Sempra Metals Ltd. v Inland Revenue & Anor

In this judgment I use the following abbreviations.

This judgment relates to another in the series of cases which have been heard, and in some cases remain to be heard, arising from the decision of the CJEC in Metallgesellschaft/Hoechst . Most of the cases are being brought and determined within the ambit of a Group Litigation Order, a GLO, of which I am the nominated judge. More details are given in the judgments of myself and the Court of Appeal in Pirelli Cable Holding NV v IRC [2003] EWHC Ch 32, [2003] STC 130; [2003] EWCA Civ 1849, [2004] STC 130.

The cases concern dividends paid by United Kingdom subsidiaries of overseas parent companies at times when the ACT system was in force as part of United Kingdom tax law. The normal operation of the system required a company which paid a dividend also to pay to the Revenue an amount of tax, called advance corporation tax, which it could later set off against what would otherwise have been its normal liability to pay corporation tax (sometimes referred to as ‘mainstream corporation tax’, or MCT) on its taxable profits. The effect was that, if a company chose to pay a dividend, it had to accept as a consequence that it would have to pay some of the corporation tax on its profits sooner than it would otherwise have had to do. On the assumption that the company did, sooner or later, have profits liable to MCT, the ACT system imposed on it a timing disadvantage and conferred a corresponding timing advantage on the Revenue. If the company did not, sooner or later, have profits liable to MCT, the ACT system was more drastic in its effects. However, the present case is concerned with companies which did, in time, have profits liable to MCT, so the disadvantage sustained by them was limited to the timing disadvantage.

There was an exception to the normal rule that a company which paid a dividend also had to pay ACT. If the dividend-paying company was a subsidiary of a United Kingdom parent company the two companies could join in making a group income election under ICTA 1988 s.247. The effect of an election was that the dividend could be paid without an accompanying liability to pay ACT, so that the timing disadvantage to the paying company was removed. However, group income elections could only be made where both the subsidiary and the parent company were resident in the United Kingdom. In Metallgesellschaft/Hoechst the CJEC decided that it was contrary to Community law (specifically article 52 of the EC Treaty, now renumbered as article 43) for United Kingdom domestic tax law to deny the ability to make group income elections to groups where the subsidiary (the dividend payer) was resident in the United Kingdom but the parent (the dividend recipient) was resident in another member state. This meant that United Kingdom law had effectively required the United Kingdom subsidiary to pay part of its mainstream corporation tax (its MCT) prematurely – earlier than, on a proper application of Community law, it ought to have had to pay it. The court also decided that United Kingdom subsidiaries of parents in other member states which had paid ACT in the past were entitled to compensation for the timing disadvantage which they had suffered. The amounts of compensation and other associated issues were to be determined by the national court.

A large number of multi-national groups, including their United Kingdom dividend-paying subsidiaries, commenced claims in the English courts. In most cases, including the present one, the parent companies which received the dividends were resident in other member states of the European Union, but some claims have also been brought where the parent company was resident in a third jurisdiction. In those cases the claims are based, not on Community law, but on non-discrimination articles in double taxation agreements. For an example see my judgment in NEC Semiconductors Ltd v IRC [2003] EWHC Ch 2813, [2004] STC 489. The cases have all been gathered together within a GLO, and I have already delivered judgments in three cases which raised issues of principle different from the one raised in the present case. I now deliver judgment in a fourth case. (I have also delivered judgment in another case which was consequential on the CJEC decision in Metallgesellschaft/Hoechst , but that case was not brought under the GLO. As with previous cases my role at this stage is limited to one of determining a question of principle.

It is clear from the CJEC judgment in Metallgesellschaft/Hoechst that United Kingdom subsidiaries of member state parents which paid ACT in circumstances where, if they had been permitted by United Kingdom domestic law to make group income elections, they would not have paid it are entitled to a remedy from the United Kingdom (represented in this connection by the Revenue), and that the remedy may be described as either restitution or compensation. In cases where the amount of ACT paid has later been set off against a liability to MCT, the restitution or compensation was to be calculated by reference to interest over the period from the date when the ACT was in fact paid until the later date when it was set off against MCT. In this judgment I refer to that period as ‘the premature tax payment period’: the basis of the CJEC’s decision that United Kingdom law was in breach of the EC treaty was that the ACT system had the effect of compelling United Kingdom subsidiaries of parents in other member states to pay part of their corporation tax prematurely.

