1. 

The present references for a preliminary ruling concern the sugar marketing years between 2001 and 2006, when there was a surplus of production over consumption in the European Union (‘EU’) and prices were significantly higher than on the world market. As one consequence, quotas were allocated to producers. Production within certain quotas could be exported with refunds financed by production levies. The calculation of the levies involved multiplying the ‘exportable surplus’ by the ‘average loss per tonne’ for each marketing year. That average loss was obtained by dividing the ‘total amount of refunds’ by the total tonnage of ‘export obligations to be fulfilled’ in the marketing year.

2. 

The main issue in these cases is the definition of the ‘total amount of refunds’ in that context. Uncertainty derived from the fact that, for some quantities contained in exported processed products, the export refunds for which they qualified were neither claimed nor paid.

3. 

In regulations determining the production levies for the marketing years between 2003 and 2006, the Commission included those quantities in the ‘exportable surplus’ but not in the ‘export obligations to be fulfilled’. Requested to rule on the validity of that calculation, the Court found in 2008, in Zuckerfabrik Jülich and Others (‘Jülich I’), ( 2 ) that all quantities exported should have been taken into account in both cases, regardless of whether refunds had been paid or not, and that the regulations in question were invalid for that reason. It did not, however, rule whether the ‘total amount of refunds’ should likewise have included all refunds available, regardless of whether they had been paid or not, or only refunds actually paid.

4. 

In 2009, the Commission adopted a new regulation correcting those which had been declared invalid. In its calculations, it included in the ‘total amount of refunds’ all refunds available, regardless of whether they had been paid or not. The recalculated levies differ little from those originally fixed but are higher than they would have been if only refunds actually paid had been included in the ‘total amount of refunds’.

5. 

A number of producers have challenged both the calculation and the legal basis for adopting the new regulation, and three national courts have asked the Court for a ruling on its validity. One of those courts also asks, in relation to repayments which are known to be due, what date should be used for applying currency exchange rates and whether interest can be awarded.

6.

The European Communities’ own resources were regulated at the material time by Council Decision 2000/597 ( 3 ) (‘the own resources decision’) and Council Regulation No 1150/2000 ( 4 ) (‘the own resources regulation’).

7.

Under Article 2(1)(a) of the own resources decision, revenue from, inter alia, ‘contributions and other duties provided for within the framework of the common organisation of the markets in sugar’ was to constitute own resources entered in the budget of the EU.

8.

Article 2(3) provided: ‘Member States shall retain, by way of collection costs, 25% of the amounts referred to in paragraph 1(a) …’.

9.

Article 6 stated: ‘The revenue referred to in Article 2 shall be used without distinction to finance all expenditure entered in the budget. …’

10.

Article 8(1) of the same decision provided: ‘The Communities’ own resources referred to in Article 2(1)(a) … shall be collected by the Member States in accordance with the national provisions imposed by law, regulation or administrative action …’

11.

Under Article 9(1) of the own resources regulation, each Member State was to open an account in the name of the Commission with its Treasury or other appointed body, and to credit own resources to that account.

12.

Article 11(1) of the same regulation stated: ‘Any delay in making the entry in the account referred to in Article 9(1) shall give rise to the payment of interest by the Member State concerned.’ ( 5 )

13.

Council Regulation (EC) No 1260/2001 ( 6 ) (‘the basic regulation’) regulated the common organisation of the markets in the sugar sector from the 2001/02 to the 2005/06 marketing years. It was repealed and replaced with effect from 1 July 2006. ( 7 )

14.

The preamble to that regulation stated, inter alia:

15.

The regulation thus provided for producers to meet the cost to the EU of disposing of surplus production by way of production levies.

16.

Article 7(3) granted production refunds on sugar originating, or in free circulation, in Member States, used to manufacture certain products of the chemical industry. The amount of the refund was to be fixed taking account of the cost of supply on the world market.

17.

Articles 27 to 29 provided for export refunds, reflecting the difference between world prices and EU prices, for specified sugar products. Conversely, Article 33 provided for an export levy where the world price of sugar was higher than the intervention price. In practice, world prices were always lower, so that all exports of A and B sugar qualified for refunds and no export levies were charged.

18.

With regard to the calculation of the production levies, Article 15 provided, ( 10 ) in particular:

‘1.   Before the end of each marketing year, the following shall be recorded:

2.   Before the end of the 2005/06 marketing year …, the following shall be recorded cumulatively for the 2001/02 to 2005/06 marketing years:

3.   … should the figures recorded under paragraph 1 and adjusted under paragraph 2 result in a foreseeable overall loss, then that loss shall be divided by the estimated production of A and B sugar … attributable to the current marketing year. The resulting amount shall be charged to manufacturers as a basic production levy on their production of A and B sugar ...

…

4.   Should the maximum permitted basic production levy not fully cover the overall loss referred to in the first subparagraph of paragraph 3, the balance not covered shall be divided by the estimated production of B sugar … attributable to the marketing year in question. The resulting amount shall be charged to manufacturers as a B levy on their production of B sugar ...

…

5.   Where the figures recorded under paragraph 1 suggest that the foreseeable overall loss for the current marketing year is unlikely to be covered by the expected proceeds from the levies because of the ceilings on the basic production levy and the B levy fixed in paragraphs 3 and 4, then the maximum percentage referred to in the first indent of paragraph 4 shall be adjusted to the extent necessary to cover the overall loss, without exceeding 37.5%.

…

6.   All the losses resulting from the grant of production refunds under Article 7(3) shall be taken into account when calculating the overall loss referred to in paragraph 1(e).

7.   The levies referred to in this Article shall be collected by the Member States.

8.   Detailed rules for applying this Article shall be adopted …, and shall cover in particular:

19.

Article 16 provided that an additional levy was to be charged if the levies provided for under Article 15(3), (4) and (5) did not fully cover the loss for the year in question. Article 16(5) provided for detailed rules to be adopted for the application of that additional levy.

20.

