appeal in the judgment under challenge before us decided both issues in favour of the Revenue: AXA Sun Life plc v Revenue and Customs Comrs [2024] EWCA Civ 1430; [2025] 1 WLR 2179. It decided to exercise its discretion to order that the decision in the Prudential test case on the existence of the cause of action did not apply to this case. It said that it would not be right for the court to decide the follower claims on a basis that is now known to be wrong in law. The court decided further that the Prudential test case had not in fact determined the start date of the limitation period for any claims other than the test case itself. It added that if it had concluded that the High Court had made such a determination, it would have done so applying an interpretation of the legislation which had also subsequently been determined by this court to be wrong. The Court of Appeal would also therefore have exercised its discretion to order otherwise on this issue as well, again because the cases should not be decided on an incorrect legal basis.[13]The appellants now appeal to this court. They argue that the Court of Appeal erred when considering how to exercise the discretion under CPR r 19.23(1)(a) to disapply the decision in the Prudential test case about the existence of the cause of action. Further, they say that the Court of Appeal was wrong in how it approached identifying whether there was a common issue about the start date for the limitation period decided by the High Court for the benefit of the parties in other proceedings on the GLO register. The appellants contend that the Court of Appeal’s approach on both points risks seriously undermining the value of the GLO process under CPR r 19.21ff.[2]The GLO regime[14]The system for group litigation orders contained in CPR r 19.23 resulted from recommendations made by Lord Woolf in Chapter 17 (headed Multi-Party Actions) of his report Access to Justice: Final Report to the Lord Chancellor on the Civil Justice System in England and Wales (July 1996). He recommended the establishment of a special procedure for multiple claims raising similar issues of fact or law. Its objectives would be:(a) to provide access to justice where large numbers of persons were affected by another’s conduct but individual losses were small;(b) to provide expeditious, effective and proportionate methods of resolving cases where individual damages are large enough to justify individual action but where the number of claimants and the nature of the issues involved make normal case management unsuitable; and(c) to achieve a balance between the normal rights of parties, to pursue and defend cases individually and the interests of a group of parties to litigate the action as a whole in an effective manner (para 2).[15]Lord Woolf identified as a basic matter that would need consideration at an early stage whether there were generic issues which could be effectively decided within the group litigation (para 32(a)). “On the whole, this is likely to be the most rational and economic way of working through the case” (para 33). At para 40, he said:
“There are, however, difficulties in relation to test cases. Firstly, both claimants and defendants need to accept that the case will be a test case in relation, say, to liability for all those claimants in the same position. It is therefore necessary to make express orders in advance of determination that parties will be bound by the results. Secondly, there are also problems if the test case turns out to be atypical—if it is disposed of on particular grounds or if the judgment is couched in such a way that it leaves undetermined the similar issues in other cases. It is therefore necessary for the difficulties of identifying cases as test or lead cases to be specifically addressed by the court at an early stage.”
[16]CPR rr 19.21–26 contains the rules for group litigation. CPR r 19.21 defines a GLO as an order made to provide for the case management of claims which give rise to common or related issues of fact or law (“GLO issues”). CPR r 19.22(2) provides that a GLO must, among other things, contain directions about the establishment of a group register on which claims managed under the GLO will be entered, specify the GLO issues which will identify the claims to be managed as a group under the GLO, and specify a management court for the claims on the group register. CPR r 19.22(3) provides that a GLO may, in relation to claims that raise one or more GLO issues, direct their transfer to the management court, order their stay until further order, direct their entry on the group register, and direct that from a specified date claims which raise one or more GLO issues should be started in the management court and entered on the group register. It may not therefore be a matter of choice for claims raising GLO issues whether they become subject to the GLO.[17]CPR r 19.23(3) sets out the effect of a GLO and contains the key provision for the purposes of this appeal in paragraph (1)(a). Paragraphs (1)–(3) are as follows:
“(1) Where a judgment or order is given or made in a claim on the group register in relation to one or more GLO issues— (a) that judgment or order is binding on the parties to all other claims that are on the group register at the time the judgment is given or the order is made unless the court orders otherwise; and (b) the court may give directions as to the extent to which that judgment or order is binding on the parties to any claim which is subsequently entered on the group register. (2) Unless paragraph (3) applies, any party who is adversely affected by a judgment or order which is binding on them may seek permission to appeal the order. (3) A party to a claim which was entered on the group register after a judgment or order which is binding on them was given or made may not— (a) apply for the judgment or order to be set aside, varied or stayed; or (b) appeal the judgment or order, but may apply to the court for an order that the judgment or order is not binding on them.”
[18]CPR r 19.24 contains case management provisions. The court may vary the GLO issues and it may provide for one or more claims on the group register to proceed as test claims, the outcomes of which will have binding effect on other group claims under CPR r 19.23, subject to the power of the court to order otherwise.[19]Lord Woolf made observations on the GLO regime in Boake Allen Ltd v Revenue and Customs Comrs [2007] UKHL 25; [2007] 1 WLR 1386, a group claim against the Revenue alleging that the tax treatment of UK subsidiaries of US and Japanese holding companies was contrary to the prohibition against discrimination contained in double taxation conventions which had effect in domestic law. He emphasised that the objective of the GLO regime was to protect all litigants from incurring unnecessary costs: para 31.[20]The importance of finality in litigation was stressed by Sir Geoffrey Vos C in Claimants in Class 8 of the CFC and Dividend Group Litigation v Revenue and Customs Comrs [2019] EWHC 338; [2019] 1 WLR 5097 a judgment concerning a particular subset of claimants in the same GLO as this appeal. The Revenue sought to raise some points which, the claimants argued, went to issues of law which had already been decided. Sir Geoffrey Vos C agreed and held that those issues could not be re-opened. They had been finally decided and there must be finality to litigation:
“The GLO process has been established in order to facilitate that outcome.” (para 133) The points could have been, but were not, raised by the Revenue when the issues were argued and decided. As Sir Geoffrey observed at para 133: “The fact that [the Revenue] might have done better if they had raised the point earlier does not assist them.”
At para 144, he said that the Revenue “cannot have two bites at the cherry …”.[21]It is clear thatthe overall purpose of the GLO regime is to facilitate the economic and efficient disposal of claims raising the same issues, for the benefit of the parties and, indeed also, for the benefit of the court, its funding and its other users, and thereby to facilitate access to justice.[3]The claims in the AXA proceedings[22]The claim giving rise to this appeal was first issued on 8 April 2003 and was brought initially by six members of the AXA Sun Life group of companies. The claim form sought declarations that certain provisions of the Income and Corporation Taxes Act 1988 (“ICTA”) were incompatible with EU law in so far as they concerned the taxation of dividends received from a company resident in another member state of the EU or the European Economic Area (“EEA”) by a company resident in the United Kingdom. The particulars of claim stated that the claimants were all companies resident in the United Kingdom and held shareholdings in numerous companies resident in members states of the EU or the EEA and received dividends from those companies. The shareholding of the claimants in any one company was always below 10%—referred to as a “portfolio shareholding”.[23]The incompatibility particularised in the claim was that a UK resident company which invests in a portfolio company resident in the EU or EEA was discriminated against as compared with a UK resident company which invested in another UK resident company in a number of ways. For example, in the latter case there was a specific exemption which meant that there was no charge to UK corporation tax on the amount of the dividend, provided that both the payer and the recipient were UK resident: see section 208 ICTA. But a dividend from an EU/EEA resident company never qualified for that exemption.[24]It was alleged that this, and other provisions in the tax regime for dividends received from portfolio companies, set up obstacles for UK resident companies which wished to invest in EU/EEA companies and thus hindered and deterred such investment. It is alleged therefore that the provisions impeded or discouraged the free movement of capital and the exercise by UK resident companies of their right of establishment in the EU. The claimants described their cause of action in various ways including as a claim for restitution of monies paid by the claimants pursuant to a mistake of law. The accounting periods covered by the claims made by most of the original claimants went back to 1990, that is much further back than six years before the claim was issued. The approximate liability to corporation tax for each claimant was set out in a schedule to the claim and the total amount of tax they had paid was £8,850,637.[25]The April 2003 claim form has been amended over the years. In July 2003 it was amended to extend the claim to cover dividends from portfolio companies resident not just in the EU or EEA but in third countries. The approximate value of tax paid rose to over £29 million. In July 2009, Guardian Royal Exchange Assurance plc (now called AXA Insurance plc) (“GREA”) was added to the claim. Its claim is different from the other claims because one of the companies from which it received dividends was a wholly owned subsidiary in the Netherlands not a portfolio company.[26]The claims by this point extended beyond the alleged discrimination in the taxation of dividends received by the claimants from their portfolio shareholdings. It now also alleged discrimination in the way that dividends paid by the claimants up the corporate chain to their own shareholders were taxed. Other provisions of ICTA provided for the payment by the claimants of ACT calculated according to the value of the dividends they paid to their shareholders. The value of those dividends reflected in some respects the dividends that they had received from their portfolio shareholdings. Again, it was alleged that the tax regime applied in a discriminatory way. If the portfolio companies had been UK residents, then there would have been a reduction in the amount of ACT the claimants were required to pay when they paid dividends to their shareholders because the dividend income they received would have attracted certain tax credits under ICTA.[27]There are now nine claimants party to the AXA proceedings. The total estimated value of the claim made by all nine claimants (including AXAIUK and GREA) in the current iteration of the amended particulars of claim dated 26 March 2019 is over £68 million. The periods in which dividends are alleged to have been received or paid go back to 1974. The current particulars of claim assert that in paying the taxes demanded of them, the claimants acted in the mistaken belief that they were legally obliged to pay the tax. The relief claimed includes restitution of payments made pursuant to a mistake of law. The relief also includes, and has included from the outset, interest pursuant to section 35A of the Senior Court Act 1981 “and/or compound (or other) interest pursuant to the rules of law and/or equity, on the sums claimed at such rate and for such period as the court shall deem just”. As we later explain, GREA is not now advancing a claim for compound interest.[4]The issues relevant to this appeal[28]This appeal concerns the claims brought by AXA Insurance UK plc (“AXAIUK”) and by GREA. The many complicated issues raised by the AXA proceedings about the extent of the incompatibility of the challenged taxing provisions with EU law have, for our purposes, been decided by judgments of the CJEU and domestic courts, including this court. There is no doubt that they have what is commonly called a San Giorgio claim, that is a right to the refund of charges levied in a member state in breach of the rules or EU law: see Amministrazione delle Finanze dello Stato v San Giorgio SpA (Case 199/82) [1983] ECR 3595.[29]AXAIUK’s claim relates to corporation tax it paid on dividends it received from its portfolio shareholdings from non-UK resident companies over a number of accounting periods. AXAIUK makes no ACT claims in respect of dividends which it paid to its own shareholders.[30]The limitation period that applies to claims in restitution is the six-year period set by section 5 of the Limitation Act 1980 for claims based on simple contract: see Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2020] UKSC 47; [2022] AC 1 (“FII SC 2”). The period now covered by AXAIUK’s claim extends back to the accounting period ending 31 December 1995. Because AXAIUK issued its claim on 8 April 2003, payments of tax in respect of that accounting period fall outside the six-year limitation period. In order to succeed in its claim for that period, AXAIUK must rely on section 32(1)(c) of the Limitation Act 1980 (“section 32(1)(c)”). That provides so far as relevant as follows:
“32. Postponement of limitation period in case of fraud, concealment or mistake. (1) … where in the case of an action for which a period of limitation is prescribed by this Act, either— … (c) 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 … mistake (as the case may be) or could with reasonable diligence have discovered it.”
[31]The question in AXAIUK’s appeal is therefore how section 32(1)(c) applies to its claim, in particular what is the date on which it could with reasonable diligence have discovered its mistake, that mistake being that it thought that the tax it was being asked to pay on dividends received for portfolios shareholdings in non-UK resident companies was lawfully due when it was not so due. This is the Limitation Issue we referred to earlier.[32]GREA’s claim relates to ACT it paid on the dividends it declared and paid to its shareholders. The periods covered by its claim are the accounting period ending 31 December 1993 to the accounting period ending 31 December 1997. Some of the ACT it paid to the Revenue, for example in the accounting period ending 31 December 1996, was then used to offset a liability for MCT which arose in a later accounting period and which it is common ground GREA lawfully had to pay. The unlawfully demanded ACT which has been used later to offset a liability for MCT is referred to as “utilised ACT”. This offsetting of unlawful ACT against lawfully due MCT took place before GREA issued its proceedings, so their claim in respect of those payments did not include a claim for reimbursement of that utilised ACT. GREA’s claim in respect of ACT relates to what has been described as the time value of money during a period when it had paid tax prematurely.[33]GREA’s claim also raises the same limitation period point as AXAIUK’s claim. Because GREA commenced its claim when it was joined as a claimant to the AXA proceedings in July 2009, it is only payments made in respect of accounting periods within six years before that date that are clearly within time. For any claim relating to payments made in respect of earlier periods, GREA also has to rely on section 32(1)(c).[34]The more fundamental issue raised by GREA’s claim is whether the common law recognises a claim to recover interest in respect of the prematurity period when the principal sum paid by mistake is no longer due to the claimant by the time proceedings are issued. This is the Set Off Issue.[35]As to whether interest should be calculated on a simple or compound basis, GREA is not seeking compound interest, although it maintains that it would be entitled to it. GREA seeks interest under section 35A of the Senior Courts Act 1981 on a simple basis. The figure has been agreed with the Revenue at £5,645,286.72. The Revenue’s case is that GREA’s only entitlement to interest in respect of the prematurity period arises under section 85 of the Finance Act 2019 which results in a materially lesser sum. Nonetheless, the development of the law as to whether compound interest can be claimed in these circumstances is relevant on this appeal. In the Prudential test case, compound interest was claimed and the state of the law at that time has a bearing on the Revenue’s decision to concede that claim for interest in respect of the prematurity period. That concession, as will be seen, is in turn relevant to the principal issue concerning when the court “orders otherwise” under CPR r 19.23.[36]If AXAIUK’s and GREA’s claims were coming before the court for determination today outside the context of the CFC and Dividend GLO, there can be no doubt that their submissions on the Set Off Issue and the Limitation Issue would be rejected. We explain why that is so in paras 80–103 below.[5]The CFC and Dividend GLO: the early stages (a). Setting up the GLO and choosing the test cases[37]The CFC and Dividend GLO was made by Chief Master Winegarten on 30 July 2003, and a group register listing all the cases included in the group was set up. It stated that the claims would constitute the “CFC and Dividend Group Litigation”. The CFC (controlled foreign companies) part of the CFC and Dividend Group Litigation has come to an end, and the Dividend part is concerned with “portfolio” holdings of less than 10% of the shares of the relevant companies.[38]The GLO provided for other claims to be added to the group register. The iteration of the register included in the papers for the court in this appeal, as amended up to 1 January 2025, lists about 60 separate issued claims though some have since been discontinued. The claims are mostly brought by more than one claimant—one claim added to the register in December 2003 has 62 claimants and one claim added in January 2014 has 140 claimants. The earliest claim recorded on the register was issued on 4 April 2003. The AXA proceedings, commenced on 8 April 2003, were added to the register on 22 August 2003.[39]To assist in the management of the Group, the GLO identified lead solicitors and directed the parties to draw up a list of the common issues of fact or law which the claimants contended arose for determination. It also directed the parties to identify which claims they wished to proceed as test cases in the different classes identified in the order. By para 12, the order stayed all claims included in the order other than the test cases.[40]Para 17 of the GLO dealt with the costs of the litigation in accordance with standard practice. It provided that all claimants whose claim fell in one or more of the classes were to be severally liable to the claimant in the test case or cases for those classes for an equal proportion of the common costs incurred by that or those test claimants. All claimants were also severally liable for an equal proportion of the lead solicitor’s costs. The detailed terms upon which the claimants would be liable for common costs were set out in schedule 4 to the order, including the hourly rates for six classes of lawyers from partner to paralegal. Further, the GLO provided that where the court awarded costs to any claimant against the defendants, the claimant had to repay a proportionate share of such recovered common costs to each claimant in whose claim the relevant issue arose.[41]A schedule to the order (as amended) established eight different classes of participants, describing the features of each class, primarily distinguishing between them on the basis of the residence of the payer and recipient of the dividend challenged. Pursuant to the directions given in the GLO, a list of the common GLO issues was agreed by the parties and incorporated into the order on 24 October 2003. The issues were divided into “EU Issues” covering liability, quantum and limitation and “Third Country Issues” also covering liability, quantum and limitation. We will come back to the wording of these issues later. For the moment we note that under both the EU Issues and the Third Country Issues, one of the issues listed was “How should compensation or relief be assessed?” and another was “… from what date does the applicable limitation period start to run?”.[42]As to the choice of test cases, an order of Park J made on 12 December 2003 identified three test cases, of which the claim of the Prudential group, as defined in the order, was to be the test claim in relation to the GLO issues concerning quantum and liability. Initially the claims covered the taxation of dividends received from non-UK resident companies in which it owned a portfolio holding, such as that advanced by AXAIUK. Later claims in respect of utilised ACT were added to the register, including GREA’s claim.[43]The December 2003 order also identified what became known as the first and second phase of the trials of the test claims. The order provided that save for the quantification of the amount of damages and compensation or restitution, all issues in the test claims “including liability for restitution” should be heard together. The order made further directions for trial. (b). The CJEU’s Reasoned Order in the CFC and Dividend Group (“the CFC Reasoned Order”)[44]On 18 March 2005, Henderson J, who was then managing the CFC and Dividend Group, referred a series of questions to the CJEU covering several of the issues raised in the three test cases. The CJEU responded by making a reasoned[45]As to the first group of questions, the CFC Reasoned Order set out the contested domestic tax provisions and the assertions of the parties. In (broad) summary the CJEU’s decision relevant for our purposes was that the provisions challenged by the test claims were contrary to the principle of free movement of capital: see para 43. The CFC Reasoned Order also addressed the questions about the consequences of any incompatibility of the tax regime with EU law. In answer to those questions, the CFC Reasoned Order said:(i) It is not for the CJEU to assign a legal classification to the actions brought before the national court by the claimants. It is for the claimants to specify the nature and basis of their actions, whether they are actions for restitution or actions for compensation for damages, subject to the supervision of the national court: para 111.(ii) However, it is well established that the right to a refund of charges levied in a member state in breach of EU law is the consequence and complement of the rights conferred on individuals by Community provisions as interpreted by the CJEU. The member state is therefore required in principle to repay charges levied in breach of Community law: para 112.(iii) In the absence of harmonising measures, it is for the domestic system of each member state to lay down detailed procedural rules governing actions provided that they adhere to the principles of effectiveness and equivalence, that is to say provided, first, that such rules are not less favourable than those governing similar domestic actions and, secondly, that they do not render virtually impossible or excessively difficult the exercise of rights conferred by Community law: para 113.(iv) In addition, where a member state has levied charges in breach of the rules of Community law, individuals are entitled to reimbursement not only of the tax unduly levied but also of the amounts paid to that State or retained by it which relate directly to that tax. That also includes losses constituted by the unavailability of sums of money as a result of a tax being levied prematurely: para 114.[46]It is that last point, namely that the losses suffered by a taxpayer who pays tax prematurely includes “losses constituted by the unavailability of sums of money” that gives rise to the prematurity period issue in this appeal.[47]The CFC Reasoned Order referred to the judgment of the CJEU in Test Claimants in the FII Group Litigation v Inland Revenue Comrs (Case C-446/04) [2012] 2 AC 436; [2006] ECR I-11814 (judgment of 12 December 2006) (“FII CJEU 1”). That judgment had been given in answer to a reference to the CJEU from the High Court in proceedings in a different group litigation concerned with franked investment income (“FII”). The FII litigation group, though different from the CFC and Dividend Group, gave rise to many of the same legal issues as were raised by the latter group. Some of the judgments in the FII proceedings on those issues were handed down by the CJEU and the English courts during the period when the Prudential test case was going through the courts.[48]The judgment in FII CJEU 1 in December 2006 itself referred back to an earlier judgment of the CJEU in Metallgesellschaft Ltd v Inland Revenue Comrs and Hoechst AG v Inland Revenue Comrs (Joined Cases C-397/98 and 410/98) [2001] Ch 620; [2001] ECR I-1727 (judgment of 8 March 2001) (“Hoechst”). Both FII CJEU 1 and Hoechst addressed the compatibility of UK taxing provisions with EU law and so also addressed the consequences of any such compatibility. In paras 87‒89 of the judgment in Hoechst, the CJEU had stressed that a taxpayer is entitled to recover compensation for the loss of the use of money as a result of the unlawful premature levy of tax and that the principal sum due is equivalent to the interest which that money would have earned during that period. The judgment in FII CJEU 1 in 2006 was thus applying the principle that had already been established in 2001 in Hoechst and the CFC Reasoned Order in turn in 2008 was repeating that same principle, namely that, as the CJEU put it in Hoechst:
“87 … where the breach of Community law arises, not from the payment of the tax itself but from its being levied prematurely, the award of interest represents the ‘reimbursement’ of that which was improperly paid and would appear to be essential in restoring the equal treatment guaranteed by article 52 of the Treaty.”
[49]We do not need to analyse the reasoning of the CJEU in those two earlier cases FII CJEU 1 and Hoechst so far as the findings of the incompatibility of the UK tax regime with EU law are concerned. But the dates on which those judgments were handed down (12 December 2006 and 8 March 2001 respectively) have proved to be significant in relation to the Limitation Issue and the application of section 32(1)(c). Those dates have been put forward as candidates for the date on which a taxpayer in the position of Prudential or the other claimants in the CFC and Dividend GLO including AXAIUK could with reasonable diligence have discovered that they had a worthwhile claim based on the illegality of the tax demands made by the Revenue.[50]Following the CJEU’s issue of the CFC Reasoned Order, the proceedings in the Prudential test case resumed before Henderson J. Although the Reasoned Order answered the questions he had referred, the CJEU’s answers raised many further questions as to how it should be implemented in the domestic legal regime. Before considering what Henderson J decided in his judgment of 24 October 2013 in the Prudential test case (Prudential Assurance Co Ltd v Revenue and Customs Comrs [2013] EWHC 3249 (Ch); [2014] STC 1236 (“Prudential HC”)), we need to describe the law as it was at that time.[6]The state of the law at the time of Henderson J’s judgment in Prudential HC[51]In Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349 (“Kleinwort Benson”),the House of Lords overruled the long-standing decision in Bilbie v Lumley (1802) 2 East 469 and held that money paid under a mistake of law (as well as under a mistake of fact) is recoverable. In Deutsche Morgan Grenfell Group plc v Inland Revenue Comrs [2006] UKHL 49; [2007] 1 AC 558 (“DMG”), the House of Lords clarified that unlawfully charged tax, as well as money paid under ordinary private transactions such as those in Kleinwort Benson, was recoverable in a common law action for restitution as money paid under a mistake of law. Such a claim was in addition to a claim for the recovery of taxes exacted ultra vires as established by the House of Lords in Woolwich Equitable Building Society v Inland Revenue Comrs [1993] AC 70 (a “Woolwich claim”). Where unlawful tax has not been repaid or set off before proceedings for its recovery are issued, simple interest may be claimed under section 35A of the Senior Courts Act 1981, running from the date that the unlawful tax was paid until the date of its repayment.[52]In this section we consider three different but related legal topics that flow from that and are directly relevant to the issues in this appeal: the existence of a common law remedy for a taxpayer in respect of the loss of use of money during the prematurity period; whether the obligation under EU law to provide taxpayers with an effective remedy requires an award of compensation calculated by reference to compound interest rather than simple interest for the prematurity period; and the date when the taxpayer is treated as having been reasonably able to discover its mistake of law for the purposes of section 32(1)(c). (a). A common law cause of action in respect of the loss of use of money during the prematurity period: the state of the law in 2013[53]Following the CJEU’s judgment in Hoechst on 8 March 2001,a group litigation order was made on 26 November 2001 to manage the claims brought against the Revenue by companies asserting the claim arising from the illegality identified in the taxing provisions at issue in Hoechst. This was called the ACT GLO. Sempra Metals Ltd was the same company as Metallgesellschaft Ltd which had been a party in the Hoechst reference to the CJEU and its claim joined the group register. Sempra Metals Ltd was a UK-resident metals trader and founded its claims on the amounts of ACT it had paid in respect of four sample dividends paid to its parent company which was resident in Germany. The intervals between the payment of the ACT and the set off of the ACT against Sempra’s liability for mainstream corporation tax ranged from just under one year to almost ten years. In all but one case, the set off of the ACT occurred before the issue of the writ. It became a test claim on the issue of the remedy for the UK’s breach of EU law.[54]The progress of the ACT GLO test case, Sempra Metals, was as follows:(a) Sempra Metals Ltd v Inland Revenue Comrs [2004] EWHC 2387 (Ch); [2004] STC 1178, judgment of Park J of 16 June 2004 (“Sempra Metals HC”).(b) Sempra Metals Ltd v Inland Revenue Comrs [2005] EWCA Civ 389; [2006] QB 37, judgment of the Court of Appeal (Chadwick, Laws, Jonathan Parker LJJ) of 12 April 2005 (“Sempra Metals CA”).(c) Sempra Metals Ltd v Inland Revenue Comrs [2007] UKHL 34; [2008] AC 561, judgment of the House of Lords of 18 July 2007 (“Sempra Metals HL”).[55]The Sempra Metals judgments were important for two of the three issues we have outlined above; the existence of the cause of action in English law for loss of the use of money during the prematurity period and whether the quantification of that loss was an amount equivalent to compound interest on that sum or only simple interest. Here we describe what Sempra Metals decided on the first point.[56]In Sempra Metals HC, Park J noted that between 50 and 70 claimants which were participants in the GLO were awaiting the outcome of Sempra’s case: para 22. Before Park J and in the Court of Appeal the focus was on CJEU’s statement in Hoechst that EU law required member states to provide a remedy in respect of the prematurity period. The principal question in the judgments of Park J and the Court of Appeal in Sempra Metals was not so much on the classification in English law of that right but rather on whether EU law required an award calculated by reference to compound rather than simple interest. Park J recorded at para 16(iv) the Revenue’s argument that English law did not recognise a cause of action which simply claimed interest on a debt paid late, citing the decision of the House of Lords in President of India v La Pintada Compania Navigacion SA [1985] 1 AC 429 (“La Pintada”). He paraphrased the CJEU’s response to that as being that whatever the position might be under English law, the remedy could not be denied where there was a breach of an article of the EC Treaty. He held that the question of remedy was a matter of EU law untrammelled by restrictions that English law may impose in comparable situations: paras 25 and 26.[57]The Court of Appeal upheld his decision save for adjusting the interest rates to be applied. The Court also recognised that English law did not give a remedy where there was no principal sum outstanding at the time when the legal action was commenced. That followed from the decision of the House of Lords in La Pintada. However, the task of the court, they held, was to give the remedy that EU law requires in circumstances where domestic law would not provide a remedy: para 41. Community law required full compensation for the loss of the use of money.[58]On appeal, the House of Lords in Sempra Metals HL went considerably beyond what Park J and the Court of Appeal had decided. They decided not to follow La Pintada. They held that an award of damages for breach of contract to pay a debt could include interest losses caused by late payment of a debt. More important for our purposes, they held that a court had jurisdiction to award compound interest where a claimant was seeking restitution of money paid under a mistake either (as per the majority of the House) in the exercise of the court’s common law restitutionary jurisdiction or (as per the minority) in the court’s exercise of the court’s discretionary equitable jurisdiction. There was therefore a claim to recover interest on a debt paid late, even if it had been paid before the start of proceedings. (b). Whether EU law required an award of compound interest for unlawfully demanded tax: the law in 2013[59]The issue of whether EU law requires a member state to calculate the remedy for loss of use during the prematurity period of the unlawfully demanded tax by reference to compound interest in order to ensure an effective remedy was also addressed by Park J and the Court of Appeal in Sempra Metals. Park J held that only an award of compound interest would fully restore the equal treatment required by EU law and remove the discrimination held by the CJEU to have been incompatible with EU law: para 27. Park J went on to hold that compound interest should be calculated on a conventional basis and so would be awarded at the same rate for all claimants. The Court of Appeal upheld his ruling on this point save for adjusting the interest rates to be applied. The House of Lords did not strictly have to address whether this was an EU law requirement since they held that there was such a remedy under English law anyway. The House held that in principle it would always be open to a claimant to plead and prove his actual interest losses caused by late payment of a debt. Accordingly, the courts had a common law jurisdiction to award interest, simple and compound, as damages on claims for non-payment of debts as well as on other claims for breach of contract and tort. That jurisdiction extended to where a claimant was seeking restitution of money paid under a mistake. Lord Nicholls took the view that the remedy in English law was therefore “plainly” an effective remedy for EU purposes, and Lord Hope and Lord Walker appear to have agreed.[60]However, in another part of this litigation forest, and interspersed with the judgments about the recovery of unlawfully paid corporation tax, a challenge to the legality of the UK’s regime for reimbursing unlawfully levied value added tax (“VAT”) was progressing through the courts. The claimants (“Littlewoods”) in that litigation begun in 2007 had submitted claims pursuant to sections 78 and 80 of the Value Added Tax Act 1994 (“VATA”) for the overpayment of VAT going back to 1973. The Revenue had repaid the VAT together with simple interest but Littlewoods claimed to be entitled to compound interest on the ground that such interest was due under common law.[61]The timeline for the Littlewoods litigation so far as completed before the judgment in Prudential HC in October 2013 was as follows:(a) Littlewoods Retail Ltd v Revenue and Customs Comrs [2010] EWHC 1071 (Ch); [2010] STC 2072, judgment of Vos J of 19 May 2010 (“Littlewoods HC 2010”).(b) Littlewoods Retail Ltd v Revenue and Customs Comrs (Case C-591/10) ECLI:EU:C:2012:478; [2012] STC 1714, judgment of the CJEU 19 July 2012 (“Littlewoods CJEU”).[62]In Littlewoods HC 2010,Vos J held as a matter of statutory construction that, if a common law claim to compound interest existed, it was excluded by the statutory scheme for reimbursement of overpaid VAT. That scheme provided only for simple interest on the sums reimbursed. Vos J made a reference to the CJEU asking the question whether the absence of a remedy calculated on the basis of compound interest for overpaid VAT was incompatible with EU law. The answers provided by the CJEU in Littlewoods CJEU were, it turned out, open to different interpretations. What the CJEU said was that EU law required the payment of interest but that “It is for national law to determine, in compliance with the principles of effectiveness and equivalence, whether the principal sum must bear ‘simple interest’, ‘compound interest’ or another type ofinterest”. See para 34 of the judgment. (c).