The main question of principle in this case is whether the calculation of interest over the premature tax payment period (or, as I think it may be more accurate to express it, the calculation of an amount equal to interest over the premature tax payment period) should be effected on the basis of compound interest, as Sempra contends, or on the basis of simple interest, as the Revenue contend. In my judgment it should be effected on the basis of compound interest (always assuming that it has been pleaded in that way). I will explain my reasons as this judgment progresses, but the key point is that a compound interest calculation will in principle result in full restitution or compensation, whereas a simple interest calculation would result only in partial restitution or compensation. In my judgment the decision of the CJEC requires full restitution or compensation, and therefore a calculation on a compound basis gives fuller effect to the decision than would a calculation on a simple basis.

There is a second question which has also been raised by the GLO and the pleadings. The decision of the CJEC in Metallgesellschaft/Hoechst shows that the entitlement to restitution or compensation crystallised when the amount prematurely paid by way of ACT was set off against a liability to MCT. So that is when the cause of action accrued. The main question addressed in this judgment concerns how the amount of restitution or compensation falls to be ascertained. But, whatever the amount of it may have been, the claimant seeks interest on it for what I will call ‘the post-utilisation period’, that is the period from the time when the cause of action crystallised until judgment. That is a different species of interest altogether from that to which the main question relates. It is not interest required by Community law. It is interest under domestic United Kingdom law, on the basis of the normal rule that there can be included in a judgment interest from the date when the cause of action accrued until the date of judgment. The question is whether this category of interest should also be calculated on a compound basis. My answer is that it should not. The reason is that the entitlement to it arises, not by virtue of Community law, but under a domestic statute, and the statute expressly provides that the interest is to be simple interest: for High Court cases see the Supreme Court Act 1981 s.35A; for County Court cases see the County Courts Act 1984 s.69.

Since the purpose of this judgment is to determine a question of legal principle, nothing much turns on the detailed facts of the case. I can therefore describe them comparatively briefly.

The claimant is now called Sempra Metals Ltd. That is not the name by which it has always been known. Its previous name was Metallgesellschaft Ltd. Thus it was the claimant in one of the two cases which were referred to the CJEC some years ago and which gave rise to the judgment of that court in Metallgeselschaft/Hoechst . Sempra is a company resident in the United Kingdom. It has at all times been a subsidiary of a German parent company. The identity of the parent company has changed on one occasion, but nothing turns on that detail. Thus dividends paid by Sempra have gone to its German parent, a company established in another member state of the European Union. During the years when ACT was in force Sempra paid several dividends, and in the case of all of them it paid the associated ACT. It is accepted by the Revenue that, if the United Kingdom legislation had permitted group income elections to be made between United Kingdom subsidiaries and parent companies established in other member states, the Sempra group (then the Metallgesellschaft group) would have made one, and that the dividends which Sempra paid to its German parent would not have required Sempra to pay ACT to the Revenue.

Four sample dividends and associated ACT payments have been agreed upon for the purposes of this test case. I do not need to give details of them. The earliest ACT payment was made on 12 October 1981, and the latest on 18 July 1994. In connection with the first three payments issues of limitation of actions could arise. However, it is accepted that, as matters now stand, my decision in Deutsche Morgan Grenfell v IRC [2003] EWHC 1779 (Ch), [2003] STC 1017, means that, since one of the bases on which Sempra’s claim is pleaded is restitution by reason of the ACT having been paid under a mistake of law, Sempra’s claim would not be statute-barred. My judgment in that case is subject to appeal, and if the judgment is not upheld the limitation argument could become important in the present case. However, I need not refer to it again in this judgment.

All the four sample payments of ACT were, sooner or later, set off against MCT. The intervals between the payments of the ACT and the times of set-off (those intervals being what I am referring to in this judgment as the premature tax payment periods) varied considerably. The longest period was almost ten years and the shortest was just under one year.

Sempra’s business involves trading, both as principal and as broker for clients, in metals listed on the London Metal Exchange. At almost all times Sempra was in a net borrowing position, in the sense that its cash borrowings exceeded its cash assets. I had extensive evidence from Sempra about its detailed financial position over the years, but, for reasons which I will explain, in my opinion the evidence does not affect my conclusion, and I hope that I will be forgiven if I refrain from attempting to summarise it. I also had evidence about the detailed financing and banking arrangements of the United Kingdom government. Again however, interesting and informative though that evidence was, I do not think that it affects this case, and I will not go into it in this judgment.