On the basis of, inter alia, Articles 15(8) and 16(5) of the basic regulation, the Commission adopted Regulation No 314/2002 ( 14 ) (‘the implementing regulation’), which provided for (among other things) the determination of the sugar quantities disposed of for consumption in the Community within the meaning of Article 15(1)(b) and (2)(a) of the basic regulation and the definition of export obligations to be fulfilled within the meaning of Article 15(1)(d) of the basic regulation.

21.

Under Article 6(4) of the implementing regulation as amended, quantities disposed of for consumption in the Community were to be calculated as (i) total quantities stored at the beginning of the marketing year, quantities produced under A and B quotas, quantities imported in the natural state and quantities contained in imported processed products, minus (ii) total quantities exported in the natural state, quantities contained in exported processed products, quantities stored at the end of the marketing year and quantities for which production refund certificates pursuant to Article 7(3) of the basic regulation had been issued – the latter quantities being, in effect, those used in the chemical industry.

22.

Article 6(5) of the implementing regulation defined ‘export obligations to be fulfilled during the current marketing year’ as, essentially: all quantities to be exported in the natural state with export refunds or levies fixed (a) by tenders opened in respect of that marketing year or (b) on the basis of export licences issued during that marketing year; (c) all foreseeable exports in the form of processed products with export refunds or levies fixed for that purpose during that marketing year; (d) quantities for which production refund certificates pursuant to Article 7(3) of the basic regulation had been issued during the marketing year; and (e) food aid.

23.

Articles 6 and 7 of the implementing regulation also made provision for advance payments to be made on production levies before the end of the marketing year, on the basis of estimates. That was explained in recital 7 in the preamble: ‘The production levies provided for in Article 15 of [the basic regulation] cannot be fixed until after the end of the marketing year in view of the fact that a large proportion of the export commitments are made in the second half of the marketing year and that the data serving to establish the production levies are available only at that point. Therefore, in order to implement as soon as possible the financial responsibility of producers, payment in advance of levies calculated on the basis of estimates should be made well before the end of the marketing year. … The levies should not be fixed or collected until information which is as precise as possible, in particular on consumption, is to hand.’

24.

The present references for a preliminary ruling concern the starting point for the calculation of the production levies, namely, the ‘overall loss’ for each marketing year.

25.

Pursuant to the provisions cited above, that ‘overall loss’ was, for each year concerned, the ‘exportable surplus’ multiplied by the ‘average loss per tonne’.

26.

The ‘exportable surplus’ was, essentially, A and B production minus EU consumption. It represented, therefore, only surplus A and B production during the year. In addition, not only did C sugar (EU production outside the A and B quotas) have to be exported without refunds, but stocks of A and B sugar, together with certain categories of imported sugar, could be exported with refunds.

27.

EU consumption was calculated by subtracting the total quantity known not to have been consumed (essentially: exports, quantities used in the chemical industry and stocks of A and B sugar at the end of the marketing year) from the total quantity known to have been available for consumption (essentially: stocks of A and B sugar at the beginning of the marketing year, A and B production during the year and imports).

28.

The ‘average loss per tonne’ was obtained by dividing the ‘total amount of refunds’ by the ‘total tonnage of export obligations’. It was thus a fraction, with ‘total amount of refunds’ as the numerator and ‘total tonnage of export obligations’ as the denominator.

29.

The main issue in these cases is whether the numerator (‘total amount of refunds’) in that fraction should include refunds to which exports were entitled but which were not claimed; it is established that the denominator (‘total tonnage of export obligations’) includes all quantities entitled to a refund, whether the refund was claimed or not.

30.

From 2003, in fixing the levies each year, the Commission interpreted ‘export obligations’ as comprising only quantities in respect of which refunds were in fact paid (and not those in respect of which refunds were available but not paid). That reduced the denominator in the fraction described in point 28 above, thereby increasing the ‘average loss per tonne’. At the same time, however, the Commission deducted all quantities contained in exported processed products (regardless of whether refunds were in fact paid) from the quantity available for consumption in the EU. As a result, the calculations set out in points 26 and 27 above produced smaller ‘forecast quantities disposed of for consumption in the Community’, and a larger ‘exportable surplus’, than would otherwise have been the case. When the ‘exportable surplus’ was then multiplied by the ‘average loss per tonne’, the ‘overall loss’ was therefore greater, which in turn increased the production levies. The effect was, apparently, significant, since in many instances sugar producers did not claim export refunds in respect of sugar incorporated in exported processed products.

31.

The production levies for the 2001/02, 2002/03, 2003/04 and 2004/05 marketing years were set by Regulations Nos 1837/2002, 1762/2003, 1775/2004 and 1686/2005 ( 15 ) respectively. Challenges to the calculation of those levies led to a number of preliminary references to the Court of Justice.

32.

On 5 May 2008, the Court gave judgment in Jülich I. It held that, under Article 15(1)(c) and (d) of the basic regulation, all quantities of exported products which fell under that article were to be taken into account for the purpose of calculating both the ‘exportable surplus’ (when they were to be subtracted from consumption) and the ‘average loss per tonne’, regardless of whether refunds had actually been paid or not.

33.

The ‘exportable surplus’, the Court noted, was the difference between EU production of A and B sugar and EU sugar consumption. Such consumption was therefore not intended to include any exported products, whether they had benefited from export refunds or not. If quantities exported without refunds were included in EU consumption, it would be overestimated and the exportable surplus underestimated, entailing a risk that the objective of ensuring that the producers themselves meet the costs of disposing of surpluses would not be achieved. ( 16 )

34.

As regards the ‘average loss per tonne’, the Court considered that the concept of ‘export obligations to be fulfilled’ (the denominator of the fraction referred to in point 28 above, defined in Article 15(1)(d) of the basic regulation) could not be construed consistently with Article 22(1) of that regulation (under which exports were subject to the presentation of an export licence, itself subject to the provision of a guarantee that the export would take place during the term of validity of the licence ( 17 )) as limited to those in respect of which export refunds were actually paid. It dismissed the Commission’s argument that such a limitation, by increasing the levels of production levies, could discourage the creation of surpluses, stressing that the calculation should not go beyond the objective of equitable self-financing by producers. ( 18 )

35.