(c) to mistake of law claims: the law in 2013

[63]Finally, there is the Limitation Issue, that is to say the issue of the correct test in section 32(1)(c) for determining when the claimant could with reasonable diligence have discovered the mistake of law for the consequences of which they now seek relief. The House of Lords in Kleinwort Benson addressed whether section 32(1)(c) would apply to mistakes of law and held, by a majority, that it did: see Lord Goff at pp 387G‒389. Although there was discussion in some of the speeches on precisely when the limitation period would start, the House did not need to address the issue and did not decide it. The issue was squarely addressed by the House of Lords in DMG, the decision in 2006 to which we have already referred as clarifying that the common law mistake of law claim existed in relation to taxes alongside the Woolwich claim. The House held that the date of constructive discovery in a mistake of law claim was the date of the judicial ruling that established the unlawfulness of the relevant UK taxing provision. The majority held that DMG’s mistake could reasonably have been discovered for the purposes of section 32(1)(c) only when the CJEU handed down its judgment in Hoechst, that is to say on 8 March 2001: see per Lord Hoffmann at para 31, Lord Hope at para 71 and Lord Walker at para 144.[64]Meanwhile Parliament had attempted to address the fact that a combination of judicial decisions which both(i) compounded the interest payable when tax that had been unlawfully demanded in some cases since 1973 and(ii) postponed the start of the limitation period under section 32(1)(c) so as to prevent such claims being time barred looked likely to create liabilities for the Exchequer of many millions of pounds.[65]The first attempt was a response to a decision of Park J in the High Court in a judgment arising from the ACT GLO litigation in Deutsche Morgan Grenfell Group plc v Inland Revenue Comrs [2003] EWHC 1779 (Ch); [2003] 4 All ER 645. That was the judgment that was ultimately confirmed by the House of Lords in DMG (in October 2006) as described above. On 8 September 2003, the Government announced that it would introduce legislation to exclude the operation of section 32(1)(c) in tax cases. This was enacted as section 320 of the Finance Act 2004, which had effect from 24 June 2004. Section 320 provided that section 32(1)(c) “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”. However, it only applied in relation to actions brought on or after 8 September 2003, the date of the Government’s announcement.[66]Following the decision of the House of Lords in DMG in October 2006 confirming Park J’s decision as to the date of reasonable discovery under section 32(1)(c), the Government announced further legislation to exclude the application of section 32(1)(c) for mistake claims relating to taxation matters. That measure was enacted as section 107 of the Finance Act 2007. It purported to disapply section 32(1)(c) retrospectively to any mistake claims which had been filed before 8 September 2003.[67]The lawfulness of these statutory provisions was then the subject of a challenge based on EU law. In Test Claimants in theFII Group Litigation v Revenue and Customs Comrs [2012] UKSC 19; [2012] 2 AC 337 (“FII SC 1”) the Supreme Court unanimously held that section 107 of the Finance Act 2007 was incompatible with EU law because by 2006 the claimants had a legitimate expectation that their claims would be adjudicated. That remedy could not be removed retrospectively by the introduction of a limitation period without some transitional notice. The court was, however, split on whether section 320 of the Finance Act 2004 (which primarily applied prospectively) was also incompatible with EU law. The majority of the court held that section 320 was incompatible, with Lord Sumption and Lord Brown dissenting. The court therefore referred the question to the CJEU and that became Case C-362/12.[68]At the time that Henderson J handed down his judgment in Prudential HC, Advocate General Wathelet had delivered his opinion (on 5 September 2013) in Case C-362/12, the reference made by the Supreme Court following FII SC 1. He agreed with the majority view of the Supreme Court that section 320 was contrary to the principle of effectiveness. Further, he concluded that it was also contrary to the principles of legal certainty and the protection of legitimate expectations. The decision of the CJEU was awaited at the time of the hearing before Henderson J in the Prudential test case. Some of the claimants had been added to the CFC and Dividend GLO register after 8 September 2003 so, if section 320 was valid, their claims in respect of early payments would be time barred because section 32(1)(c) would be disapplied by section 320.[69]We note here that by the time Henderson J made the order following his judgment in Prudential HC on 28 January 2014, the CJEU had handed down its judgment in Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (Case C-362/12)EU:C:2013:834;[2014] AC 1161 (judgment of 12 December 2013) (“FII CJEU 3”). The CJEU agreed with the Advocate General so that by the date of the order, it was clear that section 320 of the Finance Act 2004, as well as section 107 of the Finance Act 2007, was invalid and did not affect the application of section 32(1)(c) in relation to these claims.[7]The judgment and order in Prudential HC[70]Only two of the 12 substantial sections of Henderson J’s judgment are relevant to the issues in this appeal, issue X relating to remedies and issue XI headed “Limitation Issues”. As to remedies, he set out at para 162 the three agreed issues relating to remedies, the relevant ones for our purposes being whether the claimant paid tax by mistake so as to found a claim in mistake and, if so, the measure of the restitution and “on what basis is interest payable”. As to the first he noted that there was a large measure of common ground in relation to either corporation tax or ACT which had been unlawfully levied. The Revenue did not dispute that the claims were San Giorgio claims under EU law; that the overpaid tax, or its time value in the case of utilised ACT, “is in principle recoverable by either a Woolwich claim or a mistake-based restitutionary claim, subject to defences and limitation”; and that the tax was in fact paid under an operative mistake, the mistake being that it was lawfully due and payable. As to ACT claims it was also agreed that unlawful ACT which was utilised against lawful MCT “is recoverable, on the same basis as in Hoechst” (para 164(a)). In light of that agreement, the first principal issue for Henderson J was whether the Revenue could rely on a change of position defence and he held they could not both as a matter of EU law (para 189) and because the Revenue had not adduced any evidence to support the defence (para 193).[71]He then turned to the question of interest. He set out para 194 three periods over which the claimants asserted that interest should be compounded. These subparagraphs then became the short-hand terms for the different kinds of claim:(a) Category (a) is what we have referred to as the “principal amount” due in respect of the prematurity period, that is to say, interest compounded over the period before which the unlawfully levied ACT was subsequently set off against MCT, so from the day of payment by the claimants to the date of utilisation.(b) Category (b) relates to all other unlawfully levied tax from the date that the tax is paid to the date it is reimbursed by the Revenue.(c) Category (c) relates to interest that accrues on the principal sum under (a) from the date that the unlawfully levied ACT is utilised until that principal sum is reimbursed by the Revenue.[72]Henderson J said at para 195 that the Revenue accepted that compound interest was payable in respect of the utilised ACT claims (that is category (a)) because that is what the House of Lords decided in Sempra Metals. But they argued that EU law did not require compound interest to be awarded for the other categories and that simple interest under section 35A of the Senior Courts Act 1981 would provide the claimants with an “adequate indemnity” in accordance with what the CJEU had said in Littlewoods CJEU. He then described the case law relating to recovery of overpaid VAT leading up to the Littlewoods CJEU judgment. What emerged clearly from that judgment was that the right to the reimbursement of unlawfully levied tax included a right to receive interest on that tax: para 206. But the CJEU had declined to rule that the right to interest meant the right to compound interest: para 208. However, relying on Sempra Metals HL, he held that as a matter of English common law the claimants were entitled to compound interest on all their claims.[73]He analysed each of the speeches in detail and expressed his conclusions at paras 241 onwards. He repeated that it was common ground that the position as regards claims to utilised ACT was governed by Sempra Metals:
“Accordingly, compound interest is payable on the amount of the ACT prematurely paid, from the date of its payment until the date of setting-off against MCT, at conventional government rates.”
There was no rational basis for distinguishing claims for wrongly levied and unutilised ACT (that is category (b)) and for unlawfully charged corporation tax: paras 243 and 244. As to the claim for compound interest in respect of the post-utilisation period for utilised ACT (that is his category (c)), the approach of the majority in Sempra Metals HL “should logically lead to the conclusion that compound interest is also available in respect of the post-utilisation period”.[74]Turning to limitation issues at para 248, the two issues were expressed as being to what extent is the claim statute barred by a six year limitation period and was the claim barred by section 320 of the Finance Act 2004—a question that was still outstanding at that time as the judgment of the CJEU in FII CJEU 3 was awaited: paras 253–255.[75]We consider in more detail below how Henderson J approached the two relevant claimants in the Prudential test case. At this stage it is enough to note that he said that if section 320 was invalid, there would be nothing to prevent the claimant from relying on section 32(1)(c). Further, he said he did not understand the Revenue to argue that Prudential could with reasonable diligence have discovered its mistake before 14 July 1998 (that is six years before it had issued proceedings on 14 July 2004). He said “In practice, therefore, I can see no obstacle to [the second claimant] pursuing its mistake-based claims for periods before July 1998, always assuming that s 320 is invalid”: para 255.[76]The order drawn up following the Prudential HC judgment recorded in the recitals that certain matters were common ground and that the Revenue had confirmed that it was not putting forward particular arguments in relation to some issues and that it had abandoned certain other points. It then declared that the GLO Issues not otherwise agreed as set out previously were answered in the following paragraphs. Para 8 stated:
“A. In relation to those claims which are upheld these issues do not arise for decision as it is common ground that: i) such claims are to be characterised as San Giorgio claims under EU law; ii) the overpaid tax (or its time value in the case of utilised ACT) is in principle recoverable by either a Woolwich claim or a mistake-based restitutionary claim, subject to defences and limitation; iii) the tax was in fact paid under an operative mistake, the mistake being that it was lawfully due and payable; iv) unlawful ACT which was utilised against lawful MCT is recoverable, on the same basis as in Hoechst; v) unlawful ACT which was utilised against unlawful MCT is also recoverable, on the basis that the purported charge was a nullity; and vi) lawful ACT which was utilised against unlawful MCT is recoverable because the Court of Appeal so held in FII (CA), but this is subject to the Revenue's pending application to the Supreme Court for permission to appeal against that conclusion.”
[77]At para 10 the order stated:
“Compound interest computed on the conventional government basis is payable in respect of all claims which are upheld, namely, overpaid corporation tax, unutilised ACT and ACT utilised against lawful MCT for the periods both before and after utilisation (Issue Vll.2).”
[78]Para 11 stated:
“Issues VIII.1–2 are answered in light of the judgment in Case C-362/12 as follows: The claims in mistake-based restitution (that is those successful claims listed in paragraph 8.A above) are not subject to the limitation period in section 320 of the Finance Act 2004 and are in time.”
[79]The GLO issues were set out in a schedule to the Order and Issues VIII.1–2 asked to what extent is the claim statute barred by a six year limitation period and to what extent is the claim barred by section 320.[8]Developments in the law after Prudential HC[80]Each of the three legal topics we described in section 4 underwent a major reconsideration following Henderson J’s judgment in October 2013 in Prudential HC and most of what was decided is now no longer good law. There are three strands of litigation that one needs to follow to see how the law developed.[81]First, there is the further progress of the Prudential test case itself following Prudential HC:(a) Prudential Assurance Co Ltd v Revenue and Customs Comrs [2016] EWCA Civ 376; [2017] 1 WLR 4031, judgment of the Court of Appeal of 19 April 2016 (“Prudential CA”).(b) Prudential Assurance Co Ltd v Revenue and Customs Comrs [2018] UKSC 39; [2019] AC 929, judgment of the Supreme Court of 25 July 2018 (“Prudential SC”).[82]Secondly, there is the further progress of the FII GLO domestic proceedings following the CJEU’s decision in FII CJEU 1 which we have described as the precursor to the CJEU’s CFC Reasoned Order (see para 47 above). Of the many judgments given in those proceedings after Prudential HC we need to refer to the following:(a) Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2014] EWHC 4302 (Ch); [2015] STC 1471, judgment of Henderson J of 18 December 2014 (“FII HC 2”)(b) Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2016] EWCA Civ 1180; [2017] STC 696, judgment of the Court of Appeal of 24 November 2016 (“FII CA 2”)(c) Test Claimants in the FII Group Litigation v Revenue and Customs Comrs[2020] UKSC 47; [2022] AC 1, judgment of the Supreme Court of 20 November 2020 (“FII SC 2”)(d) Test Claimants in the FII Group Litigation v Revenue and Customs Comrs[2021] UKSC 31; [2021] 1 WLR 4354, judgment of the Supreme Court of 23 July 2021 (“FII SC 3”) on appeal and cross-appeal from FII CA 2 and from an earlier judgment of 23 February 2010 [2010] EWCA Civ 103; [2010] STC 1251 (“FII CA 1”).[83]Thirdly there was the further progress in the Littlewoods litigation concerning the recovery of overpaid VAT once the CJEU had handed down its judgment on 19 July 2012 on the reference made in 2010 by Vos J:(a) Littlewoods Retail Ltd v Revenue and Customs Comrs [2014] EWHC 868 (Ch); [2014] STC 1761, judgment of Henderson J of 28 March 2014 (“Littlewoods HC 2014”).(b) Littlewoods Ltd v Revenue and Customs Comrs [2015] EWCA Civ 515; [2016] Ch 373, judgment of the Court of Appeal of 21 May 2015 (“Littlewoods CA”).(c) Littlewoods Ltd v Revenue and Customs Comrs [2017] UKSC 70; [2018] AC 869, judgment of the Supreme Court of 1 November 2017 (“Littlewoods SC”). (a). A common law cause of action in respect of the loss of use of money during the prematurity period: developments after 2013[84]The Prudential CA judgment opened noting that the litigation had been going on for over 13 years:
“Whether the final end is in sight after all this time remains to be seen”
. The main issues dealt with in that judgment concerned the correct conforming interpretation to give to the legislation to remove its incompatibility with EU law and how to calculate the overpaid tax. The court turned to the question of compound interest at para 153. By this time the Littlewoods case had been decided by the Court of Appeal holding that the CJEU’s judgment in that case did require compound interest to be awarded (a decision later overturned by this court). Since the Court of Appeal’s decision was binding, the court in Prudential CA did not go into further detail: para 155. The only point which was raised on the underlying question of restitutionary remedy was the change of position defence which was summarily rejected by the court: para 147 onwards. There was no discussion of Sempra Metals. There was also no discussion of the limitation period issues because the Revenue’s application for permission to appeal on those issues had been refused: see para 158 of the judgment.[85]When the case reached the Supreme Court matters took a very different turn. At para 34 the court set out the three categories of interest to which Henderson J had referred. The court then recorded that the Revenue had accepted that compound interest was payable in respect of the utilised ACT falling within category (a), since that is what the House of Lords decided in Sempra Metals HL. But the Revenue argued that in relation to categories (b) and (c), only simple interest should be awarded since that would provide an “adequate indemnity” in accordance with what the Supreme Court had, by that time, decided in Littlewoods SC was the correct interpretation of Littlewoods CJEU. It was therefore no longer asserted that there was a right to compound interest under EU law. The question remained whether there was such a right at common law. The court noted that although the difference between simple and compound interest was modest in that case, the point arose in other cases pending against the Revenue and the total amount at stake on the Revenue’s estimate was of the order of £4–5 billion: para 36. The Revenue was not seeking to withdraw the concession as regards category (a) but Prudential still argued that the Revenue should not be permitted to contest categories (b) or (c) either since this questioned the soundness of the reasoning in Sempra Metals. The court, however, allowed the Revenue to challenge that reasoning since there had been significant developments in the law of unjust enrichment since the trial before Henderson J.[86]At para 44, the court summarised the decision of the majority in Sempra Metals HL:
“a claim would lie in unjust enrichment for restitution of compound interest on money which had been paid prematurely as the consequence of a mistake, and that the appropriate measure of restitution in the instant case was compound interest calculated on a conventional basis applicable to government borrowing.”
[87]The court referred to the Littlewoods proceedings, the aftermath of the retrospective effect of the Sempra Metals decisions in conjunction with section 32(1)(c) and the failure of attempts by Parliament to mitigate it. Paras 62–66 make clear that this court regarded what had happened as a cautionary tale against “the risks of effecting major changes to the law of restitution by judicial decision”: (para 63). Following an analysis of the case law, in particular Investment Trust Companies v Revenue and Customs Comrs [2018] AC 275, (another test case concerned with the restitution of VAT paid incompatibly with EU law), the court concluded that there was no unjust enrichment arising from the payment of money by mistake beyond the benefit of the actual payment—there was no additional transfer of value or benefit comprising the ability to use that money. What there could be is the award of interest on the debt under section 35A of the Senior Courts Act 1981 to compensate for the loss of the use of the money that should have been repaid earlier. It followed that interest could be awarded on moneys within categories (b) and (c) (assuming that the unlawful tax had not been refunded by the time proceedings were issued) but there was no restitutionary claim. The court stated that Prudential’s claim to compound interest under category (a) would also have been rejected if it had not been accepted by the Revenue: para 79.[88]An opportunity arose to revisit category (a) in FII SC 3. That appeal raised a number of issues that had been decided in the Revenue’s favour in Prudential SC but there still remained the question of compound interest for the prematurity period. The claimants argued that the Revenue were barred from denying their entitlement to compound interest for that period. There was, the claimants said, a definitive finding in the first phase of the litigation that their claim to recover the time value of money in the period of prematurity succeeded; that that claim was recognised as a claim in restitution, and that the parties had agreed in accordance with the judgment of the House of Lords in Sempra Metals that compound interest should be paid: para 58. The court rejected the arguments based on res judicata, issue estoppel, lack of jurisdiction and abuse of process. By this time sections 85 and 86 of the Finance Act 2019 had been enacted so there was now a statutory regime for providing taxpayers with the remedy that the CJEU had held in Hoechst was necessary to compensate for the loss of the use of money during the prematurity period. That meant that there was no longer any need to debate whether section 35A of the Senior Courts Act 1981 (which on its literal wording only empowered the court to award interest when the debt was itself the subject of the proceedings) could be purposively construed as also empowering a court to award interest when the debt had been repaid before proceedings were issued. The Revenue’s appeal was therefore allowed on this point.[89]It was therefore finally decided that there is no restitutionary claim at common law to the payment of interest on a debt paid late; taxpayers in the position of GREA have the statutory remedy under section 85 of the Finance Act 2019. (b). Whether EU law required an award of compound interest for unlawfully demanded tax: developments after 2013[90]Following the CJEU’s judgment in Littlewoods CJEU (in July 2012) the trial in those VAT recovery proceedings resumed in the High Court before Henderson J. In his judgment, Littlewoods HC 2014, Henderson J interpreted the CJEU’s ruling as establishing that an award calculated by reference to the compound interest that the overpaid VAT would have earned would satisfy Littlewoods’ rights under EU law. He held that sections 78 and 80 VATA, in so far as they excluded any such right, had to be disapplied to allow Littlewoods to pursue their claims. On appeal both by the claimants against the decision of Vos J in 2010 and by the Revenue against the decision of Henderson J in 2014, the Court of Appeal in Littlewoods CA affirmed both decisions. They therefore agreed with Vos J that sections 78 and 80 were a complete code for the recovery of interest which excluded any common law claim. But they also agreed with Henderson J that that those sections must be disapplied as regards VAT levied in breach of EU law because they violated the EU principle of effectiveness and that principle required as a matter of EU law an award of compound interest.[91]On appeal, the Supreme Court disagreed as regards the application of EU law. In Littlewoods SC (1 November 2017), the court agreed that the scheme for reimbursement of overpaid VAT created by sections 78 and 80 was a complete code and was therefore inconsistent with the availability of concurrent common law claims. As a matter of domestic law, therefore, the statutory scheme impliedly excluded any such claim: para 40.[92]The court went on to address whether the CJEU in Littlewoods CJEU had held that a taxpayer’s EU rights to reimbursement of overpaid VAT included a right to receive compound interest on the principal sum. They noted that there was no question concerning compliance with the principle of equivalence arising on the appeal; it turned on what the CJEU meant by “adequate indemnity” in para 29 of its judgment. The court reached “a different but none the less clear view” as to the meaning of the CJEU’s judgment in Littlewoods CJEU. Having analysed different language versions of the phrase “adequate indemnity” the court concluded that the CJEU had given member state courts a discretion to provide reasonable redress in the form of interest in addition to the mandatory repayment of any wrongly levied tax, interest and penalties (para 51). That key phrase was not tied into the idea of full compensation for the time value of money and the case law prior to Littlewoods CJEU suggested that there was no general principle of EU law that there must be full reimbursement of the use value of money. The court noted at para 60:
“Turning to the wider context of member state practice, the United Kingdom Government in its written observations to the CJEU examined the legislation in 13 other member states (Belgium, Denmark, Finland, France, Germany, Hungary, Ireland, Italy, Luxembourg, The Netherlands, Poland, Spain, and Sweden). In all but one (Sweden) simple interest is payable both on the recovery by taxpayers of taxes which were unduly paid and on the recovery by the tax authorities of taxes paid late. Thus, there was and is a widespread practice within the member states of the EU which is the same as that in the United Kingdom. In this context, if the CJEU were seeking to outlaw this practice, we would have expected clear words to that effect. They are absent.”
[93]The Supreme Court’s judgment in Littlewoods SC therefore established that, at least for VAT demanded in breach of EU law, the taxpayer’s EU right to an effective remedy did not extend to an award of compound interest but could be limited by domestic law to simple interest.[94]Given what was decided by this court in Prudential SC, it is clear that taxpayers who have paid other kinds of unlawfully demanded tax can be in no better position. (c).