A number of questions about the manner in which the United Kingdom ACT system applied to United Kingdom subsidiaries of parent companies in other member states were referred to the CJEC by the High Court. The first question to be decided was whether the absence of a power to make group income elections in such cases was contrary to Community law, and contrary in particular to the article about freedom of establishment (article 52 at the time, article 43 now). The court answered that it was. The court then moved on to the questions which are relevant to the present case. I will quote the court’s paraphrase of those questions. This is in paragraph 77 of the judgment:

In subsequent paragraphs the CJEC outlined the arguments and discussed the issues. It then gave its answers to the questions, doing so in paragraph 96 of the judgment. I shall have to look quite closely at the detailed discussion, but I think that it is convenient for me first to quote the court’s summary of its answers. This is in paragraph 96 of the judgment:

I will now go back to the discussion which preceded that answer. I would recommend anyone who is closely concerned with the subject matter of my judgment to study the whole of paragraphs 77 to 96 of the CJEC’s judgment, and not simply to rely on what I am about to say. However, I am not going to prolong this judgment excessively by reproducing all of those paragraphs. I will, however, pick out, explain, and discuss certain passages which appear to me to be particularly important for the present case. References to paragraphs are to paragraphs of the CJEC’s judgment.

The court stressed that it was not for it, but rather it was for the claimants, to choose whether to formulate their claim as one for restitution or as one for compensation for damage: paragraph 81. Sempra has formulated its claim in both ways, in the alternative.

Although the term ‘restitution’ might imply that it was necessary to identify the benefit to the national exchequer and require the government to restore that benefit to the claimant, the discussion in paragraphs 82, 88 and 89 seems to me to proceed on the basis that the amount of a restitutionary remedy would be equal to the interest which the claimant could have obtained from the use of the money if it had not paid it to the Revenue by way of ACT. In paragraph 88 the court states: ‘… in an action for restitution the principal sum due is none other than the amount of interest which would have been generated by the sum, the use of which was lost as a result of the premature levy of the tax.’

To a similar effect paragraph 89 referred to ‘interest accrued on the advance corporation tax paid by the subsidiary during the period between the payment of advance corporation tax and the date on which mainstream corporation tax became payable’ , and continued to say that ‘that sum may be claimed by way of restitution’ . Mr Glick commented in argument, and I agree, that, on the basis of the judgment, a restitutionary remedy and a compensatory remedy would both produce the same result, since both of them look to the same thing: what the taxpaying company has lost by reason of having to pay tax early, not what the Government has gained.

The court was clearly of the view that, if a claimant company formulated its claim in restitution, Community law required that in principle the claim should succeed. In particular it could not be defeated by the argument that, under the national law, the courts would not award interest if there was no principal sum owing at the time when the claim for interest was made. In a case, like Sempra’s, where the ACT had already been set off against MCT, there was no principal sum of ACT owed by the Revenue to the company when the company commenced its action. The Revenue argued that, in such a case, English law did not recognise a cause of action which simply claimed interest. That was said to be the result of several cases, including particularly President of India v La Pintada Compania Navigacion SA [1985] 1 AC 429. I can, I think, paraphrase the CJEC’s reaction to that argument as being that, whatever the position might be under domestic English law, the case before it was a matter of restitution or compensation for breach of an article of the EC treaty, and the remedy could not be denied on the basis that an English court would not give judgment for an amount of interest in similar circumstances.

The court accepted the proposition that it was for national law to settle procedural and ‘ancillary’ questions ‘relating to the reimbursement of charges improperly levied’ . It explained, however, that that principle did not apply to the present claim. The core of the court’s reasoning was set out in paragraph 87, which I believe I should quote in full:

The court took the same view when it considered the position if the claim was formulated as one for compensation rather than as one for restitution. The case was one of a breach of Community law, and Community law required that a remedy should be available. See paragraphs 91 and 95. Further, the La Pintada argument (the argument that, because no principal sum was owing – the ACT having previously been utilised to discharge an undisputed liability to MCT – the English courts would not award as damages an amount of interest by itself) could not be accepted as a defence to the claim. The court again made the point which Mr Rabinowitz encapsulated in his term ‘the primary loss’. Paragraph 93 includes this sentence: ‘However, in the present cases, it is precisely the interest itself which represents what would have been available to the claimants, had it not been for the inequality of treatment, and which constitutes the essential component of the right conferred on them.’

The foregoing are my specific comments on the critical passages in the paragraphs of the CJEC’s judgment which lead up to its formal answer to the question referred to it. I should specifically point out that, although the court frequently refers to ‘interest’ and the period over which it falls to be calculated (the period which I have called the premature tax payment period), it nowhere states whether interest should be calculated over that period on a simple basis or on a compound basis. My principal task in this judgment is to decide which of the two bases appears more precisely to accord with the principles which underlie the court’s decision.