The Court therefore held that Regulations Nos 1762/2003 and 1775/2004 were invalid, in so far as they took into account, for the purpose of calculating the ‘average loss per tonne’, only those exports which had attracted refunds. However, examination of Regulation No 1837/2002 had not disclosed the existence of any factors such as to affect its validity, as it took account of the total quantities exported in the form of processed products, regardless of whether those exports had actually benefited from refunds or not.

36.

The Court did not rule, nor had it been asked to rule, whether the ‘total amount of refunds’ should also have been calculated taking account of the total quantities exported in the form of processed products, regardless of whether refunds had actually been paid.

37.

Then, in SAFBA and Others (‘SAFBA’), ( 19 ) the Court ruled, on the one hand, that Article 6(4) of the implementing regulation was not invalid in so far as it did not, with regard to calculation of the production levies, provide for exclusion from the ‘exportable surplus’ of quantities contained in exported processed products for which no export refund had been granted and, on the other hand, that Regulation No 1686/2005 was invalid in so far as it did not respect the method of calculation approved in Jülich I.

38.

The Commission did not immediately adopt any new measure to comply with the Court’s rulings. Proceedings to recover levies paid under the regulations which had been declared invalid went ahead before various national courts. In a number of cases, orders were made for repayment of amounts by which such levies were judged to have been overcharged. The Commission, aware of those orders, considered that national courts and authorities were drawing incorrect conclusions from the judgment in Jülich I. In January 2009, it informed Member States that it was drafting a new regulation which would indicate the correct amounts of levy for the marketing years in question, and would inform them in due course as to the action to be taken. That draft was discussed in the relevant management committee in September and October 2009. However, the committee was reluctant to accept the calculations used by the Commission, and many Member States wished to use other figures, proposed by the presidency.

39.

Essentially, the Commission calculated ‘total refunds’ by multiplying the figure for all relevant exports, whether a refund had been paid or not, by the average amount of refund payable for such exports; the numerator and the denominator of the fraction referred to in point 28 above thus both reflected the total amounts of exports carrying a right to a refund. The presidency’s calculation, by contrast, included in the numerator only the actual amount of refunds paid, while the denominator remained the same. Under that calculation, the ‘overall loss’ and, consequently, the level of the production levies were both lower. ( 20 )

40.

However, the Council was unable to adopt a different regulation within the requisite time-limit. ( 21 ) Consequently, the Commission adopted the draft in the form proposed, as Regulation No 1193/2009 ( 22 ) (‘the contested regulation’), fixing new production levies both for the years concerned by the regulations which had been declared invalid and for the 2005/06 marketing year, for which the same method of calculation had been used. ( 23 )

41.

The legal bases cited for the contested regulation in its preamble were the EC Treaty and the basic regulation, in particular the first indent of Article 15(8) and Article 16(5) thereof.

42.

Recitals 5 and 6 in the preamble to the contested regulation state:

43.

Articles 1 to 4 of the contested regulation fix new levies to replace those previously set in the relevant regulations. Of the 21 levies concerned, 12 are unchanged, 6 are slightly lower and 3 slightly higher than the amounts previously calculated.

44.

Under Article 6 of the contested regulation, Articles 1 to 4 apply from the dates of entry into force of the respective regulations whose provisions they replace.

45.

Zuckerfabrik Jülich AG (‘Jülich’) is a German sugar producer. Its challenge to the levies for the marketing years 2002/03, 2004/05 and 2005/06 on the basis of Regulations Nos 1762/2003, 1686/2005 and 164/2007 led to the judgment in Jülich I. Following the adoption of the contested regulation, the national agency established new production levies for the marketing years in question. Jülich then challenged the notices determining those levies before the Finanzgericht (Finance Court) Düsseldorf, arguing that the contested regulation was invalid.

46.

That court notes that the Commission recalculated the total amount of refunds for the marketing years in issue by including fictitious refunds for exports in respect of which no refund was paid. It questions the compatibility of such a revision, in so far as it goes beyond the conditions imposed by the judgment in Jülich I and is applicable to marketing years which are now closed, with the EU principle of non-retroactivity.

47.

The Finanzgericht Düsseldorf has therefore asked the Court for a preliminary ruling on the question: ‘Is [the contested regulation] valid?’

48.

British Sugar plc (‘British Sugar’), a manufacturer of sugar products, seeks repayment, with interest, of the amount of production levies which it claims it was overcharged by the competent national agency during the marketing years 2002/03 to 2005/06. On the basis of Jülich I, it calculates the amount as around EUR 12531000, before interest. It claims, first, that the contested regulation has the same fundamental defect as Regulations Nos 1762/2003, 1775/2004 and 1686/2005. The method of calculation, in particular as regards the ‘average loss per tonne’, is based on hypothetical losses reflecting export refunds which were notionally available but never actually paid. The ‘overall loss’ is therefore inflated. Second, the contested regulation is invalid in so far as it requires non-euro repayments to be made at the exchange rates applicable at the time the levy was originally calculated rather than the time of repayment. Finally, British Sugar claims interest on the sums to be repaid. It has therefore brought proceedings before the High Court of Justice of England and Wales.

49.

The national agency contends that the contested regulation now sets out the legal formula for the calculation of the repayment due to British Sugar and is legally binding. Under that calculation, the amount due is GBP 366590,79. The agency argues that the Court endorsed the methodology of Regulation No 1837/2002, in so far as it did not find that regulation invalid. The contested regulation follows the same scheme and is therefore equally valid. The appropriate exchange rate is that which applied when the original production levies were calculated. Finally, British Sugar should not be awarded interest. Any repayment will be matched by a payment from the Commission to the agency under the EU’s own resources system. There is no legal basis in the own resources legislation for Member States to recover interest from the Commission, and the same principle must apply to any repayment to British Sugar.

50.

The High Court therefore seeks a preliminary ruling on the following questions:

51.