(c) to mistakes of law: developments after 2013

[95]At the time of Prudential HC and the order made by Henderson J to give effect to that judgment, the prevailing law as to when a taxpayer could with reasonable diligence have discovered its mistake for the purposes of section 32(1)(c) was that it was the date on which a court had finally determined that the relevant statutory provisions—or perhaps analogous provisions—were incompatible with the law. The Supreme Court and the CJEU had also held that Parliamentary attempts to mitigate the consequences of this by legislation disapplying section 32(1)(c) were invalid and dropped out of the picture: see paras 64–69 above.[96]Some years after DMG, the same issue came up to the court in the FII GLO. In FII SC 2, a panel of seven justices unanimously overruled DMG. In its judgment of 20 November 2020 the court held that the limitation period started to run at an earlier time ie when the taxpayers could reasonably have realised that they had a worthwhile claim that the regime was unlawful. In their judgment (with which Lord Lloyd-Jones and Lord Hamblen agreed) Lord Reed and Lord Hodge recorded that the Revenue had initially limited its submissions about the date of reasonable discovery to an argument about which CJEU judgment should have alerted the FII claimants to their mistake and so triggered the start of the limitation period for the purposes of section 32(1)(c), applying the DMG test. The Revenue had argued that the date when the taxpayer could reasonably have discovered its mistake was not, as the taxpayer contended, 12 December 2006 (the date of the judgment in FII CJEU 1) but 8 March 2001 (the date of the CJEU’s judgment in Hoechst). The FII CJEU 1 date was the date on which the CJEU had first held that the statutory provisions which were directly at issue in the FII GLO were incompatible with EU law and so would, applying the test laid down in DMG, be the relevant date. The Revenue argued for the earlier date of the Hoechst judgment on the basis that that litigation “turned on closely analogous points of EU law”: see the Revenue’s argument recorded at p 20C of the law report.[97]However, by the time the appeal came before this court, the Revenue challenged the correctness of DMG in tying the date to a final judgment of a court which authoritatively determined the illegality of the precise or analogous statutory provisions. The court overruled DMG and held that the test for ascertaining the date of discoverability for the purposes of section 32(1)(c) requires the court to identify the point in time when the claimant knows, or could with reasonable diligence know, that he had made a mistake of law with sufficient confidence to justify embarking on the preliminaries to the issue of a claim form, such as submitting a claim to the proposed defendant, taking advice and collecting evidence. To put it another way, it was the date when the claimant discovers or could with reasonable diligence discover the mistake in the sense of recognising that a worthwhile claim arises: see paras 193, 209(1) and 213(13) of the judgment of Lord Reed and Lord Hodge.[98]The court remitted the case to the High Court to determine the date of the start of the limitation period. That date was then determined by Richards J in a further test case in the FII GLO: see BAT Industries plc v Inland Revenue Comrs [2024] EWHC 195 (Ch); [2024] STC 305 (upheld on appeal: [2025] EWCA Civ 1271; [2025] STC 1718).[9]The answer to the issues on the current state of the law[99]On the current state of the law, as regards the Set Off Issue, GREA would have no claim at common law for compensation by way of interest or otherwise for the loss of the time value of the ACT it had paid during the prematurity period, as held by this court in FII SC 3.[100]Instead, GREA has the statutory remedy provided by section 85 of the Finance Act 2019. That section applies where proceedings have been brought against the Revenue on any date before 12 December 2012 and they include a claim for unlawful ACT which was paid on or after 1 January 1996 or within six years ending with the start of proceedings, but where the ACT was set off or repaid before the proceedings were started. In such proceedings the claimant is entitled to claim(a) “the principal amount”, namely simple interest during the prematurity period, and(b) simple interest on the principal amount from the end of the prematurity period until the date that the principal amount is paid. Section 85(4) sets out the interest rate payable for different periods, starting with 1 October 1993 to 31 March 1997 (8%) and ending with 27 January 2009 to 29 October 2018 (0.5%). The rate from 30 October 2018 is 0.5%, subject to the power of the Treasury to specify other rates. The effect of the Court of Appeal’s decision in the present case is that GREA’s only entitlement to interest in respect of the prematurity period arises under section 85. If that is correct, then the limitation point does not arise in relation to GREA’s claim because the matter is governed by section 85 and there is no claim for mistake of law to which section 32(1)(c) applies.[101]As regards AXAIUK’s claim to recover its unlawfully demanded corporation tax, it is common ground that AXAIUK has a common law claim for the recovery of the unlawful corporation tax charged on its portfolio dividends as money paid under a mistake of law, based as we earlier explained on the decisions of the House of Lords in Kleinwort Benson and DMG. Further, it is common ground that AXAIUK is entitled to rely on the change of the law made by Kleinwort Benson and DMG to assert a cause of action that arose from payments it made at a time when Bilbie v Lumley was still binding authority that no such cause of action existed.[102]The Limitation Issue arises because AXAIUK’s claims extend to payments made earlier than six years before it issued proceedings, that is earlier than 8 April 1997. For those claims, it must rely on the postponement of the limitation period under section 32(1)(c). As this court decided in FII SC 2, it can do so if it can show that it could not reasonably have discovered that its belief that the tax was due was mistaken until a date which was within the period of six years before the issue of the proceedings. To comply with that test, AXAIUK would need to show the date on which it could have realised that it had a worthwhile claim to challenge the legality of the tax regime. The six-year limitation period starts to run from that date. That date has not yet been determined so far as AXAIUK is concerned. Therefore, on the law as it now stands, the matter would need to be remitted to the High Court to determine the date, unless the parties could reach agreement on the date. Depending on the correct date, the claim in respect of corporation tax paid in the accounting period ended 31 December 1995 might be statute barred.[103]AXAIUK and GREA say, however, that they are entitled to orders that are based not on the current state of the law but on what the law was thought to be as at either 24 October 2013 or 28 January 2014, those being the respective dates of the judgment and the order of Henderson J in the Prudential test case. That is because their claims were on the GLO register for the CFC and Dividend GLO at those dates and so the decisions in the Prudential test case should be binding on the Revenue as far as disposing of their claims is concerned.[10]The further progress of the AXA proceedings[104]Following the final disposal of the Prudential test case, the stay imposed in 2003 on the AXA proceedings was lifted by order of Falk J on 18 November 2021. The parties then formulated their rival contentions as to the effect of the orders made in the Prudential test case on the issues in these proceedings.[105]The taxpayers’ submissions as to how the issues should be resolved followed from their contention that they can rely on the determination of issues that were raised and decided in the Prudential test case chosen and pursued in the CFC and Dividend GLO of which they formed a part. If that is correct, then the order they are entitled to is based on the law as at October 2013 or January 2014:(a) As regards GREA’s claim, that would mean that (i) they can recover on the basis of a common law claim in restitution for money paid under a mistake of law; (ii) they have a claim which entitles them to a remedy in respect of the prematurity period and (iii) the proper quantification of that remedy is the award of a sum equivalent to the interest that the principal sum would have earned during the prematurity period. (As earlier mentioned, GREA claims simple interest, although an unqualified application of the decision in Prudential HC would entitle it to compound interest).(b) As regards the limitation period raised by GREA and AXAIUK, they can (i) assert that they both fall within section 32(1)(c); (ii) postpone the start of the six year limitation period until the date on which they could with reasonable diligence have discovered their mistake; (iii) identify that date by applying the test set out in DMG namely the date of the final judicial judgment which established the illegality on which they rely. That date is, they assert, either 12 December 2006 (the date of the judgment of the CJEU in FII CJEU 1) or alternatively some other date but a date no earlier than 14 July 1998. The significance of that date is that it is six years before the date on which the first ACT claim was made in the Prudential test case and the date referred to by Henderson J in Prudential HC (see para 75 above). If either of those is the earliest date on which they could with reasonable diligence have discovered their mistake, then none of AXAIUK’s and GREA’s claims is statute barred and there is nothing more they need to prove to show that their claims are in time.[106]In the High Court, Richards J held that neither the Set Off Issue nor the Limitation Issue had been decided in the Prudential test case as GLO issues. He therefore decided them in accordance with the law as it now stands, following the decisions of this court since Prudential HC, as discussed above.[107]On appeal by AXAIUK and GREA, the Court of Appeal upheld the decision that the Limitation Issue had not been decided as a GLO issue and should therefore be decided in accordance with the current law. However, it held that the Set Off Issue had been decided as a GLO issue in Prudential HC and went on to consider whether it should exercise the power under CPR r 19.23 to “order otherwise”, so that Henderson J’s decision on that issue was not binding in the AXA proceedings.[108]The judgment of the court (Lewison, Nugee and Falk LJJ) noted that exercise of the power must take into account its context and purpose (para 52) and that, having regard to the important and useful benefits of the GLO regime, and the principle of finality and the need for efficiency and economy in the conduct of litigation, there needs to be “a strong reason to depart from that important general rule” set out in CPR r 19.23(1)(a) (para 53). It would be quite wrong to order otherwise merely because a dissatisfied litigant in a different claim on the group register may have better or different arguments that could lead to a different result on the law (para 57).[109]The court found that the principles of res judicata, particularly as applied to issue estoppel, a helpful guide (paras 58–67), referring to the decision of the House of Lords in Arnold v National Westminster Bank plc [1991] 2 AC 93 (“Arnold”) and the decision of this court in Virgin Atlantic Airways Ltd v Zodiac Seats UK Ltd [2013] UKSC 46; [2014] AC 160 (“Virgin Atlantic”). The court held that “the particular, and highly unusual, facts of this case make it appropriate to ‘order otherwise’ because, by analogy with the principles applying to issue estoppel, there are special circumstances which would otherwise create injustice” (para 69).[110]The court gave four reasons for this conclusion. First, the Revenue could not fairly be criticised for not seeking to overturn Sempra Metals in the Prudential test case, given the state of the law at that time (paras 70–72). Secondly, the effect of FII SC 3 is that parties to the FII GLO and claimants in other group actions must rely on section 85 of the Finance Act 2019 for an award of interest in respect of the prematurity period. Building on the decision in Prudential SC that Sempra Metals did not represent the law, this court held in FII SC 3 that interest was not recoverable at common law in respect of the prematurity period. Unless the power to order otherwise was exercised in the present case, claimants on the CFC and Dividend GLO group register “would benefit from a decision that is now known to be wrong in law, and further in circumstances where that has been established by the test case itself” (paras 73–76). Thirdly, the GREA claim has yet to be determined. Unless the court exercised its power to order otherwise, the court trying GREA’s claim:
“77. …would be required to decide the Set Off issue in a way that is known to be wrong. This is very different in nature to seeking to reopen an existing decision made on the basis of an incorrect understanding of the law. The court would be required to act in contravention of the basic principle that it should decide a case in accordance with its understanding of the law.”
Fourthly, unless the court ordered otherwise, category (a) claims would be decided “on a basis that is not only known to be wrong in law but is also logically inconsistent with the basis on which they are awarded interest under categories (b) and (c)” (para 81).[11]The discretion to order otherwise: in general[111]The power conferred by CPR r 19.23(1)(a) to order that the decision in the test case is not binding on follower claims is expressed in general terms. The drafter of the rule has not specified or limited the circumstances in which it may be exercised.[112]Mr Bremner KC, acting for the appellants, argued for a very narrow application of the “orders otherwise” discretion. He argued first that CPR r 19.23(1)(a) should be read as if it said that the judgment or order in the test case is binding unless that judgment or order itself provides that it is not to bind some or all of the follower claims. As Henderson J’s judgment in Prudential HC and the order made giving effect to it contained no such limitation, Mr Bremner submitted that there is no discretion in the court now to disapply the decision on the GLO issues determined. We do not accept that reading of the rule. If that had been the drafter’s intention, it would have been made clear by different wording. We consider that the court does have power to make an order in follower claim proceedings, after the test case has been disposed of, identifying any follower cases in respect of which the test judgment or order should not be binding as regards any GLO issue.[113]Accepting that there must be some circumstances in which it is contemplated the power is used, Mr Bremner gave four examples of when it might be appropriate to order otherwise. First, the court could order in advance that the decision in a test case would not be binding on certain sub-groups because of differences between their situation and that of the test claimant. Second, it may become apparent during or after the hearing of the chosen case that it is unsuitable to be a test case. Third, there might be some procedural unfairness or error in the test case which made it unfair for its decision to bind follower claims. Fourthly, the ruling in the test case might be overbroad in its wording so that it should not be applied in its full force or to all follower claims.[114]We consider, however, that these examples, other than the fourth, would seem to follow on from a realisation, either at the time of designation or later, that the chosen case was not in fact suitable to stand as a test case for all or some of the follower claims. We do not accept that the discretion is limited to that kind of circumstance.[115]When considering what it is helpful to say in this judgment as to the circumstances in which the discretion should be exercised, we note that the kinds of disputes in which the GLO procedure is being used and the nature of the GLO issues identified are very varied. A brief examination of the list of group litigation orders published on HM Courts & Tribunals Service website clearly shows it is being adopted in a wide range of legal claims, from claims arising from maltreatment of pupils at a residential school to claims in private nuisance against the owners of a meat processing plant. It would be rash therefore to try to provide guidance for every GLO set of claims.[116]The Court of Appeal considered that it was appropriate to order otherwise in the present case because not to do so would “create injustice” (see para 69 of its judgment). It is true that the avoidance of injustice is the principle which must guide the court in the exercise of its power to order otherwise, as it guides the exercise of any discretion. That does not mean, however, that the rule confers a discretion with a wide ambit on the judge exercising it, as we make clear below. Although it is not appropriate in this judgment to attempt to give an exhaustive description of the circumstances in which a court may decide to order otherwise, it is clear that having regard to the importance of not undermining the efficacy of the GLO regime, there will need to be some exceptional quality about the circumstances before it becomes appropriate for the court to consider whether to exercise the power. We focus here on the exercise of the discretion in a case, such as the present appeal, where one party challenges the binding nature of the test case judgment because the legal basis of the decision in that judgment has been shown to be wrong.[117]Issue estoppel provides another circumstance in which the law imposes a general rule that a party may be precluded from putting forward potentially strong arguments in support of its case because of the binding nature of an earlier order. But the doctrine also contemplates that there may be an exception to that rule so that the court may take account of a change in the law in deciding, when making a new order, whether a party remains bound by a previous order. The different kinds of estoppel that can arise in proceedings were fully explored by this court in Virgin Atlantic.Lord Sumption (with whom Lady Hale, Lord Clarke and Lord Carnwath agreed) described issue estoppel at para 17 as “the principle that even where the cause of action is not the same in the later action as it was in the earlier one, some issue which is necessarily common to both was decided on the earlier occasion and is binding on the parties”.[118]There is, of course, strictly no estoppel of any kind between the claimants in the AXA proceedings and the Revenue because there has been no determination of any issue in that claim, only in the proceedings between Prudential and the Revenue. However, one can see by an albeit imperfect analogy that the aim of the GLO is to prevent follower claims from seeking, as Lord Sumption put it in Virgin Atlantic “to reargue in materially altered circumstances an old point which had previously been rejected”: para 21. Whereas the bar on cause of action estoppel is, Lord Sumption said, absolute in relation to all points which had to be and were decided in order to establish the existence or non-existence of a cause of action, issue estoppel presents a bar “except in special circumstances where this would cause injustice”.[119]CPR r 19.23(3) clearly recognises the possibility of an exception to the binding force of the test case order, so we consider that the cases on issue estoppel are helpful in indicating what kinds of exceptional circumstances can justify a departure from the rule. The leading case on when exceptional circumstances justify lifting an issue estoppel is Arnold. The question in that case was whether when operating a rent review clause under a lease, the tenants were bound by the construction given to the very same clause by Walton J in earlier litigation between the same parties during the previous rent review. The Court of Appeal had subsequently, in other cases, cast doubt on Walton J’s construction. In his analysis of Arnold, Lord Sumption in Virgin Atlantic stressed that the critical point was that it was not a case where the tenant had failed to bring his whole case forward before Walton J. The tenant had argued the very point he now wished to reopen. The question was whether he should be allowed to reargue it in materially altered circumstances, in particular where the material alteration was that it had been shown subsequently that Walton J had wrongly applied the law. Counsel for the landlord submitted that it had hitherto been accepted that it was never an answer to a plea of estoppel to say simply that the earlier decision was wrong: “a later decision demonstrating beyond doubt that the former decision was wrong does not overcome the absolutely binding nature of the estoppel.” (see p 99A of the report).[120]Their Lordships disagreed. Lord Keith said at p 103 that there appeared to be “powerful grounds for the view” that Walton J’s construction of the rent review clause was wrong. He said (p 109B) that “One of the purposes of estoppel being to work justice between the parties, it is open to courts to recognise that in special circumstances inflexible application of it may have the opposite result”. The further relevant material that a party might be allowed to bring forward in the later litigation could be a change in the law. At p 110, Lord Keith approved the statement of Sir Nicolas Browne-Wilkinson V-C who had decided the matter at first instance: [1989] Ch 63, 70–71:
“In my judgment a change in the law subsequent to the first decision is capable of bringing the case within the exception to issue estoppel. If, as I think, the yardstick of whether issue estoppel should be held to apply is the justice to the parties, injustice can flow as much from a subsequent change in the law as from the subsequent discovery of new facts. In both cases the injustice lies in a successful party to the first action being held to have rights which in fact he does not possess. I can therefore see no reason for holding that a subsequent change in the law can never be sufficient to bring the case within the exception. Whether or not such a change does or does not bring the case within the exception must depend on the exact circumstances of each case.”