Before I turn to the main issue (should interest be computed on a compound basis or not) there is one other issue which I should discuss. Should the restitution or compensation be calculated on what Mr Glick called a ‘conventional basis’, that is by reference to a rate of interest which is the same for all claimants, or should it be calculated (or at least be capable of being calculated) on an actual basis, that is by examining the particular circumstances for each claimant company? I think that Mr Rabinowitz was inclined to say that it could be done either way, depending on how the particular claimant chose to formulate the quantum of its claim. Mr Glick submitted that it should be done on a conventional basis in all cases. On this issue I agree with Mr Glick.

It seems to me that a calculation of restitution or compensation on a conventional basis is what the CJEC had in mind, though I accept that in saying that I am to a degree reading what is between the lines rather than what is explicitly set out in them. The question submitted to the CJEC by the national court was whether Community law conferred a right ‘to claim a sum of money by way of interest on the advance corporation tax which the subsidiary paid’, a formulation which seems clearly to have assumed an ascertainment of the amount on a conventional basis. There is nothing in the CJEC’s discussion of the question or in its answer to it which suggests that the question had made an inappropriate assumption. I repeat a sentence from paragraph 87 of the judgment: ‘… 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 .. .’ The vice in the United Kingdom law was that it compelled tax to be paid prematurely, so that the subsidiary suffered a loss of the use of money for a time. The natural method of measuring the time value of the use of money, or of the loss of such use, is by an interest calculation, and in my view that is what the CJEC had in mind.

If, contrary to my view, it is not implicit in the CJEC’s decision that restitution or compensation should in all cases be measured by reference to interest and on a conventional basis, then I would take the view that the CJEC left it to the national court to resolve the precise basis of measurement. On that footing I would unhesitatingly choose that the measure should indeed operate by reference to interest and on a conventional basis. (Whether the conventional basis should assume compound or simple interest is another question, to which I turn in paragraphs 23 et seq below.)

The alternative to the use of a conventional basis would appear to leave it open to any particular claimant to tender evidence to the effect that, in its business, the return on capital employed was higher than generally prevailing interest rates, and to argue that therefore it should receive a greater sum by way of restitution or compensation. The Revenue, who are likely to have reasonably full records of companies of the substance to be participants in the GLO, might wish to put in evidence of their own to a contrary effect. Conversely, in the case of some other company it would be possible for the Revenue to make the first move and tender evidence that the company’s return on capital employed was below prevailing interest rates, or even negative, so that very low restitution or compensation, or even none at all, should be awarded. The company would probably wish to respond by adducing its own evidence by way of challenge to the Revenue’s analysis. I do not believe that it would be appropriate for the cases covered by this aspect of the GLO to become embroiled in details of that sort. Sempra has produced eight lever arch files of detailed documentation about its trading activities over the premature tax payment periods, and even then I do not think that it has proposed a specific rate of interest by reference to which its restitution or compensation should be calculated.

I was told that between 50 and 70 claimants which are participants in the GLO are awaiting the outcome of Sempra’s case. The point of principle with which they are all concerned is whether interest over the premature tax payment periods should be calculated on a compound basis or on a simple basis. For reasons which I am going to explain, in my opinion it should be calculated on a compound basis, but in my judgment the rate of interest to be used should be derived from prevailing levels of interest rates in the market generally, and should in principle be the same for all claimants. It should not vary with the particular levels of profitability or non-profitability of the large number of separate claimant companies. I add in that connection that Mr Glick, in submitting that restitution or compensation should be calculated on a conventional basis, acknowledged that some claimants might thereby secure a higher level of payment than they would achieve if they had to prove the individual levels of profitability which they themselves were attaining in the premature tax payment periods. He nevertheless submitted that, on pragmatic grounds even if for no other reason, the conventional basis of calculation should be the one to be adopted. I agree with him, and I do not think that Mr Rabinowitz regarded this as a major point of principle on which he would strenuously argue for a different basis.

I turn now to what is the major point of principle. Should the restitution or compensation be calculated by reference to compound interest or by reference to simple interest? As I have indicated, my judgment is that it should be calculated by reference to compound interest.

It is true that English law has in some contexts been reluctant to include interest in awards of damages, and I was referred to several English cases which demonstrate that proposition. I was also referred to cases which show that, although English courts may in some instances award compound interest, it is exceptional for them to do so. It is also true that the intervention of Parliament was necessary before most awards of damages could carry interest for the period from the accrual of the cause of action until judgment. However, although both counsel helpfully and thoroughly took me through much of the English case law, they both accepted that the present case ought to be determined by reference to Community law, and especially by reference to the principles laid down by the CJEC judgment in Metallgesellschaft/Hoechst . I agree with them, and in the circumstances I do not propose to examine the difficult issues of English law which were explained to me.