Believing that it had been overcharged for levies under Regulation No 1686/2005, the French sugar producer Tereos applied on 2 May 2007 to the relevant national agency for partial repayment. Receiving no response, it brought an action before the Tribunal de grande instance (Regional Court), Nanterre, arguing that Regulation No 1686/2005 was invalid and seeking repayment of EUR 11 600 782. Following the adoption of the contested regulation, Tereos asked the national court to seek a preliminary ruling on its validity in the light of Article 15 of the basic regulation and to order repayment to Tereos of the sum claimed, together with interest.

52.

The national court notes that, in the contested regulation, the Commission did not recalculate the levy by strictly applying the method of calculation taken from Article 15 of the basic regulation as interpreted by the Court in Jülich I but used the method already applied for the marketing year 2001/02, the Court having stated that the examination of Regulation No 1837/2002 had not disclosed the existence of any factors such as to affect its validity.

53.

It has therefore referred the following questions to the Court of Justice for a preliminary ruling:

54.

The three references for a preliminary ruling have been joined by the Court for the purposes of the oral procedure and the judgment.

55.

Written observations have been submitted by the claimants in the three sets of proceedings, by the German, Spanish, French, Lithuanian, Austrian and United Kingdom Governments, and by the Commission. At the hearing on 30 June 2011, submissions were made by the three claimants, the German and French Governments, and the Commission. All those who have made submissions, apart from the Commission and, to a certain extent, the United Kingdom, consider that the contested regulation is invalid.

56.

In addition, Jülich, British Sugar and certain other sugar producers have brought actions before the General Court seeking annulment of the contested regulation, and the Republic of Poland has brought an action seeking annulment of Article 3 of that regulation. ( 24 ) The procedure in all those cases has been suspended pending the outcome of the present proceedings.

57.

The central issue in all three cases is whether the Commission’s calculation in the contested regulation was valid, in so far as it included in the ‘total amount of refunds’ refunds to which exporters were entitled, but which were not in fact claimed.

58.

However, the contested regulation is also challenged, in particular, as regards its formal validity, on the ground that the basic regulation, cited as its legal basis, had been repealed before the contested regulation was adopted. That issue, it seems to me, should logically be addressed first. I shall then consider the validity of the calculation, followed by a number of further points related to its compatibility with the judgment in Jülich I.

59.

In addition, two further issues are raised in the British Sugar reference: the correct reference date for determining exchange rates for sums to be repaid in currencies other than the euro, and the availability of interest on such sums. I shall deal with them last, beginning with the availability of interest, an issue which concerns all Member States and which seems to precede that arising from fluctuating exchange rates.

60.

Articles 15(8) and 16(5) of the basic regulation provided for the adoption of detailed rules for applying those two articles, in accordance with the procedure referred to in Article 42(2) thereof, which, in turn, referred to the comitology decision. ( 25 ) In short, they gave the Commission a legal basis for adopting regulations fixing the production levies for each marketing year in accordance with the management committee procedure.

61.

However, Regulation No 318/2006 ( 26 ) repealed the basic regulation with effect from 1 July 2006.

62.

Consequently, the latter was no longer in force when the contested regulation was adopted on 3 November 2009. Yet the legal basis cited is, first, the EC Treaty and, second, the basic regulation ‘and in particular the first indent of Article 15(8) and Article 16(5) thereof’.

63.

The Commission acknowledges that on 3 November 2009 the basic regulation had been repealed and replaced by a new instrument which set up a different regime for the sugar sector and provided no legal basis for an act fixing levies under the previous regime. It submits, however, that it was required by Article 233 EC (now Article 266 TFEU) to take the measures necessary to comply with the rulings in Jülich I and SAFBA. Those rulings, the Commission contends, thus had the effect of ‘overriding the apparent absence of legal base’ and ‘had given new life, as it were, to Articles 15(8) and 16(5) of the basic regulation, to the extent necessary to enable it to rectify the illegalities found by the Court’.

64.

I cannot agree. The binding nature of any act intended to have legal effects must be derived from a provision of EU law which prescribes the legal form to be taken by that act and which must be expressly indicated therein as its legal basis. ( 27 ) The provision which forms the legal basis of an act and empowers the Union institution to adopt the act in question must be in force at the time when the act is adopted. ( 28 ) A provision which has been repealed does not meet those criteria. Nor can a judgment of the Court resuscitate a repealed measure other than by declaring invalid the measure which repealed it.

65.

That does not mean, however, that the Commission had no means of taking – as it was required to do by Article 233 EC – the measures necessary to comply with the rulings in Jülich I and SAFBA. At least two routes were open to it.

66.

First, the Commission could have asked the Council to grant it competence to adopt a regulation fixing new levies for the marketing years in question. I am unconvinced by the Commission’s submission that such a request would have been pointless because the two institutions differed as to the correct calculation. The measure required would have concerned only the Commission’s formal competence to fix the levies anew, not the way in which those levies were calculated.

67.

Second, as pointed out by the claimants, adoption of a regulation was just one way of achieving compliance. ( 29 ) Repayments are a matter, first, between the sugar producers and the national agencies (with adjudication, where appropriate, by the national courts). ( 30 ) Clearly, it is desirable, following Jülich I, for levies to be adjusted consistently throughout the EU. To that end, the Commission could have issued guidance as to the appropriate calculations. Any question as to that guidance could have been referred to this Court for a preliminary ruling, much as has happened in the present cases (although it might not have been possible to bring actions for annulment before the General Court).

68.

However, in so far as the contested regulation also cited the EC Treaty as its legal basis, it has been queried whether Articles 37 or 233 EC (now, respectively, Articles 43 and 266 TFEU) might not have conferred competence on the Commission. I think not.

69.

First, the Joint Practical Guide for the drafting of Community legislation ( 31 ) indicates: ‘If the direct legal basis of the act is a Treaty provision, the general citation is accompanied by the words “, and in particular”, followed by the relevant article … If, by contrast, the direct legal basis of the act is to be found in secondary legislation …, the particular act concerned is cited in a second citation, with the relevant article, preceded by the words “, and in particular”.’ On the assumption that the Commission was following its own drafting guidelines, therefore, there was no reliance on any particular Treaty provision as a direct legal basis for the contested regulation. Only Articles 15(8) and 16(5) of the basic regulation were relied on.