[121]One must recognise at once that this case is not on all fours with the Arnold case. In that case the parties were in a continuing relationship and the decision of Walton J would have regulated four further rent reviews until the end of the term. Further, any potential appeal by the tenant had been blocked by Walton J himself when he refused to certify the matter as fit for appeal; something that he had clearly been wrong to refuse. Mr Bremner submitted that Arnold provided no assistance in the present case because the public interest engaged in issue estoppel is different from the public interest engaged in the operation of the GLO system. While we accept that the public interests are not the same, we do not think that the public interests involved in the GLO system are such as to exclude or reduce in importance the considerations of justice to which Sir Nicolas Browne-Wilkinson and Lord Keith referred.[122]Arnold therefore stands as authority for the proposition that the underlying principle is to work justice between the parties but also that allowing a litigant to reopen a point decided against them is likely to be just only in exceptional circumstances. That is how it was regarded by the court in Virgin Atlantic although Lord Neuberger sounded a useful note of caution. He said at para 62 of Virgin Atlantic that it may be dangerous simply to invoke the observation of Lord Keith in Arnold that estoppel is intended to work justice between the parties. It is only too easy, Lord Neuberger said, “to fall back on it as an excuse for an unprincipled departure from, or an unprincipled exception to, the rule”. However, he added that “in a case where the rule has been relied on, I consider that it is helpful for a court which is inclined to accept the argument that it does not prevent a point being taken, to consider whether that outcome would work justice between the parties”.[123]We said earlier that the analogy between the power to order otherwise and the Arnold exception to issue estoppel is imperfect. That is because issue estoppel does not generally raise the additional weighty factor that making an exception in any case might have wider and potentially undesirable repercussions for the operation of the overall GLO regime. There are therefore two conflicting key considerations of great importance, namely the need to maintain the integrity of the GLO regime by treating the test case as binding on follower claims on the one hand and on the other hand a party’s right to have their dispute decided by the application of a correct understanding of the law as it stands at the time that the order is made.[124]The first key consideration directs the court to approach the exercise of the discretion with great caution so as not to jeopardise a central purpose of a GLO which is to enable multi-party actions to be determined efficiently on a collective basis. Test cases on issues of law common to the class of claims subject to a GLO are a central feature of the GLO system. It is intended to provide a mechanism giving finality and consistency in the determination of all the cases on the register. The re-litigation of a GLO issue decided in a test case gives rise to the risk of inconsistent decisions which it is the purpose of the GLO process to prevent. It risks a follower claimant being placed in a better or worse position than the claimant in the test case and may affect the willingness of claimants on the GLO register to agree to another claim going forward as the test case.[125]But the counterweight to that in a case such as the present is that it will ordinarily be unjust for a person to have their rights determined otherwise than in accordance with the right legal rules. In general, it would be contrary to the rule of law for a court to make an order imposing an obligation on a party to proceedings when it is clear that, as at the date that the order is made, there is no legal basis for imposing such an obligation. The consequences of such an order are not just that the defendant is liable to pay the judgment amount but, if it is unable to do so, it may be wound up as an insolvent company or, if an individual, made bankrupt. Not only is there no legal basis for such an order at the time it is made, other than the overruled test case decision and the effect of CPR r 19.23(1)(a), but, because of the declaratory effect of judgments in a common law system, no legal basis existed for the decision in the test case at the time it was made.[126]Two points follow from the need to balance those two key considerations. First, a challenge to the binding effect of the GLO test case judgment can only be mounted where it has been clearly established that the legal basis for that judgment was wrong. It is not enough that the follower claimant or the unsuccessful defendant in the test case wishes to raise different—potentially better—arguments on the law than were raised and decided in the test case. Further, it is not enough for the challenger to be able to point to dicta in subsequent cases that cast doubt on the correctness of the test case judgment. In this way the exceptionality required is stronger in this context than it is in issue estoppel. It was not entirely clear in Arnold that Walton J’s decision was wrong, but the issue estoppel was still lifted. In the present context, in our judgment, that would not be enough. It will be a rare case where, between the date when the test case is finally decided and the date on which the follower claims come to be disposed of, there has been a reversal of the law so as to show that the test case was wrong. But it is right that the discretion should arise only in such a rare case.[127]The second point that follows on from the balance of those key considerations is that GLO decisions set out in the judgment or order are binding on the follower claims whether they are decided after full argument or by way of concession or admission on the part of the party which later asks the court to order otherwise. The extended rights of appeal conferred on GLO register parties by CPR r 19.23(2) remove any unfairness that might thereby be created.[128]Once the challenger has established that the legal basis for the test case has been shown to be wrong, the countervailing need to apply the correct law in determining the follower claims comes into operation. That is not, however, the end of the exercise of the discretion because the court will need to consider whether, in the particular circumstances of the GLO proceedings in question, it will still be unjust to order otherwise. Factors that should be addressed include the following.[129]An important factor is whether the result of disapplying the test case judgment will be to impose on the party relying on the test case judgment further lengthy and costly proceedings, in effect relitigating the issue that has been decided in their favour by the test case. It may be that there will need to be further stages in the follower claims in any event, for example where the GLO issues are limited to liability and each follower claim will need to be separately resolved as regards quantum. But if an order that the test case is not binding will then lead to substantial additional work for the parties and for the court, that may point towards refusing to order otherwise.[130]We agree with the observation of this court in FII SC 2 when considering whether the Revenue should be allowed to challenge the earlier judgment in DMG. In words that apply equally to this judgment they said:
“93. … It is important that there be discipline in the conduct of actions which are the subject of group litigation orders and it is important that there be finality in the determination of issues raised in such actions. An appellate court, in the interests of justice, will normally seek strenuously to avoid an outcome which results in the parties, who have already gone to trial on the quantification of a claim, having to amend their pleadings and to adduce further evidence to apply its ruling on a new issue of law to the facts of their case. In a normal litigation, the need for a retrial would be a strong and normally determinative pointer against allowing a party to withdraw a concession which had influenced the way in which a litigation had been conducted.”
[131]Another factor is whether there is any particular unfairness to follower claimants arising from the fact that they were not chosen to be the test case in this particular GLO. Their own proceedings will have been stayed, perhaps for many years, and they will have had limited control over the progress of the test case. In some circumstances it will be unfair to deprive them of the benefit of a judgment in favour of the test case claimant on a GLO issue if they may be prejudiced by the lapse of time that has occurred and by the fact that they may have assumed that they did not have to progress their own claim in tandem with the test case. We note the comments of Lord Woolf in Boake Allen to which we referred earlier. He said (paras 32–33) that to achieve the GLO regime’s objective of saving parties’ costs, parties to the follower claims are not required themselves to take all the steps normally needed in litigation but will leave the parties to the test case to take all the steps needed to make good the claim.[132]It is true, however, that being a follower claimant rather than a test case claimant can have distinct advantages. Even though the follower claimant may ultimately be called on to contribute to the legal costs, they are spared the substantial investment of management time and resource that must be made by the test claimant who must actively direct the proceedings, liaising with the court and engaging with the defendant. That is also a factor that may influence whether it is just or unjust to deprive the follower claimant of the benefit of the test case judgment.[133]It may also be relevant for the court to be alert to a party having attempted to manipulate proceedings by making tactical decisions in the course of the test case with a view to inviting the court at a later date not to treat the test case decision as binding. For example, it may be necessary, as it is in this case, to consider why a particular concession was made or a point not pursued on appeal in the test case when weighing up whether the challenger should now in effect be released from the consequences of that decision being effectively carried forward to the follower claims.[134]Having explained the competing interests at stake and the factors that the court should take into account, we turn to consider how to apply those in the present case.[12]The discretion to order otherwise: the Set Off Issue in this appeal[135]Prudential was expressly designated as a test case to resolve common issues shared by those on the group register for the CFC and Dividend GLO, including the Set Off Issue. The Revenue accept that the Set Off Issue was an issue common to those claims and they did not challenge the designation of Prudential as a test case. As the Revenue further accept, GREA’s claim is indistinguishable from that in Prudential on this point. The choice of Prudential as the test case directly affected all other follower claims, including that of GREA. There was some dispute in the courts below as to whether the judgments in Prudential CA or Prudential SC had undermined that conclusion. That suggestion was firmly rejected by Richards J and by the Court of Appeal in the judgment under appeal and is not pursued by the Revenue before us.[136]We are of the clear view, as was the Court of Appeal, that to enter judgment in a follower case on the basis of the judgment in Prudential HC, particularly in the very unusual circumstances that have arisen in this GLO, would be unjust to such an extent that it is appropriate to exercise the discretion to order otherwise.[137]The primary consideration as regards those unusual circumstances is the extraordinary nature of the developments of the law in the 20 years since these proceedings were started. This case does, therefore, meet the stringent criterion we described in the previous section. It has now been definitively decided by this court that the two planks on which the Prudential HC decision was based as regards the Set Off Issue were wrong.[138]First, for many years it was clear, as confirmed by the House of Lords in La Pintada,that there was no cause of action in English law for recovery of the loss of the use of money when the debt had been paid before the start of proceedings. In 2007 the House of Lords decided in Sempra Metals HL that such a cause of action did exist. But 11 years later in Prudential SC this court indicated that, but for the Revenue’s concession as regards the prematurity period and category (a), they would have held that there was no such claim. This was confirmed in 2021 in FII SC 3 where the court considered and rejected an argument that in Prudential SC the court had at least decided that there was some right to a remedy for the prematurity period which was more generous than that by then provided in section 85 of the Finance Act 2019. The court in FII SC 3 reiterated its conclusion that the right arising from the breach of EU law “is not itself a restitutionary claim for the recovery of money paid under a mistake of law”: para 105. Because no such right existed—or had ever existed—the court rejected the contention that section 85 had retrospectively deprived the claimants of a right.[139]Secondly, as regards the availability of compound interest, we now know that it has in fact never been the law that the EU principle of effectiveness required a member state to award compound interest when unlawfully levied tax is repaid by the taxing authority to the taxpayer. The CJEU has never held that there is such a requirement. The House of Lords in Sempra Metals HL did not need to decide whether there was such a requirement because they held that compound interest was available as a matter of English law. The court rejected that analysis of the position in Littlewoods SC in November 2017.[140]This court has already remarked on the scale and speed of legal developments arising from the several GLOs challenging different aspects of the tax regime. The combined litigation has been described as being of “exceptional complexity and novelty” raising “legal issues of unparalleled complexity” which were at “the frontier of legal developments”: see FII SC 2 paras 18 and 78. Similarly in FII SC 3 the court said:
“78. … The FII GLO litigation and the related GLO litigations proceeded against a background in which both domestic and EU law were in a state of significant development and interacted with each other in this GLO litigation. Henderson J in FII HC 2 [2015] STC 1471, para 468 correctly spoke of ‘a complex and evolving legal landscape’. The three judgments of the CJEU on references in the FII GLO litigation in 2006, 2012 and 2013 together with judgments on references in other relevant proceedings, and the now three appeals to this court in the FII GLO litigation as well as the appeals to the House of Lords in Sempra Metals and to this court in Littlewoods and Prudential, are testimony to the evolving nature of that landscape. Issues which affect the FII GLO litigation have been decided in the other legal proceedings such as Littlewoods and the portfolio dividends GLO (including in Prudential) and vice versa. Against that background, it is unsurprising that questions that are of central importance to the claims in the FII GLO litigation have only recently been decided or are yet to be decided.”
[141]We turn then to consider whether there are factors which outweigh the injustice that would be caused to the Revenue if the follower claims are decided on an incorrect legal basis. In our view, there is nothing arising from the conduct of these particular GLO proceedings which militates against applying a correct understanding of the law to dispose of the follower claims.[142]First, the exercise of the power in this case does not have the result of condemning the parties to yet more years of litigation to resolve factual or legal points. In so far as there remain any factual issues concerning the amounts and dates of tax payments by GREA, those would not have been resolved anyway by the Prudential test case. We understand that the computation of the section 85 remedy, complicated though it no doubt is, is not made more difficult because the order on the prematurity period in Prudential HC does not apply.[143]Mr Bremner’s core submission that the power to order otherwise should not be exercised to allow points of law to be re-argued does not, therefore, arise in this case. There is no question of a second bite at the cherry. The legal position is established by Supreme Court authority and there will be no further argument as a result of the Court of Appeal’s order. Mr Ewart KC accepted that it would not be an appropriate use of the power to enable renewed argument on a common issue decided by the test case. In this respect, this is a stronger case than Arnold. In that case, although Lord Keith said that the earlier decision of Walton J was plainly wrong, the issue of construction of the lease remained to be argued and decided.[144]Secondly, as regards the manner in which the test case proceedings were conducted, Mr Bremner argues that the Revenue could have challenged the existence of the restitutionary cause of action for the prematurity period by appealing the judgment in Prudential HC in respect of category (a). They decided not to do so for their own tactical reasons. A similar submission was addressed by this court in FII SC 3. The taxpayers argued there that it would be an abuse of process for the Revenue to deny that the claimants had a common law cause of action for the prematurity period because they had not challenged the correctness of Sempra Metals HL earlier in the FII proceedings. The court in FII SC 3 recognised that the impact on the claims in the FII GLO may be significant since the claims for the period of prematurity were a major portion of the claimants’ claims and represented the entire claim for some of the claimants. But the court rejected that and other points relied on by the claimants, holding that the Revenue’s conduct was not abusive and that the effect on the claim of the new challenge did not provide a good ground to exempt the claimants from the application of the law as it stood in the light of recent developments of the law of unjust enrichment: para 82.[145]The exercise of the discretion to order otherwise under CPR r 19.23(1)(a) is not the same as the decision whether a party is abusing the process of the court by seeking to raise a particular legal argument. But the factors discussed by the court in FII SC 3 are relevant to the consideration of what justice requires in the present appeal. It would be a serious injustice for the Revenue if they now have to meet claims valued at many millions of pounds when there is no legal basis for them.[146]Further, as regards suggested unfairness, Mr Bremner argued that it is unfair that Prudential obtains the benefit of the decision in Prudential HC but the claimants in the follower claims do not. Fairness requires a consistent outcome for all these cases. We do not accept that this consideration outweighs the injustice of a judgment based on law known to be wrong. The existence of the exception provided in the rule necessarily envisages that the result of the proceedings for the test case and the follower claims will be different—that is an inevitable result of the inclusion of the power to order otherwise. That factor cannot, without more, be enough to preclude the exercise of the discretion.[147]In any event, there would not be, or may well not be, consistency between all claims on the register. A judgment on a common issue in a test case is not binding on parties to claims subsequently added to the register except to the extent that the court may so order under CPR r 19.23(1)(b). If it does so order, any party to a claim later added to the register may not appeal the judgment but may apply for an order that the judgment is not binding on that party. The situation could well arise that a decision in a test case which was later overruled would not apply to claims added subsequently to the register.[148]A further point on unfairness on which Mr Bremner relied was that the claimants in the follower cases were required, in accordance with usual practice, to contribute to Prudential’s costs. As we have mentioned, schedule 4 to the GLO made by Master Weingarten in March 2003 (as amended) provided that all claimants falling within one of seven classes were severally liable to the claimant in the test claims for those classes for an equal proportion of the common costs incurred in relation to that issue by that test claimant. Common costs were defined as including the costs reasonably incurred by the test claimant in relation to its own claim and also(a) any liability on the part of the test claimant to pay the defendant’s costs incurred in relation to the test claim,(b) the costs incurred by the solicitors acting for the test claimant in administering the group litigation and(c) costs incurred in ascertaining, calculating, allocating and collecting the costs from each claimant including the costs of chasing for payment and processing the payments when received. Further, all claimants are liable for the lead solicitors’ costs (the lead solicitor may be different from the solicitors acting for the test claimants) in administering the group litigation.[149]We do not consider that this question of costs, whether alone or with other factors, outweighs the injustice of imposing a substantial liability on the Revenue on a basis which is wrong in law, all the more so as the liability greatly exceeds the amount of costs in issue. It should be noted that AXA will have recovered its share of Prudential’s costs paid by the Revenue on the standard basis.[150]For these reasons, we agree with the Court of Appeal that this is an appropriate case in which to order under CPR r 19.23(1)(a) that the decision in Prudential HC on the Set Off Issue is not binding in the claim made by GREA against the Revenue.[13]The Limitation Issue[151]The significance of the debate as to the nature of the claim for the prematurity period was not only the quantification of the loss but also the application of section 32(1)(c). Given our decision on the prematurity period, this point now is only relevant for AXAIUK’s claim. GREA’s claim in respect of the utilised ACT does not give rise to a cause of action for mistake of law, so section 32(1)(c) is not relevant. This point is therefore now only relevant for the portfolio corporation tax claim which it is common ground is a claim for relief from the consequences of a mistake of law.[152]We described earlier that if AXAIUK’s claim were being decided on the law as it applies today, it would be able to postpone the start of the limitation period by relying on section 32(1)(c) only until the date at which it could reasonably have realised that it was mistaken in thinking that it was under a legal liability to pay the ACT that the Revenue was demanding from it, rather than from the date of the judicial decision which authoritatively held that there was no such legal liability.[153]The question to which we now turn is what was decided about limitation in the High Court judgment in the Prudential test case? As earlier explained, it is AXA’s case that Henderson J decided as a GLO issue, so as to be binding on follower cases, that the limitation period did not begin to run until 12 December 2006, the date of the CJEU’s judgment in FII CJEU 1, or, alternatively, on any view not before 19 October 2003, six years before GREA made its ACT claims.[154]The Revenue’s case, which was accepted by Richards J at first instance and by the Court of Appeal, is that Henderson J made no decision on the start of the limitation period as a GLO issue. The judgments in Prudential CA (April 2016) and Prudential SC (July 2018) did not address the Limitation Issue substantively: see Prudential CA at para 158. We must therefore now step back into the Prudential test case to analyse what Henderson J decided in Prudential HC handed down on 24 October 2013.