In my judgment it is as a matter of Community law that the CJEC required restitution or compensation to be paid by the Revenue to United Kingdom subsidiaries which the ACT system (in breach of Community law) effectively compelled to pay parts of their corporation tax liabilities prematurely. The court has entrusted the ascertainment of the restitution or compensation to the national court, but the remedy to which the national court gives effect is a remedy required by Community law. It is irrelevant whether national law would also give a remedy, and it is also irrelevant to enquire whether an analogous remedy under national law would be measured on a compound basis or on a simple basis.

I have described in paragraph 16(iv) above how, in Metallgesellschaft/Hoechst , the Revenue argued, in reliance on the La Pintada case (supra), that, because English law does not recognise a cause of action for interest where there was no principal debt outstanding at the time when the action was commenced, no restitution or compensation should be payable to the claimant companies. I paraphrased the CJEC’s answer as follows: whatever the position might be under domestic English law, the case before the court was a matter of restitution or compensation for breach of an article of the EC treaty, and the remedy could not be denied on the basis that an English court would not give judgment for an amount of interest in similar circumstances. It seems to me that to argue that, although a remedy must be given (because the CJEC has so decided) and although it must be calculated by reference to interest (because the CJEC has decided that as well), nevertheless the interest should be restricted to simple interest because that is what would happen in the case of a claim brought under English law, would be to reintroduce essentially the same argument as that which the CJEC rejected. It is true that in the CJEC the Revenue were attempting to use principles of English law in order not to have to pay restitution or compensation at all, whereas the question now is whether principles of English law can be invoked to secure that the Revenue should pay lower amounts by way of restitution or compensation. However, the principle is the same. I believe that I have to look for the measure of restitution or compensation which Community law requires, untrammelled by restrictions which English law may impose in situations which might be said to be comparable in some respects.

In my judgment Community law requires the remedy to be a full one and not a partial one. In paragraph 87 of the CJEC’s judgment the court said that ‘the award of interest … would appear to be essential in restoring the equal treatment guaranteed by article 52 of the Treaty’ . In my opinion only compound interest will fully restore equal treatment.

The following example shows why that is so.

Assume two United Kingdom subsidiary companies, S1 which is a subsidiary of a parent company resident in a member state other than the United Kingdom, and S2 which is a subsidiary of a parent company resident in the United Kingdom.

On the same date S1 and S2 each paid a dividend of an identical amount to its parent company.

S1, acting in accordance with the provisions of the domestic United Kingdom tax legislation, paid ACT of, say, £1m by reference to the dividend.

S2, having made a group income election with its parent company, paid no ACT, but kept £1m which it would otherwise have had to pay to the Revenue by way of ACT.

Five years go by before either company has sufficient taxable profits to be liable to pay any MCT to the Revenue. Each company is then in principle liable to pay MCT of more than £1m.

S1 discharges £1m of its MCT liability by setting off against it the £1m of ACT which it had paid five years earlier.

S2 pays the full amount of its MCT to the Revenue in cash, but it does so only after the five years, when it has a liability to MCT.

S1 and S2 have been treated unequally: S1 had to pay £1m of its corporation tax liability five years prematurely; S2 did not.

For equal treatment to be restored (the concept stated by the CJEC in paragraph 87 of the judgment) S1 must receive restitution or compensation from the Revenue of five years interest, and the interest must be compound interest.

Five years simple interest (that is one year’s interest multiplied by five) would still leave S1 and S2 in unequal positions. Suppose that an annual interest rate of 5% is taken. If S1 received restitution or compensation of five years simple interest the amount would be £250,000 (one year’s interest of £50,000 multiplied by five). However, S2 had retained £1m for five years whereas S1 had paid £1m to the Revenue. If S2 had invested its retained £1m by laying it out at interest for five years (an assumption which it is appropriate to make for the reason explained in the next paragraph), S2 would have earned five years’ compound interest. If I assume the same interest rate of 5% per annum and I also assume yearly rests, S2’s interest return over the five years would have been £276,282. So a payment by the Revenue to S1 of £276,282 will restore equal treatment, whereas a payment of £250,000 will not.