70.

Second, Article 37 EC conferred no legislative competence on the Commission in the agricultural field other than to submit proposals to the Council. Any measure adopted on that basis would have had to be adopted by the Council.

71.

Third, Article 233 EC, while requiring compliance with the judgment in Jülich I, did not itself provide the Commission with a legal basis for adopting a regulation for which no other authority existed in the Treaty or in secondary legislation – at least when, as in the present case, alternative modes of compliance were available.

72.

I am thus of the view that the contested regulation is formally invalid, lacking a legal basis in force at the time of its adoption.

73.

That does not, however, mean that the Court should refrain from examining the conformity with the basic regulation of the calculation on which the contested regulation was based. Even if the latter is formally invalid, the production levies must still be adjusted to take account of Jülich I, with or without further guidance from the Commission, by national agencies to which claims have been addressed and by national courts seised of disputes. That adjustment must conform to the requirements of the basic regulation, as interpreted by the Court, and it would not be sensible to delay that interpretation until it has to deal with a future Jülich III case.

74.

The essence of the question is simple: in carrying out the calculation required by Article 15(1)(d) of the basic regulation, should the Commission have included, in the ‘total amount of refunds’, refunds which could have been claimed in respect of export obligations but which were never in fact claimed or paid out?

75.

Unfortunately, as the present cases have shown, the phrase used (‘total amount of refunds’) could bear either meaning: ‘total amount of refunds claimable’ or ‘total amount of refunds paid’.

76.

The most obvious inference from that lack of precision is that, when the provision was first drawn up, the legislature did not recognise that there could be any ambiguity. I find it difficult to imagine that the legislature, drafting and debating the method for calculating the production levies, could have been aware of alternative interpretations of ‘total amount of refunds’, leading to two different results when the calculation was carried out, without specifying which of them was to be used, or at least providing for the choice to be made in implementing legislation.

77.

One plausible explanation is that the legislature presumed that all refunds claimable would be in fact claimed and paid out. When a word like ‘refunds’ is used without further qualification, and unless the context dictates otherwise, the only natural interpretation is that it refers to refunds actually paid out.

78.

It seems to me, therefore, that the starting-point must be a strong, though not necessarily irrebuttable, presumption that ‘total amount of refunds’ means only refunds actually paid out, to the exclusion of purely potential refunds never claimed or paid out.

79.

However, the Commission has put forward a number of reasons for interpreting the phrase to include all potential refunds, and those submissions must be examined. In summary: (i) that was the interpretation on which Regulation No 1837/2002, not found to be invalid by the Court in Jülich I, was based; (ii) it is consistent with the inclusion of export obligations on which no refunds were claimed in the denominator of the fraction; (iii) it is consistent with the references to forecast quantities and amounts in Article 15 of the basic regulation; and (iv) it ensures that producers cover the costs of export refunds. I shall examine those submissions in turn.

80.

The Commission argues that it merely used the same basis of calculation for the contested regulation as for Regulation No 1837/2002, examined by the Court in Jülich I and not found to be invalid. To ensure consistency over the five marketing years covered by the basic regulation, it had to revert to that method in the contested regulation. Moreover, without any challenge from sugar producers, it had used the same basis of calculation in previous marketing years under different regulations containing substantially identical wording to that of the basic regulation at issue here.

81.

I cannot accept that justification, even if the calculations in those previous years did take ‘total amount of refunds’ and ‘total tonnage of export obligations’ to include, respectively, refunds which could have been, but were not, claimed and export obligations on which refunds could have been, but were not, claimed. ( 32 )

82.

In Jülich I, the Court was asked to interpret the basic regulation with respect to two elements in the overall calculation: the ‘exportable surplus’ and the ‘total tonnage of export obligations’ used in the further calculation of the ‘average loss per tonne’. It was not asked to rule, nor did it rule, on the correct interpretation of the other element in the latter calculation, namely, the ‘total amount of refunds’. Consequently, its finding that ‘examination of Regulation No 1837/2002 has not disclosed the existence of any factors such as to affect its validity’ can have no bearing on the way in which the Commission interpreted the ‘total amount of refunds’ in that regulation. Even if, as the Commission states, all the elements used in the calculation were presented to the Court, the questions referred and, consequently, the Court’s examination thereof were confined to certain of those elements. They did not include the element here at issue.

83.

Nor can it be of any relevance whether, under previous regulations governing the sector and containing substantially similar rules, the Commission had consistently used the same interpretation of ‘total amount of refunds’, when calculating the ‘average loss per tonne’. Even if such a basis was in fact used and not objected to, its legality must depend not on repeated use and acquiescence but on a correct interpretation of the relevant enabling legislation.

84.

I accept that, when rectifying calculations found to be incorrect by the Court, the Commission could find it reasonable to revert to a time-honoured method which had not been found incorrect. However, the fact that its interpretation of ‘total amount of refunds’ in Regulation No 1837/2002 had not previously been challenged or, consequently, censured by the Court does not mean that the alternative interpretation, used in the regulations declared invalid, was incorrect. That, too, had been neither challenged nor censured by the Court.

85.

I accept also that, all other things being equal, provisions which are substantially the same should be interpreted consistently. However, the basic regulation laid down a regime for the sugar sector which was appreciably different from both its predecessor and its successor. In the absence of continuity of context, there was no decisive need for continuity of interpretation in every regard. Moreover, as regards the five marketing years covered by the basic regulation, the Commission could – assuming that an adequate legal basis was available – have rectified consistently all five regulations fixing the production levies, if that was required by the judgment in Jülich I. It does not, however, appear to have considered that possibility.

86.

I conclude that the fact that the phrase ‘total amount of refunds’ had been previously (and even systematically) interpreted by the Commission as including refunds which could have been, but were not, claimed does not affect the way in which that phrase should have been interpreted in the context of Article 15(1)(d) of the basic regulation.

87.