(a) The limitation issues arising in the Prudential test case

[155]To recap, at the time of Henderson J’s judgment in the Prudential test case:(a) The governing authority on identifying the commencement of the limitation period under section 32(1)(c) was DMG decided in October 2006. Applying that test, it was generally accepted by the parties in the Prudential test case that the earliest date for constructive discovery of the illegality of the provisions at issue in the CFC and Dividend GLO was 12 December 2006, the date of the decision in FII CJEU 1.(b) Parliament had enacted section 320 of the Finance Act 2004 which retrospectively limited the effect of section 32(1)(c) in a way which would have barred some of the Prudential claims. The issue of the compatibility of section 320 with EU law was at that time pending before the CJEU having been referred to it by this court following its decision in FII SC 1 handed down on 23 May 2012. The Advocate General’s opinion had been delivered on 5 September 2013, advising that section 320 was invalid, but the CJEU had not yet handed down its judgment.(c) This court had held, also in FII SC 1, that section 107 of the Finance Act 2007 was invalid and so did not disapply section 32(1)(c).[156]The significant dates in the procedural history of the Prudential test case were set out by Henderson J in para 249 of his judgment in Prudential HC. There were two claimants party to the Prudential test case whose claims were particularly affected by the start date for the purposes of section 32(1)(c). They were the original claimant, the Prudential Assurance Co Ltd (“PAC”) which commenced its claim on 8 April 2003 and Prudential Holborn Life Ltd (“PHL”) which was added to the claim by amendment on 14 July 2004. Initially both PAC’s and PHL’s claims were made in relation to portfolio dividends and were claims (following various amendments) in respect of accounting periods going back to 1990. By amendment dated 19 October 2009, claims in respect of ACT were added by them both.[157]One of the important points considered by Henderson J was whether claims added by way of amendment were to be treated as having been made at the date of the amendment or whether they related back to issue of the proceedings. Relation back was particularly important at that stage of the Prudential litigation because the issue of the validity of section 320 was still undecided; section 320 disapplied section 32(1)(c) only in respect of proceedings commenced after 8 September 2003 so if claims added in by way of amendment related back to the start of PAC’s proceedings (April 2003) they would not be caught by section 320 even if it were valid.[158]Henderson J referred to the standard provision in the GLO that amendments took effect from when they were made, unless the court held that the amendment did not add a new claim or only added a claim that arose out of the same or substantially the same facts as already pleaded, in which case they related back to the start of the proceedings (para 251). Applying that principle, Henderson J held:(a) All PAC’s claims in relation to corporation tax on portfolio holdings related back to the start of its claim on 8 April 2003 even if they arose in accounting periods that were only added later by way of amendment. That meant that its claims in respect of accounting periods later than 8 April 1997 were not statute barred but its claims for earlier periods were barred unless it could rely on section 32(1)(c).(b) PAC’s claims for ACT added by amendment in October 2009 could relate back to its original corporation tax claim and so would be in time, regardless of section 32(1)(c), in so far as they extended back to 8 April 1997.(c) PHL’s claims to corporation tax on portfolio holdings related back only to the date when it was added to the claim on 14 July 2004 and not to the date when the proceedings were first started by PAC. That meant that its claims for periods after 14 July 1998 were not statute barred but claims for periods earlier than that were barred unless it could rely on section 32(1)(c).(d) PHL’s claims in relation to ACT also added by way of amendment related back to the amendment which had brought it into action (14 July 2004) and so all claims for ACT in respect of periods before 14 July 1998 could only be made if section 32(1)(c) applied.[159]The Revenue’s application for permission to appeal against the relation back of ACT claims to the start of PAC’s or PHL’s proceedings was refused by the Court of Appeal: see para 158 of Prudential CA. What Henderson J decided in addition to the relation back point is the issue raised in this appeal which we now address.

(b) The limitation issue for AXAIUK

[160]The timeline in the current proceedings is as follows. AXAIUK issued its claim on 8 April 2003. That, as we have said, relates only to the corporation tax it paid on dividends it received from its portfolio shareholdings, not to utilised ACT on dividends it paid to its shareholders. Its claims go back to the accounting period ending 31 December 1995, so it needs to rely on section 32(1)(c) for periods earlier than 8 April 1997.[161]On the limitation point, the appellants addressed the court as to three steps: first, was the date on which the limitation period started to run a GLO issue in the CFC and Dividend GLO? Second, if it was a GLO issue, what was decided by Henderson J on that issue in Prudential HC? Third, if the limitation period point was decided by Henderson J in Prudential HC on the basis of the law as it then stood (that is according to DMG) should the court order otherwise so as to prevent AXAIUK relying on that decision? (c). The CFC and Dividend GLO[162]Schedule 3 to the CFC and Dividend GLO made in agreed form in 2003 included among the issues for the test case:
“Limitation (A) Can a claimant’s claim properly be brought as a claim for restitution for mistake of law or must such a claim be brought only as: (a) a claim for damages; and/or (b) a claim for restitution in respect of payment made pursuant to an unlawful demand? (B) In either event, from what date does the applicable limitation period start to run?”
[163]Para (B) clearly identified the start date of the limitation period as an issue, which could encompass the start date applicable by virtue of section 32(1)(c).[164]In fact, the pleadings in the Prudential case did not raise any issue under section 32(1)(c). In their defence, the Revenue pleaded that the claims were largely statute barred. It was for the claimants to serve a reply pleading that section 32(1)(c) applied and the facts on which they relied. No reply was served.[165]In December 2012, Henderson J directed the parties to agree a list of issues to be decided at the trial of the Prudential case. As regards limitation, the agreed list included:
“VIII. Limitation 1. To what extent is the claim statute barred by a 6 year limitation period? 2. To what extent is the claim for recovery under a mistake of law barred by section 320 FA 2004.”
[166]The second issue raises only the issue of the validity under EU law of section 320 of the Finance Act 2004. The first issue does not raise any issue as to the date on which a limitation period would start under section 32(1) nor, as noted above, had any particular dates been pleaded by the claimants.

(d) The judgment in Prudential HC

[167]Henderson J did not refer to the original limitation issues stated in the GLO. On the assumption that section 320 was invalid, Henderson J said at para 255 that he did not understand the Revenue to argue that PHL could with reasonable diligence have discovered its mistake before 14 July 1998, so that it would be entitled to pursue its mistake-based claims for periods before that date. The next question was whether the ACT claims which were added by amendment could be related back to the dates of the respective claim forms. While he answered that question in the affirmative, he prefaced his analysis with this important observation at para 257:
“The question is probably academic, since it seems to me that the claimants would probably be able to rely on section 32(1)(c) on the ground that they could not have been aware of the invalidity of the ACT provisions before, at the earliest, the decision in [FII CJEU 1] in December 2006: compare FII (High Court [2009] STC 254 at para 267. In case it matters, however, I will briefly state my views on the question.”
[168]The order made by Henderson J in January 2014 included the following declaration: “11. Issues VIII. 1–2 are answered in light of [FII CJEU 3] as follows: The claims in mistake-based restitution (that is those successful claims listed in paragraph 8.A above) are not subject to the limitation period in section 320 of the Finance Act 2004 and are in time.” (e). Was the start date for limitation purposes decided in Prudential HC as a GLO issue?[169]In interpreting what was said by Henderson J in his judgment and the order, it is important to avoid hindsight and therefore to recognise that if DMG was the prevailing law, the date of reasonable discovery was not a fact specific issue but the same date for everyone. The only date that is expressly mentioned by Henderson J was that it was not earlier than 14 July 1998. That was a date specific to the Prudential proceedings because it was six years before PHL was added to the proceedings. It was not a date that was likely to be relevant for other proceedings on the register which were lodged at a different time.[170]The bald statement in the order of 28 January 2014 that the claims are “in time” mentions no date and must be interpreted by reference to what Henderson J said in his judgment. As para 257 of his judgment indicates, there appears to have been an underlying assumption, based no doubt on the law as stated in DMG, that for the purposes of section 32(1)(c), the Prudential claimants could not have been aware of their mistake in paying ACT before the decision in FII CJEU 1. But it is not an issue on which Henderson J was addressed or which he decided. At para 252 of his judgment, he said that two questions “were briefly argued before me”, neither of which was the start date for limitation purposes, and even those questions were not “argued in detail or at any length”. The discussion in his judgment as regards the ACT claims, which were added by amendment, is limited to questions relevant to whether they related back to the dates of the claim forms. We agree with the Court of Appeal’s view that the date of discoverability for the purposes of section 32(1)(c) “was simply not an issue” (para 104).[171]We stress that the fact that the answer to a GLO question may be agreed or admitted, rather than determined after adversarial argument, does not prevent it from being a binding decision on that GLO issue. A party cannot attempt to keep their powder dry by conceding something which is then decided against them in the test case in the hope of arguing it out in full in a different case on the register. But that is not what happened here.[172]The law that applies to AXAIUK’s claim is the law established in FII SC 2. It may be that given the result in the BAT Industries case the parties will be able to agree on a start date.