In sub-paragraph (x) above I assumed that S2 (which, instead of paying £1m of ACT when it paid the dividend, kept the £1m and should be assumed to have retained it for five years until it used it to pay part of an MCT liability) would have invested the £1m at interest over the five years. That assumption is the logical corollary of the point which I have decided in earlier paragraphs, namely that restitution or compensation should be calculated on a conventional basis, assuming a market-based interest rate and not seeking to identify what each separate claimant company, if it had been permitted to make a group income election, would have done with the money which, on that hypothesis, it would have kept rather than paying over to the Revenue. I am comparing two companies, one of which has to pay money to the Revenue prematurely and the other of which does not. If the restitution or compensation to the first company is to be calculated by reference to a conventional rate of interest and a period equal to that by which the payment was premature, then the second company, which retained money instead of paying it to the Revenue, must be assumed to have earned interest on the money at the conventional rate over that same period: otherwise the comparison would not be valid.

Another point which I wish to make is that, in my view, it does not make any difference (or at least not any major difference) whether or not the company which paid the ACT (Sempra in the sample case which was argued before me) was in a cash surplus position or in a net borrowing position at the time. Going back to my example, if S1 (the subsidiary of the non-United Kingdom parent company which, according to the United Kingdom statute, could not make a group income election) had cash of £1m or more at the time when it paid the £1m of ACT, it would not have needed to borrow to pay the ACT, but it would have lost the opportunity to lay out its £1m and earn compound interest on it over five years. Conversely, if S1 was in a net borrowing position it would have had to borrow a further £1m to fund the ACT payment, and would have had to bear interest on the £1m over five years. On the assumption that the borrowing remained outstanding for the five years (an assumption which in my judgment ought logically to be made) the interest cost would be a compound interest cost, not a simple interest cost.

It may be asked whether the logical analysis which I have attempted to explain could be affected by what happens to the United Kingdom subsidiary’s ordinary business operations over the premature tax payment period (a period of five years in my examples). In my view it could not be affected. For example, if in the first instance the subsidiary (S1 in the hypothetical case which I assumed above) borrowed the money which it needed to pay the ACT, but later was able to repay the borrowing from cash flows from other activities, there would still have been a compound interest cost, but now it would have arisen in a more complex way. As long as the borrowing was outstanding the subsidiary was liable to pay compound interest to its lender. When the borrowing was paid off that particular compound interest-bearing liability was no longer in existence, but the subsidiary could not lay out £1m of its own money so as to earn compound interest: it had chosen to use that £1m to repay a liability which it would not have had but for the apparent liability to ACT, and if it had not had that liability it would have been able to earn a compound return by investing its cash surplus of £1m which arose from other activities. A different permutation of events would be if the subsidiary was able in the first instance to pay the ACT without borrowing, but later (still within the premature tax payment period) had to borrow in order to fund its ordinary trading activities. For the first part of the period the company would have suffered a compound interest disadvantage in the sense that it would have had to pay to the Revenue an amount of money (£1m in my example) which it would otherwise have been able to lay out at compound interest. For the second part of the period the company would be suffering a compound interest disadvantage in the different sense that, for continuing trading purposes, it would have had to borrow at compound interest a sum of money (for example £1m) which it would not have needed to borrow if it had not previously had to pay the ACT to the Revenue.

There is one detailed point which I wish to make before I move on. In all of the foregoing discussion I have assumed a conventional rate of interest, and I have also assumed, tacitly or expressly, that it would be the same rate in all cases. It is, however, almost invariably the case that the interest rate which a company can obtain on money deposited by it is lower than the interest rate which it will have to pay on money borrowed by it. It is a commonplace that financial institutions like banks always seek to achieve a margin through charging borrowers interest at higher rates than they pay to depositors. It would be possible to calculate restitution or compensation on a conventional basis, but to adopt one conventional rate for companies which (like Sempra) were in a net borrowing position when they made the payments of ACT, and a different (and lower) conventional rate for companies which were in a cash surplus position. This point was briefly alluded to in argument by Mr Glick, but it was not explored. I will be willing to hear further submissions on it if either counsel wishes me to do so. My present inclination, however, is that the same conventional rate should be taken for all cases. As long as it is applied on a compound basis it will produce full restitution or compensation taking the claimant companies collectively, and for pragmatic reasons I would be content with that. I have said in paragraphs 18 and 19 above that, in my judgment, all calculations should be made on a conventional basis and not by reference to what any particular company says it would have done if it had not had to pay the ACT. For essentially the same reasons I am unattracted by the idea of having two conventional rates of interest, not one.