The Commission submits that it is logically consistent, when determining the ‘average loss per tonne’ (the ‘total amount of refunds’ divided by the ‘total tonnage of export obligations’), to take both the numerator of the fraction (‘total amount of refunds’) and the denominator (‘total tonnage of export obligations’) as each comprising either all cases in which refunds could have been granted, whether they were in fact paid out or not, or only those cases in which refunds were in fact paid out. It would be inconsistent for the numerator to comprise only those cases in which refunds were in fact paid out while the denominator included all cases in which refunds could have been granted, whether they were paid out or not. Since, by virtue of Jülich I, the denominator must include all cases in which refunds could have been granted, the same must be true of the numerator.

88.

That superficially attractive argument is not, ultimately, convincing.

89.

If the ‘average loss per tonne’ had to be determined in relation only to the tonnage of export obligations in respect of which refunds were actually paid, then both numerator and denominator would have had to be confined to cases in which refunds were paid out. That was the approach taken in the regulations found to be invalid, but the Court ruled that the denominator must include all quantities of exported products, whether refunds had been paid or not.

90.

That being so, and unless the phrase ‘average loss per tonne’ is to be given a meaning which departs from any normal usage, ( 33 ) the total of actual losses (that is to say, of actual refunds) must be divided by the total number of tonnes actually exported. More laboriously, the amount of refund paid on each tonne exported could be calculated (being sometimes zero) and the average determined by adding all those amounts together and dividing by the number of tonnes (including those on which zero amounts were refunded). To replace those zero amounts by notional amounts seems quite the opposite of logical consistency.

91.

As counsel for Tereos put it at the hearing, if a trader offers four kilos of sugar for the price of three, the average price per kilo paid by the customer is the total divided by four, and not by three. If the total were divided by three, that would amount to disregarding the fourth, ‘free’ kilo in the denominator of the fraction. Transposing that analysis to the calculation in Article 15(1)(d) of the basic regulation, it would amount to disregarding exported quantities on which no refunds had been paid – which would be contrary to the judgment in Jülich I.

92.

The Commission did not take quite that approach in the contested regulation, but achieved much the same result by adding to the numerator. If (in the ‘four for the price of three’ example) the normal price of one kilo is added to the numerator, the result is the same as if the fourth kilo is omitted from the denominator. The Commission’s approach largely neutralised the modification of the denominator required by the judgment in Jülich I – which goes far to explain why the new levies fixed in that regulation differ little from those fixed in the previous regulations declared invalid by the Court.

93.

I can see no way in which that modification of the numerator was dictated by any consideration of internal consistency in defining the elements of the division to be performed.

94.

The Commission has drawn attention to the fact that Article 15 of the basic regulation refers repeatedly to forecast ( 34 ) quantities and amounts, and to obligations ‘to be fulfilled’. In particular, the calculation in Article 15(3), which uses figures first estimated for each marketing year, then adjusted in the light of cumulative totals for the five marketing years concerned, refers to a ‘foreseeable overall loss’ even after that adjustment. In addition, in Jülich I, the Court stated that ‘the method for calculating the estimated total loss is designed, in any event, to establish, in a forward-looking and conventional way, the losses caused by disposing of surplus Community production’. ( 35 ) The Commission infers that the amounts and quantities to be taken into account in the calculations are not intended to be actual figures as definitively established after receipt of all the data, but rather to reflect the need to anticipate expenditure and income in the relevant budget. Consequently, it reasons, the calculations are intended to take account of all refunds potentially payable, and not those actually paid, regardless of whether definitive figures are subsequently established.

95.

It is true that the basic regulation uses predictive terminology. That seems inevitable if – as the regulation prescribes – figures are to be recorded (under Article 15(1)) individually before the end of each marketing year or (under Article 15(2)) cumulatively before the end of the five-year period. It does not, however, mean that definitive figures are not to be used once they are available; indeed, Article 15(2)(a) – which affects all the calculations in Article 15(2), even though they are to be performed before the end of the five-year period – refers to the ‘definitive’ figure, and recital 14 in the preamble speaks of the ‘total loss recorded’.

96.

I note also that the contested regulation was adopted in November 2009, over three years after the end of the last marketing year concerned, and it has not been suggested that definitive figures for export refunds actually paid out were not available by then. (Indeed, the regulations corrected by the contested regulation were all adopted more than three months after the close of the relevant marketing year.) Moreover, recital 7 in the preamble to the implementing regulation, adopted by the Commission itself, indicates that the intended approach was to obtain advance payment of production levies on the basis of estimates, then to fix the levies definitively once accurate information was available.

97.

If it had been intended to confine the calculation to foreseeable export refunds, regardless of whether they were paid or not and even once definitive figures were available, I would have expected an explicit statement to that effect in the legislation. The wording as it stands, read in its context, is not indicative of such an intent.

98.

The Commission’s argument does not, therefore, in my view, rebut the presumption that ‘total amount of refunds’ means only refunds actually paid out, to the exclusion of purely potential refunds never claimed or paid out.

99.

In its observations to the Court, the Commission produced a table indicating, for each of the five marketing years governed by the basic regulation: (i) the amount of refunds granted to sugar producers, ( 36 ) (ii) the levies as calculated in the contested regulation and (iii) the levies recalculated in accordance with the method favoured by the claimants in the main proceedings. Those figures purport to show that, over the five-year period, the levies imposed by the contested regulation fell some EUR 60 million short of the refunds granted, whereas levies determined according to the claimants’ method would fall EUR 346 million short. Thus, the Commission argues, the cost of the refunds is already not covered by the levies under the contested regulation, and the claimants’ method would benefit producers even more.

100.

The Commission’s figures for the amounts of refunds granted have been hotly disputed. In particular, it is claimed that they included export refunds not actually paid out ( 37 ) and/or production refunds. At the hearing, the claimants produced alternative figures, purporting to show that, over the five-year period, the levies imposed by the contested regulation exceeded refunds granted by between EUR 325 million and EUR 338 million, whereas levies determined according to the claimants’ method would exceed refunds by between only EUR 39 million and EUR 53 million. (In addition, I find that none of the totals for refunds granted can readily be related to the figures given in the management committee documents drawn up by the Commission and the presidency – and that the latter documents appear to differ between themselves as to the amounts in question.)