Conclusion

[173]For the reasons given in this judgment, we would dismiss the appeals on both the Set Off Issue and the Limitation Issue.[1]Introduction[174]Although I agree with Lady Rose and Lord Richards that the appeal should be dismissed, I wish to explain my reasons for affirming that, in a case of this kind, it will necessarily be rare that a court may properly “order otherwise” under CPR r 19.23(1)(a). By a case of this kind, I mean a case in which:(i) a court has finally decided in a judgment given in a test claim in group litigation a GLO issue which is an issue of law;(ii) an order is sought that the judgment does not bind the parties to another claim that was on the group register when the judgment was given; and(iii) the ground on which such an order otherwise is sought is that there is said to have been a subsequent development in the law which shows that the test claim was wrongly decided.[175]The power to order otherwise under CPR r 19.23(1)(a) must, like any power given to the court by the Civil Procedure Rules, be exercised in accordance with the overriding objective of dealing with cases justly and at proportionate cost. But, as Lady Rose and Lord Richards make clear at para 116 above, that does not mean that the rule confers a discretion with a wide ambit. The principles governing its exercise must be deduced from the scheme, context and purpose of CPR r 19.23 and the wider regime applicable to group litigation. I therefore think it important to situate the rule within that larger landscape.[2]Group litigation[176]Multi-party claims pose particular challenges for any legal system. In a complex society in which the activities and decisions of corporations and government bodies often affect large numbers of people, a single event or course of conduct can cause widespread harm and give rise to a multitude of claims for redress. To require each such claim to be litigated and decided separately would, at best, be hopelessly inefficient and, at worst, completely impractical. It would also risk inconsistent outcomes, which is itself an injustice. To avoid these mischiefs, procedures are needed to enable claims raising the same or similar issues to be dealt with collectively.[177]Three such procedures are currently available in England and Wales. Each has advantages but also significant limitations. The most longstanding is the representative action, which allows one or more persons to sue or be sued as representatives of others. Such an action can provide an effective means of collective redress, but its availability is limited by a requirement that the representative(s) and all those represented must have the “same interest” and by its inability to deal with any individual differences between claims: see Lloyd v Google LLC [2021] UKSC 50; [2022] AC 1217. A second and potent procedure available since 2015 is to bring “collective proceedings”, a form of class action. But this procedure is limited to proceedings before the Competition Appeal Tribunal to which section 47A of the Competition Act 1998 applies. The third and currently the most common form of multi-party procedure is “group litigation”. That is the procedure with which this appeal is concerned.[178]Group actions are, in essence, no more than groups of individual claims which are dealt with together. Many of the techniques used in managing group litigation were developed by judges on an ad hoc basis using their general powers of case management before any formal regime was established. Those techniques include: the appointment of a single designated judge to manage all claims of a specified description; maintaining a register of claims included in the group; setting cut-off dates for joining the group; appointing lead solicitors; identifying common issues suitable to be decided as preliminary issues; selecting one or more claims to proceed as “test” or “lead” cases while other claims are stayed; and making orders for the sharing of costs among claimants or defendants.[3]The limited effect of judgments in individual cases[179]The fact that, in group litigation, the claim of each member of the group is formally separate imposes a significant constraint on the provision of collective redress. Notably, it limits the court’s ability to decide common issues in a way that binds the parties to all the claims in which those issues arise. There can be hundreds or thousands of such claims. Unless a judgment on a common issue is capable of binding all the parties affected by the issue, the issue can be re-litigated, in principle many times over, thus bringing about the twin mischiefs of inefficiency/waste of resources and potential inconsistency of outcome.[180]English law has several legal rules designed to serve the strong public and private interest in the finality of litigation by preventing a matter from being re-litigated once it has been the subject of a final judgment. The two main rules go by the opaque names of “cause of action estoppel” and “issue estoppel”. These rules preclude the parties to a claim from reopening in any other proceedings between them either the actual subject matter of the claim (cause of action estoppel) or the determination of an issue that was necessary to the resolution of the claim (issue estoppel).[181]These rules, however, apply only to the parties to the claim and their “privies” (that is, persons whose claim or liability arises through that of the party). They have no effect in relation to another claim involving another party, even if the issues are identical. Indeed, on an issue of fact, a judgment is not even admissible in another claim: see Rogers v Hoyle [2014] EWCA Civ 257; [2015] QB 265. As Lord Neuberger observed in Virgin Atlantic Airways Ltd v Zodiac Seats UK Ltd (formerly Contour Aerospace Ltd) [2013] UKSC 46; [2014] AC 160, para 46:
“It is well established that the fact that an identical issue is determined differently in two different sets of proceedings is irrelevant to the rights of the parties to each set of proceedings inter se.”
Each claim in group litigation counts as a different set of proceedings for this purpose.[182]In managing group litigation, it is sometimes feasible to try issues as preliminary issues simultaneously in every claim in which the issues arise. But often such a procedure would be cumbersome or infeasible. It is therefore usual to select one or more individual claims as test claims and to use these as a vehicle for deciding common issues. The purpose of doing so is liable to be defeated if a judgment given on such an issue in a test claim does not bind parties to other claims which raise the same issue.[183]Before legislation was introduced to address this problem, the only procedural means of preventing the re-litigation of issues decided in a test claim was to rely on the court’s power to prevent an abuse of its process. That power was invoked in Ashmore v British Coal Corpn [1990] 2 QB 338. Some 1,500 claims for equal pay were brought by female canteen workers employed by British Coal. Test claims were selected. After a hearing involving extensive evidence, all 14 test claims were dismissed. One of the other claimants then sought to proceed with her claim. The Court of Appeal upheld the tribunal’s decision to strike out the claim on the basis that to allow re-litigation of the same issues would defeat the purpose of the procedure adopted and amount to an abuse of process.[184]The power to prevent an abuse of process is broad and flexible, but it is not a panacea. The burden is always on the party seeking to rely on the doctrine to establish that it is oppressive or a misuse of the court’s process to allow a claim to proceed (or a defence to be raised). That is a high bar. Further, in deciding whether to exercise the power the court is required to make “a broad, merits-based judgment which takes account of the public and private interests involved and also takes account of all the facts of the case”: see Johnson v Gore Wood & Co (a firm) [2002] 2 AC 1, 31 (Lord Bingham of Cornhill). If group litigation is to provide a reliable method of resolving multi-party claims effectively, a harder edged rule is needed.[4]The group litigation rules[185]Such a rule is one of the main benefits of the special regime for group litigation introduced by the Civil Procedure (Amendment) Rules 2000 (SI 2000/221) on the recommendation of Lord Woolf’s Access to Justice Final Report (July 1996), ch 17. The group litigation rules are contained in section III of CPR Part 19 and are now numbered 19.21 to 19.26.[186]CPR r 19.21 defines a Group Litigation Order (“GLO”) as an order made to provide for the case management of “claims which give rise to common or related issues of fact or law (the ‘GLO issues’)”. CPR r 19.22(1) gives the court power to make a GLO where there are or are likely to be a number of claims giving rise to such issues. The claims to be managed as a group under the GLO are identified by specifying the GLO issues: see CPR r 19.22(2)(b).[187]As noted in Class Actions in England and Wales (ed Damian Grave, Maura McIntosh and Gregg Rowan), 2nd ed (2022), para 5–061, the fact that decisions on the GLO issues automatically bind all claims on the register is perceived as one of the key advantages of using a GLO to resolve multiple claims. This effect is achieved by CPR r 19.23. That rule provides (with emphasis added):
“Effect of the GLO 19.23 (1) Where a judgment or order is given or made in a claim on the group register in relation to one or more GLO issues— (a) that judgment or order is binding on the parties to all other claims that are on the group register at the time the judgment is given or the order is made unless the court orders otherwise; and (b) the court may give directions as to the extent to which that judgment or order is binding on the parties to any claim which is subsequently entered on the group register. (2) Unless paragraph (3) applies, any party who is adversely affected by a judgment or order which is binding on them may seek permission to appeal the order. (3) A party to a claim which was entered on the group register after a judgment or order which is binding on them was given or made may not — (a) apply for the judgment or order to be set aside, varied or stayed; or (b) appeal the judgment or order, but may apply to the court for an order that the judgment or order is not binding on them …”
[188]The critical provision is CPR r 19.23(1)(a), which establishes the rule that a judgment or order made in a claim in relation to a GLO issue is binding on the parties to all other claims on the group register at the time the judgment is given, unless the court orders otherwise. (The term “judgment” in this context refers to a type of order, typically an order embodying the final decision of the court in a claim; but for convenience I will use the term “judgment” as a shorthand for the phrase “judgment or order”.) Before focusing on the critical provision, two broader features of CPR r 19.23 are worth noting.[189]First, the rule that a judgment or order is binding on the parties to other claims in the group needs to be viewed together with the right conferred on those parties by CPR r 19.23(2) to seek permission to appeal. This is an important quid pro quo. Just as a judgment does not ordinarily bind a non-party, so too a non-party ordinarily has no ability to appeal from it. A rule which makes a judgment binding on parties to other claims could be unfair if those parties had no opportunity to appeal from a judgment which adversely affects them. But, conversely, the ability to seek permission to appeal limits the scope for arguing that a party to another claim ought in justice not to be bound by what the judgment decided.[190]On the face of it, where a party adversely affected by a judgment has the opportunity to appeal, that is the proper way to challenge the correctness of the decision. Once that possibility has been exhausted, there is no justification for allowing the decision to be reopened. There are exceptions: for example, where evidence emerges that the judgment was obtained by fraud. But the purpose of the rules limiting the time for bringing an appeal is to ensure that litigation is brought to a definitive end and that parties who have received a judgment in their favour are not left in a state of uncertainty about its value. In Smith v Brough [2005] EWCA Civ 261; [2006] CP Rep 17, para 54, Brooke LJ stressed three reasons for declining to grant a long extension of time for appealing:
“(1) that it is a fundamental principle of our common law that the outcome of litigation should be final; (2) that the law exceptionally allows appeals out of time; (3) that this, and … [the doctrine that a judgment may be impugned for fraud], are the exception to a general rule of high public importance and reserved for rare and limited cases where the facts justifying the exception can be strictly proved.”
These reasons for not allowing a decision to be challenged after the time for an appeal has passed apply just as strongly whether the party which had the opportunity to appeal is a party to the claim or another party which is bound by the judgment pursuant to CPR r 19.23(1)(a).[191]The second point worth noting is the distinction drawn in CPR r 19.23 between claims that are on the group register at the time the judgment is given and claims which are subsequently entered on the group register. A party to a claim which is entered on the group register after a judgment on a GLO issue has been given cannot appeal the judgment: see CPR r 19.23(3)(b). Equally, there is no presumption that the judgment will be binding on the parties to such a claim. The court may give directions as to the extent to which the judgment is binding on those parties: see CPR r 19.23(1)(b). If the court directs that the judgment is binding on the parties to such a claim, a party may apply to the court for an order to the contrary: see CPR r 19.23(3).[192]It is to be expected that the power under CPR r 19.23(3) to order that the judgment is not binding will be exercised more readily than the power to make such an order under CPR r 19.23(1)(a). That is because the party applying for an order under CPR r 19.23(3) will not have had an opportunity to appeal the judgment—unlike a party who seeks an order otherwise under CPR r 19.23(1)(a).[5]No time limit for an order otherwise[193]Against that background, I come to the question of when the court may properly exercise the power under CPR r 19.23(1)(a) to order otherwise. For the appellants, it was submitted that CPR 19.23(1)(a) contemplates a judgment or order which is either binding or not binding from the time it is made. I understood the suggestion to be that the rule, properly interpreted, requires any order otherwise to be made when (or before) the judgment which otherwise has binding effect is given and does not permit the court to make such an order at any later time.[194]Like Lady Rose and Lord Richards, I do not consider this to be a tenable interpretation. A party to a claim other than the claim in which the judgment is given may not be aware of what was decided by the judgment until after it has been given and promulgated. It is unreasonable to interpret the rule as requiring such a party to have applied for an order that the judgment is not binding on them before knowing that the judgment exists and what it has decided.[195]In making the rule, the Civil Procedure Rule Committee could have chosen to fix a time limit by which any application for an order otherwise under CPR r 19.23(1)(a) must be made. But they have not done so.[196]In this respect CPR r 19.23 may be compared with the rule that applies to decisions made in lead cases in the First-tier Tribunal (Tax Chamber). Rule 18 of The Tribunal Procedure (First-tier Tribunal) (Tax Chamber) Rules 2009 applies if two or more cases have been started before the tribunal which give rise to common or related issues of fact or law. The rule, in paragraph (2), gives the tribunal power to specify one or more such cases as lead cases and stay the others (referred to as “the related cases”). Paragraphs (3) and (4) provide: “(3) When the Tribunal makes a decision in respect of the common or related issues—(a) the Tribunal must send a copy of that decision to each party in each of the related cases; and(b) subject to paragraph (4), that decision shall be binding on each of those parties. (4) Within 28 days after the date that the Tribunal sent a copy of the decision to a party under paragraph (3)(a), that party may apply in writing for a direction that the decision does not apply to, and is not binding on the parties to, that case.” (a) the Tribunal must send a copy of that decision to each party in each of the related cases; and (b) subject to paragraph (4), that decision shall be binding on each of those parties.[197]There is no equivalent or comparable provision in CPR r 19.23. If the intention was to limit the period within which a party may apply for an order that the judgment is not binding on the parties to another claim, it would be necessary to include one.[198]This is not to say that the time when a party seeks an order otherwise under CPR r 19.23(1)(a) is inconsequential. In principle, delay in applying for such an order is a relevant consideration which the court should take into account in deciding whether it is just to make the order. But there is no cut-off date, let alone one that occurs when the judgment which has binding effect is given.[6]Subsequent events should generally be disregarded[199]The general question raised by this appeal is when, if ever, does it accord with the overriding objective to order otherwise on the basis of a new development, such as a judicial decision overruling earlier authority, occurring after the judgment in relation to a GLO issue was given. There are cogent reasons why any such order should be rare.[200]As discussed above, the principal purpose of using one or more test claims as a vehicle for deciding a GLO issue is to obtain a judgment on that issue which will bind, not only the parties to the test claim itself, but the parties to all other claims in which the issue arises. The underlying goals which the procedure is designed to serve are efficiency (avoiding the costs and delays involved in litigating issues common to multiple claims more than once) and consistency (avoiding the injustice of deciding such common issues differently as between claims in the same group). Those goals would generally be undermined if a party to one of the follower claims in which the issue arises were permitted to re-litigate the issue in circumstances where a party to the test claim is not permitted to do so.[201]In support of their contention that a development in the law occurring after a GLO issue was decided in a test case may justify such re-litigation, counsel for the Revenue emphasised the court’s duty to apply the law correctly. It is a basic tenet of the rule of law that courts should seek to determine cases on the basis of the law. Thus, it will ordinarily be unjust for a person to have their rights determined otherwise than in accordance with the correct legal rules.[202]As a general proposition, this is clearly correct. But the right to have a claim determined on the basis of the law as it currently stands is not unqualified. It is qualified by the principle of finality in litigation. I have referred above to the rules such as cause of action estoppel, issue estoppel and the time limits for appealing which restrict the ability of a party to reopen a matter that has been decided. A material development in the law may in some circumstances justify granting permission to appeal out of time: see Property and Reversionary Investment Corpn Ltd v Templar [1977] 1 WLR 1223. If the possibility of seeking permission to appeal remains open to a party to a test claim in which a GLO issue was decided, then it also remains open to a party to any other claim falling within CPR r 19.23(1)(a). But if appeal rights have been exhausted, the law does not treat the value of ensuring that cases are decided correctly as a sufficient reason to allow a party to the test claim to reopen the judgment on the ground that new material is now available which shows that the decision was wrong. That being so, the relevant question is what, if anything, justifies permitting a party to another claim in which the GLO issue arises and whose appeal rights have also been exhausted to contend that the test claim was wrongly decided when the parties to the test claim cannot do so. Appealing to the principle that rights should ordinarily be determined in accordance with the correct legal rules as currently understood does not help to answer that question.