There is one other important matter which I must consider. I need to evaluate an argument advanced by Mr Glick. He contrasts the circumstances of a taxpayer company like Sempra, which in the event was able to set off against MCT all the ACT which it paid (‘utilised ACT’), with other companies which paid ACT but did not have sufficient liabilities to MCT to be able to set off some or all of the ACT (‘unutilised ACT’). Assume that a company with unutilised ACT was a subsidiary of a parent company established in a member state of the European Union other than the United Kingdom. The decision of the CJEC in Metallgesellschaft/Hoechst did not specifically deal with such a case, because both of the companies which were parties to that reference to the CJEC had utilised all of the ACT which they had paid. (One of them was, of course, Sempra, then called Metallgesellschaft.) However, it has never been disputed by the Revenue (and I agree) that an inevitable effect of the court’s decision was that a subsidiary with unutilised ACT was entitled to have it repaid by the Revenue. In addition, and importantly for purposes of the present argument, such a subsidiary was also entitled to receive interest over the period from the payment by it of the ACT until the repayment. But in such a case the view of the Revenue, which has not so far as I am aware been disputed (at least not yet) is that the interest which the company receives (or which it would be held to be entitled to receive if the issue was taken to court rather than being conceded by the Revenue) is regulated by s.35A of the Supreme Court Act 1981, and is therefore simple interest, not compound interest:

I do not agree with the argument. Let me assume for the moment that the Revenue are right that the interest which is payable to a company which recovers unutilised ACT is simple interest. The reason why it is simple interest is because a United Kingdom statute so provides. However, the United Kingdom statute does not apply at all to the case (which is Sempra’s case) where there is no unutilised ACT to recover. In that case the entitlement to interest arises solely by virtue of Community law and the CJEC’s decision. If, as I believe, the true effect of Community law and the CJEC’s decision is that interest for the premature tax payment period should be compound interest, I do not think that it would be right to depart from that result in order to achieve consistency with the treatment of other taxpayer companies under a national statute which does not apply to the particular case before the court.

If there might be a difference between the nature of the interest received in the two different situations (compound interest in the case of utilised ACT, and simple interest in the case of unutilised ACT), why should they be assimilated so that the treatment of interest on unutilised ACT is caused to prevail over what would otherwise be the treatment of interest on utilised ACT? I see no reason why they should. Of the two treatments, which one is out of line with the true merits of the situation? The answer, in my view, is the treatment of interest on unutilised ACT (if that treatment is indeed restricted to simple interest by virtue of s.35A of the Supreme Court Act 1981). Compound interest gives full restitution or compensation; simple interest gives partial restitution or compensation. I note in passing that the Law Commission has recently recommended that s.35A should be changed so as to permit compound interest to be awarded.

There are other points to be made in relation to this particular argument advanced by Mr Glick. I believe that, in the contemplation of the ACT system, the normal case which was visualised was one where ACT would be utilised. Although companies with large amounts of unutilised ACT (sometimes referred to as surplus ACT mountains) did develop over the years while the ACT system was in force, they would have been seen by the original architects of the system as the exception and not the norm. I am unattracted by the argument that what is in my view the clearly preferable decision for the normal case (a company like Sempra which does utilise its ACT) should be departed from in order to achieve consistency with the exceptional case (a company which does not utilise its ACT).

Further, although I said in paragraph 33 above that so far as I am aware no-one has yet challenged the Revenue’s view that, in the case of unutilised ACT which is repaid by reason of the CJEC decision, the interest payable for the intervening period must be simple interest because the right to it derives from s.35A, I do not wish to be understood as necessarily accepting that that is correct. I can see substantial arguments that, in the case of unutilised ACT as well as in the case of utilised ACT, the entitlement to interest for the intervening period does not depend on s.35A, but rather depends on the principles of Community law explained by the CJEC in Metallgesellschaft/Hoechst .

Suppose that s.35A had never been enacted, and suppose also that the CJEC had dealt, not just with the two cases before it, but also with a third case in which the subsidiary had paid ACT, some of all of which it had not utilised. What would the CJEC have decided as respects that third case? I am sure that it would have decided that the unutilised ACT should be repaid, and I am equally confident that it would have directed that the Revenue must also pay interest (or an amount equal to interest) for the period between the payment of the ACT and the repayment of it. If interest (or an amount equal to it) had to be paid where the ACT had already been utilised, it follows as the night follows the day that interest (or an amount equal to it) also had to be paid where the ACT had not yet been utilised. The CJEC would have required that result regardless of whether United Kingdom law did or did not include a provision like s.35A.

I also note that there are discretionary elements in the wording of s.35A which would not easily reconcile with the CJEC’s decision that restitution or compensation must be paid. Under s.35A interest ‘may’ be included in a sum for which judgment is given. It may be on the whole or ‘any part of’ the debt. And it may be for the whole or ‘any part of’ the period from accrual of the cause of action until judgment. The CJEC’s requirement that there must be restitution or compensation is more mandatory than the impression given by the wording of the United Kingdom statute.

In the circumstances I have my doubts even about the starting point for this argument which Mr Glick advanced.