101.

‘Judex non calculat’, as the agent for the German Government put it at the hearing. Presented with such a variety of figures, the Court would indeed be well advised to resist any temptation to seek an arithmetical solution.

102.

However, one point is clear from both sets of figures presented to the Court. None of the calculations proposed produces a match between export refunds and production levies. Either amount can be higher than the other and, whichever method is used, the deviation varies from year to year (for example, under both methods, production levies are lower than export refunds for the 2002/03 marketing year, and higher for 2003/04).

103.

Clearly, it would be desirable to be able to interpret the basic regulation in such a way as to keep the deviation to the minimum. However, the degree of variability in the figures presented strongly suggests that no such outcome can be guaranteed, whether the ‘total amount of refunds’ includes or excludes potential refunds not actually paid out. That may be because the overall result is influenced by other factors unrelated to the calculation in Article 15(1) of the basic regulation. Whatever the explanation, however, it does not justify distorting the parameters of the calculation to achieve a particular result.

104.

I note also, from Articles 2(3) and 6 of the own resources decision, ( 38 ) that only 75% of the amount of production levies collected is paid into the EU budget and the amount paid in is not earmarked for any particular use. Consequently, the justification for seeking a precise match between levies and refunds is considerably diminished.

105.

I propose, therefore, that the Court should consider the way in which the basic regulation is designed to ensure that production levies cover the cost of export refunds, rather than the way in which specific calculations may deviate from what may be deemed an ideal result.

106.

It is the aim of the basic regulation that ‘producers should bear full financial responsibility for the losses incurred each marketing year from disposing of that part of Community production under quota which is surplus to the Community’s internal consumption’. ( 39 ) However, that outcome is to be achieved ‘in a fair yet efficient way’ ( 40 ) and not merely with the arithmetical rigour of a simple division of losses by surplus. ( 41 )

107.

It may be helpful to view the steps in the calculation in an order different from that in Article 15(1) of the basic regulation.

108.

The ‘overall loss’ to be covered by the production levies is not the total loss incurred by the EU in respect of all the refunds paid in connection with exports during the marketing year. It is the proportion of that loss attributable to the ‘exportable surplus’. The ‘exportable surplus’ is that part of EU production within the A and B quotas which is surplus to internal consumption. Regardless of which specific quantities of sugar are actually exported, the existence of such a surplus creates a need to export and, consequently, gives rise to payment of export refunds. However, even if production did not exceed consumption, some exports would still qualify for refunds. It is therefore logical – and fully in line with the self-financing principle underpinning the basic regulation – to attribute to the exportable surplus only that part of the cost of export refunds which would not have been incurred but for the existence of that surplus, and to limit production levies to levels which cover that part alone of the total cost.

109.

The proportion of the total cost to be attributed to the exportable surplus is calculated pro rata. For the total tonnage of export obligations in a given marketing year, there is a corresponding total amount of refunds. That total amount (the numerator), divided by that total tonnage (the denominator), gives an average refund in respect of each tonne, which is then multiplied by the tonnage of the exportable surplus to obtain the amount attributable to that surplus – the ‘overall loss’ which will determine the amount of the production levies.

110.

I have explained at point 89 et seq. above why it seems to me intrinsically logical, if all the quantities exported are to be included in the denominator of the fraction, to include in the numerator only refunds actually paid out in order to determine the ‘average loss per tonne’. But is that view consistent with the nature and conception of the calculation of the ‘overall loss’ from which the production levies are determined? I believe that it is.

111.

The ‘exportable surplus’ does not account for all quantities exported with entitlement to a refund (if I understood the Commission at the hearing, it is equivalent to about half of those quantities). Internal consumption is satisfied partly by imports or by stocks carried over from the previous marketing year. Consequently, of production within the A and B quotas during any given marketing year, some quantities will be consumed within the EU, some exported and some placed in storage to be carried over to the next marketing year. The proportions will not, however, correspond to the division between internal consumption and exportable surplus in Article 15(1) of the basic regulation.

112.

The purpose of identifying the ‘exportable surplus’ as a step in determining the ‘overall loss’ is thus not to be able to estimate the amount of refund actually paid in respect of each tonne. It is rather to determine the extent to which the total amount of refunds is increased by the existence of the exportable surplus. The quantities actually exported are made up of those exported in the natural state and those exported in processed products, and the latter comprise quantities in respect of which refunds are paid and those in respect of which no refunds are paid. It seems reasonable to assume that the relevant proportions apply to all export obligations fulfilled, whether attributable to the exportable surplus or not. Those proportions should therefore be reflected in the ‘average loss per tonne’ attributed to the surplus. That outcome is achieved by excluding refunds not actually paid out from the ‘total amount of refunds’ in Article 15(1)(d) of the basic regulation, but not by including them in that amount.

113.

Again, therefore, I find no reason to depart from the presumption which I derived from the wording of the provision at point 74 et seq. above. I thus consider that the expression ‘total amount of refunds’ in Article 15(1)(d) of the basic regulation covers only refunds actually paid out, and that the contested regulation is invalid in so far as the calculation of the levies fixed for the marketing years concerned includes, in the ‘total amount of refunds’ within the meaning of that provision, refunds which could have been claimed in respect of export obligations but which were never in fact claimed or paid out.

114.

The above considerations seem sufficient to answer the questions referred to the Court as to the formal and substantive validity of the contested regulation and the correct interpretation of Article 15(1)(d) of the basic regulation. However, a number of other objections to the validity of the contested regulation have been advanced during the proceedings. I shall address them more briefly.

115.