[203]The Revenue sought to answer it by relying on an analogy drawn by the Court of Appeal between the binding effect of a judgment on a GLO issue in a test claim and the doctrine of issue estoppel. As mentioned earlier, this doctrine generally prevents a party from contesting in later proceedings the determination in earlier proceedings between the same parties of an issue which was necessary to the resolution of the earlier claim. In Arnold v National Westminster Bank plc [1991] 2 AC 93, however, the House of Lords held that the rule is not absolute. Lord Keith (with whose speech the rest of the appellate committee agreed), at p 109B, recognised a potential exception to issue estoppel “in the special circumstance that there has become available to a party further material relevant to the correct determination of a point involved in the earlier proceedings … being material which could not by reasonable diligence have been adduced in those proceedings”. Lord Keith added that such further relevant material is not confined to matters of fact and may, at least in some circumstances, include a change in the law. On this basis the House of Lords concluded that, on the facts in Arnold, tenants were permitted to reopen on a later rent review a question of interpretation of the lease which had been decided in the landlord’s favour in earlier proceedings.[204]Here the Court of Appeal considered that, by analogy with this approach, a change in the law may amount to “special circumstances” which justify making an order otherwise under CPR 19.23(1)(a): see para 69 of their judgment.[205]I do not find the suggested analogy persuasive. In a case such as Arnold the claimant is bringing a new claim which has arisen since the judgment said to create an issue estoppel was given. There has also in the meantime been a change in the law. It can readily be accepted that in this situation there are competing considerations of justice and that the balance may sometimes favour allowing the issue to be reopened. In a case of the present kind where CPR 19.23(1)(a) applies, the position is materially different. There is a group of claims in all of which the relevant cause of action has already arisen and which are proceeding in parallel. A test claim has been selected as a vehicle for deciding a common issue because it is inefficient or unworkable to try all the claims together. A final judgment on that issue has been given in the test claim. The aim of the procedure is that, unless there is a relevant factual difference, the decision will be binding on the parties to all the claims in the group. If the law later changes, it is in principle unjust to permit the parties to some, but not all, materially identical claims to reopen the common issue. To allow this is contrary to the underlying precept of one for all, all for one.[206]That potential injustice is magnified by the fact that it may be a matter of happenstance which claim has been used as a test claim and how long the litigation continues after the judgment in the test claim is given. The law is never in a permanently fixed state. It is always capable of development, which sometimes results in judicial decisions being overruled or departed from in ways that might lead to a different outcome if an issue already decided in litigation were to arise for decision again. If group litigation lasts for many years—as these proceedings have done and are an extreme example, as they have now been running for more than 23 years—it is only to be expected that there may be relevant legal developments. If such developments can be used as a basis for reopening issues previously decided in the litigation, legal certainty will be undermined and yet further delay caused.[207]Counsel for the Revenue also argued that the very existence of the power to order otherwise necessarily envisages that there are circumstances in which the power may properly be exercised. I agree. But it does not follow that those circumstances include events which occur after the judgment deciding a GLO issue has been given and all rights of appeal from it have been exhausted. Nor does it follow that the circumstances include a subsequent change in the law.[208]In this context, it is important to recognise that, where a GLO is in place, the GLO issues are not limited to common issues of law. They may comprise issues of fact as well as law and may be “common or related issues” (emphasis added). Not all GLO issues may be truly common to all the claims in the group, in the sense that they arise in precisely the same way or must necessarily be decided uniformly. Limitation is a good example of an issue which is often included among the GLO issues but may not strictly be a common issue. That may be so where, for example, the issue is about when the claimant acquired the knowledge required to start time running, or is whether the court should exercise its discretion under section 33 of the Limitation Act 1980 to extend the limitation period in a personal injury case.[209]An example of group litigation in which such issues arose is the Atomic Veterans Litigation. The claims were for personal injury allegedly caused by exposure to radiation as a result of nuclear tests carried out by the Ministry of Defence in the 1950s. Reversing the decision of the judge, the Court of Appeal found that nine of the ten lead claimants had the knowledge required to start time running more than three years before their claims were commenced and that the judge had been wrong to exercise his discretion under section 33 of the Limitation Act 1980 to extend the limitation period. The Supreme Court dismissed the appeals, which meant that the nine lead claims could not proceed. Lord Wilson JSC observed that, if there would be any particular injustice in visiting adverse judgments in these lead claims upon other, materially similar, claims within the group, the power under CPR r 19.23(1)(a) to order otherwise would cater for it: see AB v Ministry of Defence [2012] UKSC 9; [2013] 1 AC 78, para 15.[210]It may be said that, in cases such as these, there is no need for an order otherwise because, if there is a relevant difference between the facts of the test claim and the facts of another claim on the group register, the judgment in the test claim can be distinguished. But it cannot, in my view, be assumed that this will be so. At the very least, there may be room for argument about whether the decision in the test claim of the GLO issue as formulated dictates the outcome in another case, despite a relevant factual difference. Making an order otherwise may be the simplest, or the only, way of giving effect to the relevant legal distinction.[211]There are, in any case, multifarious situations in which an order may be made to which CPR r 19.23(1)(a) applies. They include giving procedural directions for dealing with the determination of GLO issues, where it may well be appropriate to make orders specific to certain claims or a sub-group of claims, or which do not apply to certain claims or a sub-group of claims, so that an order otherwise is needed. Other examples were suggested by Mr Bremner KC in argument (see para 113 above). I agree with Lady Rose and Lord Richards that it would not be sensible, even if it were feasible, to try to envisage all the circumstances in which an order otherwise might be called for. The court must focus on the kind of case with which this appeal is concerned and which I have described at para 174 above. There is no a priori reason to assume that an order otherwise will sometimes be appropriate in a case of this kind. There are good reasons, which I have sought to explain, why in any such case the starting-point should be that, unless there is a material difference between the facts of the test claim in which a GLO issue has been decided and the facts of another claim that was on the group register when the judgment was given, both claims should stand or fall together.[7]How the question arises here[212]As Lady Rose and Lord Richards have shown, whether the present claims are barred by limitation was not decided by the Prudential test case. So the question whether the court should order otherwise does not arise in relation to “the Limitation Issue”.[213]The question does arise in relation to “the Set Off Issue”, as the Revenue now accept.[214]To put that issue in context, the claims made in this litigation include claims for repayment of advance corporation tax (“ACT”) found to have been unlawfully levied on dividends received by the claimants from foreign companies. The charges were held to be unlawful in so far as the UK tax regime treated dividends received from companies not resident in the UK less favourably than dividends received from UK-resident companies. ACT could be set off against corporation tax which later became payable and, in many cases where ACT was unlawfully levied, sums paid as ACT were later set off against corporation tax lawfully due.[215]Questions arose about the extent of the restitutionary remedies to which claimants were entitled. The GLO issues included the quantum of relief and, more particularly, the basis on which interest was payable. A claim brought by the Prudential Assurance Co and other companies in the Prudential group was designated as a test claim for GLO issues which included these issues.[216]Claims for interest fell into three categories:(a) interest claimed on unlawfully levied ACT which was subsequently set off (or “utilised”) against lawfully levied corporation tax, from the date of payment to the date of set-off;(b) interest claimed on all other unlawfully levied tax (including ACT), from the date of payment to the date of repayment by the Revenue; and(c) interest claimed on the amount of the interest referred to in (a) above, treated as a principal sum, from the date of set-off to the date of payment by the Revenue.[217]In Sempra Metals Ltd (formerly Metallgesellschaft Ltd) v Inland Revenue Comrs [2007] UKHL 34; [2008] AC 561 the House of Lords decided (by a majority) that, in a claim falling within category (a) above, compound interest calculated at conventional government borrowing rates was payable. The rationale for this decision was that the recipient of the unlawfully levied tax (ie the Revenue) had been unjustly enriched by having the use of the money during the relevant period, and that the value of this enrichment was appropriately measured by the cost of borrowing an equivalent amount in the market.[218]Based on what the House of Lords had decided in Sempra Metals, the Revenue conceded in the Prudential test claim that compound interest at conventional government rates was payable in relation to category (a). But, in relation to categories (b) and (c), the Revenue contended that the claimants were entitled only to simple interest under section 35A of the Senior Courts Act 1981.[219]Following a trial in the High Court which encompassed these issues, Henderson J held that there was no rational basis for distinguishing claims in category (b) from those in category (a) and that the reasoning of the majority in Sempra Metals should logically also lead to the conclusion that compound interest was payable in relation to claims in category (c): see Prudential Assurance Co Ltd v Revenue and Customs Comrs [2013] EWHC 3249 (Ch); [2014] STC 1236 (“Prudential HC”), paras 243–246. The judge made an order dated 28 January 2014 reflecting the Revenue’s concession and these conclusions. This order, which I will call “the Prudential HC judgment”, is the judgment or order otherwise binding on the parties to the present claims in respect of which the Revenue has asked the court to order otherwise under CPR 19.23(1)(a).[220]On an appeal to the Supreme Court in the Prudential test case, the Revenue maintained the position it had taken in the High Court, including its concession that compound interest was payable in relation to claims in category (a). But it now advanced an argument which challenged the rationale on which compound interest had been awarded in Sempra Metals. In deciding the appeal, the Supreme Court subjected the reasoning in Sempra Metals to critical analysis and concluded that the majority had been wrong to regard the basis on which interest was payable as the reversal of unjust enrichment. Properly analysed, the receipt of money paid under a mistake of law gives rise to an immediate obligation to repay the amount, which represents a debt. It followed that interest could be awarded on the claims within categories (b) and (c) under section 35A of the 1981 Act; and interest on an equivalent basis could have been awarded on the claim in category (a). The court accordingly decided that Prudential’s claim to compound interest under categories (b) and (c) must be rejected and that its claim under category (a) would also have been rejected, if it had not been accepted by the Revenue: see Prudential Assurance Co Ltd v Revenue and Customs Comrs [2018] UKSC 39; [2019] AC 929 (“Prudential SC”), para 79.[221]Legislation was later enacted in sections 85 and 86 of the Finance Act 2019 which provided a statutory right to be paid simple interest at a prescribed rate for categories (a) and (c). In proceedings relating to a different group of claims, the Supreme Court held that the claimants were restricted to this remedy and could not recover interest under section 35A of the 1981 Act: see Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2021] UKSC 31; [2021] 1 WLR 4354 (“FII SC 3”).[222]Of the two present appellants, only the second (“GREA”) paid ACT which was unlawfully charged. In the light of this court’s decisions in Prudential SC and FII SC3, GREA accepts that only statutory interest can be awarded under section 35A of the Senior Courts Act 1981 in respect of category (b) and under section 85 of the Finance Act 2019 in respect of category (c) and that section 85 will also apply to its category (a) claim if the Prudential HC judgment is not binding on this issue.[223]It is common ground on this appeal that, pursuant to CPR 19.23(1)(a), the Prudential HC judgment is binding on the parties to this claim in relation to the Set Off Issue unless the Court of Appeal was entitled to order otherwise. GREA contends that the Court of Appeal was wrong to do so.[8]Why an order otherwise is justified[224]I have explained why the fact that there has been a material change in the law since the Prudential HC judgment was given is not enough to justify making an order otherwise. The reasons which persuade me that it is nevertheless appropriate to do so are that, on the unusual facts of this case, neither of the two key purposes of treating the judgment as binding on the parties to other claims is undermined by making such an order. As discussed (see paras 179 and 200 above), those key purposes are avoiding, first, inefficiency/waste of resources and, second, the potential for inconsistent outcomes involved in re-litigating a GLO issue that has already been finally decided.[225]In agreement with Lady Rose and Lord Richards, I think it critical that an order otherwise will not result in a GLO issue decided by the Prudential test case being re-argued. That is because it has been conclusively established by the Supreme Court’s decisions in the Prudential case itself and in FII SC 3 that, on the question of interest, the Prudential HC judgment was wrong in law. The appellants therefore (rightly and inevitably) accept that, if an order otherwise is made, the Set Off Issue must be decided against them without the need for any further argument on the issue. So reopening the issue will not result in any additional cost and delay or other waste of resources.[226]There remains the injustice of reaching inconsistent results on exactly the same issue of law in the test claim and in other claims that were on the group register when the test claim was decided. That inconsistency is indubitable. But its significance is, in my view, outweighed in this case by the incoherence that would result from treating the Prudential HC judgment as binding on the parties to GREA’s claim in relation to the claim for interest in category (a).[227]The critical point here is that, as was recognised in the Prudential test case itself, there is no rational basis for reaching different outcomes in relation to the three categories of claims for interest referred to at para 216 above. The reasoning in Sempra Metals which, if valid, justified awarding interest that reflected the cost of borrowing the sums unlawfully levied by the Revenue applied equally to all the categories of interest claimed. Conversely, if that reasoning was invalid, awarding interest on that basis was wrong across the board. That was why Henderson J, on whom Sempra Metals was binding as a precedent, decided that interest must logically be awarded on the same basis in relation to claims in categories (b) and (c) as for claims in category (a). On appeal the Supreme Court agreed with that logic but drew the opposite conclusion because they held that Sempra Metals has been wrongly decided on this issue. For that reason, having rejected Prudential’s claim to compound interest under categories (b) and (c), the court said that Prudential’s claim under category (a) would also have been rejected, if it had not been conceded by the Revenue.[228]As GREA accepts, the decision in Prudential SC has the effect that GREA cannot recover interest on the Sempra Metals basis on unlawfully levied ACT which was not utilised by subsequent set-off against lawfully levied corporation tax (category (b)). Nor can it recover interest on that basis on the time value of utilised ACT since the date of set-off (category (c)). There is no suggestion that GREA could properly seek an order otherwise to disapply the binding effect of that decision. Nor could or does GREA take issue with the application to its claims in categories (b) and (c) of the rates of interest prescribed by (respectively) section 35A Senior Courts Act 1981 andsection 85 of the Finance Act 2019.[229]In these circumstances it would be irrational and incoherent (as well as incorrect in law) to assess interest on utilised ACT during the period from the date of payment by GREA to the date of set-off (category (a)) on the Sempra Metals basis. As the Court of Appeal pointed out, the illogicality would be made even more stark by the fact that the “principal” amount which forms the basis of the award under category (c) is derived from category (a): AXA Sun Life plc v HMRC [2024] EWCA Civ 1430; [2025] 1 WLR 2179, para 81.[230]The palpable irrationality of assessing interest on GREA’s category (a) claim on a different basis from its category (b) and (c) claims is, in my opinion, far more offensive to the basic principle that like cases should be decided alike than the inconsistency involved in assessing such interest on a different basis from Prudential’s category (a) claim (because of the concession made by the Revenue in the Prudential test case). When this factor is combined with the fact that no further argument is required to decide the Set Off Issue, it was right to make an order otherwise.[231]I have not overlooked the fact that a long time elapsed from the date of the decision in Prudential SC which left the Prudential HC judgment intact (25 July 2018) before the Revenue asked the court, if necessary, to order otherwise (which appears to have been at or around the time when the Set Off Issue was identified as a preliminary issue in this case in September 2022). Normally a delay of this length would be a compelling factor against making an order otherwise. But in this exceptionally complex and protracted litigation it made sense to leave the stay on proceeding with other claims in place until the related FII Group Litigation was resolved; and no serious attempt has been made to argue that the Revenue was guilty of unreasonable delay.