There is one last matter to be considered in connection with the issue of whether restitution or compensation should be calculated by compound or simple interest. It was submitted to me by Mr Glick that the case law of the Court of First Instance indicated that under Community law interest awarded by way of compensation is always simple interest. The case referred to was Corus UK Ltd v The Commission of the European Communities (Case T- 171/99) [2002] 1 WLR 970. In my opinion, however, no general principle can be extracted from it.

The company had been subjected to a fine by the Commission. It paid the fine, but it appealed to the Court of First Instance. The appeal was partially successful, and part of the fine was repaid. The company argued that the repayment should be made with interest, and the court agreed. There was some discussion of the difference between compound and simple interest. The court said that the generally accepted principle in the domestic laws of member states was that the recovery of interest was dependent on the right to recover the principal sum (judgment paragraph 56), and that interest would normally be awarded without compounding (paragraph 60). However, those observations do not seem to me to be a guide for the present case. They were a generalised description of the domestic laws of member states, whereas the present case depends on Community law. They assumed a case where there was a principal sum owing, whereas in the present case there was not. Further, the Commission had calculated what amount the excessive part of the fine had earned for it, and had made that calculation on a compound basis. The actual amount of interest which the court awarded was the amount so calculated, so that in the event it was compound interest. In the circumstances I would not cite the case as supporting my view that the restitution or compensation in this case should be computed on a compound basis, but I certainly do not agree that it supports a conclusion that the computation should be made on a simple basis. It is a very special case, and I do not think that it assists on the question which I have to decide in the different circumstances of the present case.

For all of the foregoing reasons, on the main point of principle which I am asked to decide in this judgment my conclusion is that the restitution or compensation payable to Sempra and to other companies in a comparable situation should be calculated by reference to a conventional interest rate applied by way of compound interest, not by way of simple interest, over the premature tax payment period. There are other details which will still have to be resolved, including the rate of interest and the intervals between the rests at which compound interest is calculated. However, I am not asked to decide those questions now. I imagine that the Revenue will hope to resolve them by negotiation with the various claimant companies, and if that happens the questions may never arise for formal decision by the court.

Sempra utilised the four sample payments of ACT, by setting the amounts of ACT against MCT, on five different dates, the earliest of which was 1 July 1990 and the latest of which was 1 July 1996. One example is that on 1 July 1994 it set off (that is, it utilised) £369,143.12 of ACT which it had paid on 10 July 1989. The effect of the decisions of the CJEC and of myself is that, as respects that ACT, it is entitled to judgment for an amount computed by reference to compound interest over the period of just under five years between the payment in 1989 and the utilisation in 1994. Although the judgment will be computed by reference to interest, it will be judgment for what Mr Rabinowitz described as ‘the primary loss’ (see paragraph 16(v) above). At this stage in this judgment I am no longer concerned with that element in Sempra’s claim, but with a further claim which Sempra makes for interest over what I might call the post-utilisation period. In the particular instance which I have just described the post-utilisation period is the period from 1 July 1994, when the ACT was utilised and the cause of action crystallised, until judgment in the present case. It is accepted that Sempra is entitled to interest for that period, but is it entitled to it on a compound basis or on a simple basis? Similar questions arise as respects the post-utilisation periods applicable to the other three sample payments of ACT.

In my opinion the element of interest in what Sempra will recover for the post-utilisation periods falls to be calculated on a simple basis. The reason is that it is interest payable by virtue of s.35A of the Supreme Court Act 1981, and therefore it can only be simple interest, because that is what s.35A provides. I have quoted the relevant statutory wording in paragraph 33 above. In paragraph 86 of the CJEC’s judgment in Metallgesellschaft/Hoechst the court said:

However, the claim for interest attributable to the post-utilisation period is different. It is truly ancillary: it is a claim for interest on the primary loss, and it is a claim for interest over the period between the time when the primary loss accrued and the time when judgment is given for the primary loss to be paid to Sempra by way of restitution or compensation. It may seem anomalous that Sempra’s recovery for the period from payment of the ACT until utilisation is in an amount computed by reference to compound interest, but its recovery for the period from utilisation until judgment is in an amount computed by reference to simple interest. However, the anomaly lies in the wording of s.35A. That section is the source of the claim for interest for the post-utilisation period, and it plainly restricts the interest to simple interest. I have already mentioned that the Law Commission has recommended a change in the law in this respect, but the change has not happened yet, and as the law now stands the interest payable to Sempra for the post-utilisation period can only be simple interest.

I believe that I have now dealt with all of the issues which require to be decided at this stage, and I have no other observations to make.