Conclusion

[232]For these reasons I concur in the decision that the appeal should be dismissed. 268. ANNEX: KEY CASES ABBREVIATION CASE NAME DATE Kleinwort Benson Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349 29 October 1998 Hoecsht Metallgesellschaft Ltd v Inland Revenue Comrs and Hoechst AG v Inland Revenue Comrs (Joined Cases C-397/98 and 410/98) [2001] Ch 620; [2001] ECR I-1727 8 March 2001 DMG Deutsche Morgan Grenfell Group plc v Inland Revenue Comrs [2006] UKHL 49; [2007] 1 AC 558 25 October 2006 Sempra Metals Litigation Sempra Metals HC Sempra Metals Ltd v Inland Revenue Comrs [2004] EWHC 2387 (Ch); [2004] STC 1178 16 June 2004 Sempra Metals CA Sempra Metals Ltd v Inland Revenue Comrs [2005] EWCA Civ 389; [2006] QB 37 12 April 2005 Sempra Metals HL Sempra Metals Ltd v Inland Revenue Comrs [2007] UKHL 34; [2008] AC 561 18 July 2007 Littlewoods Litigation Littlewoods HC 2010 Littlewoods Retail Ltd v Revenue and Customs Comrs [2010] EWHC 1071 (Ch); [2010] STC 2072 19 May 2010 Littlewoods CJEU Littlewoods Retail Ltd v Revenue and Customs Comrs (Case C-591/10) EU:C:2012:478; [2012] STC 1714 19 July 2012 Littlewoods HC 2014 Littlewoods Retail Ltd v Revenue and Customs Comrs [2014] EWHC 868 (Ch); [2014] STC 1761 28 March 2014 Littlewoods CA Littlewoods Ltd v Revenue and Customs Comrs [2015] EWCA Civ 515; [2016] Ch 373 21 May 2015 Littlewoods SC Littlewoods Ltd v Revenue and Customs Comrs [2017] UKSC 70; [2018] AC 869 1 November 2017 Prudential Litigation Prudential HC Prudential Assurance Co Ltd v Revenue and Customs Comrs [2013] EWHC 3249 (Ch); [2014] STC 1236 24 October 2013 Prudential CA Prudential Assurance Co Ltd v Revenue and Customs Comrs [2016] EWCA Civ 376; [2017] 1 WLR 4031 19 April 2016 Prudential SC Prudential Assurance Co Ltd v Revenue and Customs Comrs [2018] UKSC 39; [2019] AC 929 25 July 2018 Class 8 CFC and Dividend Group Litigation CFC and Dividend Group Litigation Claimants in Class 8 of the CFC and Dividend Group Litigation v Revenue and Customs Comrs [2019] EWHC 338 (Ch); [2019] 1 WLR 5097 20 February 2019 FII Group Litigation FII CJEU 1 Test Claimants in the FII Group Litigation v Inland Revenue Comrs (Case C-446/04) EU:C:2006:774; [2012] 2 AC 436 12 December 2006 FII CA 1 Test Claimants in theFII Group Litigation v Revenue and CustomsComrs [2010] EWCA Civ 103; [2010] STC 1251 23 February 2010 FII SC 1 Test Claimants in theFII Group Litigation v Revenue and CustomsComrs [2012] UKSC 19; [2012] 2 AC 337 23 May 2012 FII CJEU 2 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (formerly Inland Revenue Comrs) (No 3) (Case C-35/11)EU:C:2012:707; [2013] Ch 431 13 November 2012 FII CJEU 3 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (Case C-362/12); EU:C:2013:834; [2014] AC 1161 12 December 2013 FII CA 2 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2016] EWCA Civ 1180; [2017] STC 696 24 November 2016 FII SC 2 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs[2020] UKSC 47; [2022] AC 1 20 November 2020 FII SC 3 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs[2021] UKSC 31; [2021] 1 WLR 4354 23 July 2021 DATE CASE NAME ABBREVIATION 29 October 1998 Kleinwort Benson Ltd v Lincoln City Council [1999] 2 AC 349 Kleinwort Benson 8 March 2001 Metallgesellschaft Ltd v Inland Revenue Comrs and Hoechst AG v Inland Revenue Comrs (Joined Cases C-397/98 and 410/98) [2001] Ch 620; [2001] ECR I-1727 Hoecsht 16 June 2004 Sempra Metals Ltd v Inland Revenue Comrs [2004] EWHC 2387 (Ch); [2004] STC 1178 Sempra Metals HC 12 April 2005 Sempra Metals Ltd v Inland Revenue Comrs [2005] EWCA Civ 389; [2006] QB 37 Sempra Metals CA 25 October 2006 Deutsche Morgan Grenfell Group plc v Inland Revenue Comrs [2006] UKHL 49; [2007] 1 AC 558 DMG 12 December 2006 Test Claimants in the FII Group Litigation v Inland Revenue Comrs (Case C-446/04) EU:C:2006:774; [2012] 2 AC 436 FII CJEU 1 18 July 2007 Sempra Metals Ltd v Inland Revenue Comrs [2007] UKHL 34; [2008] AC 561 Sempra Metals HL 23 April 2008 Test Claimants in the CFC and Dividend Group Litigation v Inland Revenue Comrs (Case C-201/05) EU:C:2008:239; [2008] STC 1513; [2008] ECR I-2875 The CFC Reasoned Order 23 February 2010 Test Claimants in theFII Group Litigation v Revenue and CustomsComrs [2010] EWCA Civ 103; [2010] STC 1251 FII CA 1 19 May 2010 Littlewoods Retail Ltd v Revenue and Customs Comrs [2010] EWHC 1071 (Ch); [2010] STC 2072 Littlewoods HC 2010 23 May 2012 Test Claimants in theFII Group Litigation v Revenue and CustomsComrs [2012] UKSC 19; [2012] 2 AC 337 FII SC 1 19 July 2012 Littlewoods Retail Ltd v Revenue and Customs Comrs (Case C-591/10) EU:C:2012:478; [2012] STC 1714 Littlewoods CJEU 13 November 2012 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (formerly Inland Revenue Comrs) (No 3) (Case C-35/11)EU:C:2012:707; [2013] Ch 431 FII CJEU 2 24 October 2013 Prudential Assurance Co Ltd v Revenue and Customs Comrs [2013] EWHC 3249 (Ch); [2014] STC 1236 Prudential HC 12 December 2013 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs (Case C-362/12); EU:C:2013:834;[2014] AC 1161 FII CJEU 3 28 March 2014 Littlewoods Retail Ltd v Revenue and Customs Comrs [2014] EWHC 868 (Ch); [2014] STC 1761 Littlewoods HC 2014 18 December 2014 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2014] EWHC 4302 (Ch); [2015] STC 1471 FII HC 2 21 May 2015 Littlewoods Ltd v Revenue and Customs Comrs [2015] EWCA Civ 515; [2016] Ch 373 Littlewoods CA 19 April 2016 Prudential Assurance Co Ltd v Revenue and Customs Comrs [2016] EWCA Civ 376; [2017] 1 WLR 4031 Prudential CA 24 November 2016 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2016] EWCA Civ 1180; [2017] STC 696 FII CA 2 1 November 2017 Littlewoods Ltd v Revenue and Customs Comrs [2017] UKSC 70; [2018] AC 869 Littlewoods SC 25 July 2018 Prudential Assurance Co Ltd v Revenue and Customs Comrs [2018] UKSC 39; [2019] AC 929 Prudential SC 20 February 2019 Claimants in Class 8 of the CFC and Dividend Group Litigation v Revenue and Customs Comrs [2019] EWHC 338 (Ch); [2019] 1 WLR 5097 CFC and Dividend Group Litigation 20 November 2020 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs [2020] UKSC 47; [2022] AC 1 FII SC 2 23 July 2021 Test Claimants in the FII Group Litigation v Revenue and Customs Comrs[2021] UKSC 31; [2021] 1 WLR 4354 FII SC 3 26 April 2023 AXA Sun Life plc v Inland Revenue Comrs [2023] EWHC 944 (Ch); [2023] STC 1167 27 November 2024 AXA Sun Life plc v Inland Revenue Comrs [2024] EWCA Civ 1430; [2025] 1 WLR 2179

order

Test Claimants in the CFC and Dividend Group Litigation v Revenue and Customs Comrs (Case C-201/05) [2008] ECR I-2875; [2008] STC 1513 (“the CFC Reasoned Order”). The questions can, for our purposes, be grouped broadly into two kinds. The first sought answers as to whether the tax regime challenged was incompatible with EU law. The second asked about the nature of the remedy to which taxpayers were entitled if they had paid tax which the CJEU now determined was not lawfully demanded.