All those objections derive, essentially, from the fact that, in the contested regulation, the Commission did not confine itself to implementing strictly the interpretation of the basic regulation given in the ruling in Jülich I. I have made it clear that I agree with the claimants in the main proceedings – and with all but one of the governments which have submitted observations – that the change in the calculation of the ‘total amount of refunds’ in the contested regulation was not only not required by that judgment but was incompatible with it in so far as it neutralised the effect of the ruling that ‘export obligations to be fulfilled’ included all quantities exported, regardless of whether refunds had been paid or not. On the basis of that incompatibility, it has been submitted that the contested regulation infringed the principles of legal certainty, non-retroactivity and protection of legitimate expectations; that it contained an inadequate statement of reasons; and that the Commission committed a misuse of powers.

116.

As regards the first three principles alleged to have been infringed, it is settled law that, although in general the principle of legal certainty precludes an EU measure from taking effect from a point in time before its publication, it may exceptionally be otherwise where the purpose to be achieved so demands and where the legitimate expectations of those concerned are duly respected. ( 42 ) Subject to those conditions, therefore, it must be accepted that, when a Commission regulation has been declared invalid by the Court on the basis of specific defects, the Commission may enact new legislation having retroactive effect to redress those defects where the situation arising from the invalidity is also incompatible with EU law and where that incompatibility cannot be put right without such new legislation.

117.

In the present instance, as I have pointed out, ( 43 ) new legislation was only one means of dealing with the situation. It is not, therefore, certain that the purpose to be achieved in fact demanded the adoption of a new regulation. Be that as it may, however, the ruling in Jülich I did not require a change to the method of calculating the ‘total amount of refunds’, whether explicitly or by implication. To the extent that that aspect of the calculation had not been the subject of any assessment by the Court, all those concerned could entertain a legitimate expectation that it would not be altered retroactively. I agree, therefore that the contested regulation infringed the principles of legal certainty, non-retroactivity and protection of legitimate expectations.

118.

As regards the Commission’s obligation to state its reasons for changing the method of calculating the ‘total amount of refunds’, it must be acknowledged that the preamble to the contested regulation does contain an explanation – though not, in my view, a valid one – for reverting to the method of calculation used in Regulation No 1837/2002. It is also clear that the Commission’s detailed calculations and the figures on which they were based were made available to the Member States in the management committee. Moreover, the sugar producers do not appear to have had any difficulty in obtaining those data in sufficient time to bring annulment actions before the General Court, and the relevant documents have been produced to this Court as annexes to a number of sets of observations. In those circumstances, I would hesitate to find the contested regulation invalid solely on the ground that it did not contain a full statement of the reasons on which it was based.

119.

Finally, the allegation of misuse of powers has been put to the Court in two forms. More generally, it is argued that the Commission’s power to adopt a new regulation did not extend to making changes to the method of calculating the ‘total amount of refunds’. Whilst I agree with that proposition, I do not consider it necessary to view it as a separate ground of invalidity. More specifically, it has been submitted that the Commission’s approach was motivated not by a concern to ensure compliance with Article 15 of the basic regulation but by the knowledge that there was insufficient provision in the own resources budget to make all the repayments that would be required if no change were made to the ‘total amount of refunds’, thus necessitating further contributions from the Member States. That, if established, would in my view amount to a clear misuse of powers. However, in the context of the present references for a preliminary ruling, the Court does not have the evidence before it to reach a concluded view on whether that was, or was not, the case.

120.

I now turn to the two questions raised only in British Sugar, of which one is none the less relevant in all similar proceedings and the other may be relevant in some other Member States.

121.

It is clear that some amounts of production levy have been overpaid on the basis of invalid EU legislation. When duly claimed, they must be repaid, some years after they were collected, to producers by the national agencies and, in turn, to the national agencies by the EU. In any proceedings seeking such repayment, it is important to know whether those sums should bear interest.

122.

In the main proceedings in British Sugar, a claim for interest is available in principle under English law but the national agency argues that no award can be made, because it cannot itself reclaim interest from the EU. It relies on point 4 of the Commission’s guidance to Member States for the application of the contested regulation, ( 44 ) which states, essentially, that any interest on amounts repaid to producers by Member States cannot be deducted from the EU’s own resources, as the relevant legislation provides no basis for such deduction.

123.

The referring court asks, therefore, whether EU law precludes awarding interest to producers if no corresponding interest can be claimed by the national agency from the EU’s own resources, whether it is true that no such interest can be claimed and whether EU law precludes a national court or agency from exercising a discretion, under national law, to award no interest.

124.

It is settled law that, in the absence of provisions of EU law regarding reimbursement of charges improperly levied on the basis of EU regulations that have been declared invalid, all ancillary questions, such as the payment of interest, including the rate of interest and the date from which it must be calculated, are to be determined in accordance with national rules. ( 45 ) However, the referring court is unsure of the extent to which EU rules, in particular those relating to the possibility – or impossibility – for a Member State of obtaining interest from EU funds, may constrain the application of those national rules.

125.

The starting point, it seems to me, must be the underlying reason for awarding interest at all. And that reason is to be found in the prohibition of enrichment which is, in common law parlance, ‘unjust’ or, in many civil law systems, ‘without cause’ or undue. ( 46 )

126.

In Masdar, ( 47 ) the Grand Chamber stated: ‘According to the principles common to the laws of the Member States, a person who has suffered a loss which increases the wealth of another person without there being any legal basis for that enrichment has the right, as a general rule, to restitution from the person enriched, up to the amount of the loss. … [L]egal redress for undue enrichment, as provided for in the majority of national legal systems, is not necessarily conditional upon unlawfulness or fault with regard to the defendant’s conduct. … On the other hand … it is essential that there be no valid legal basis for the enrichment. … Given that unjust enrichment, as defined above, is a source of non-contractual obligation common to the legal systems of the Member States, the Community cannot be dispensed from the application to itself of the same principles where a natural or legal person alleges that the Community has been unjustly enriched to the detriment of that person.’ Those dicta appear particularly applicable to the circumstances of the present cases.

127.

When the EU budget has received from an economic operator, whether directly or indirectly, a sum to which it was not lawfully entitled, it is unduly enriched. To a considerable extent, that situation is redressed by repayment of the principal sum concerned. However, during the period between collection and repayment, the sum has been at the disposal of the EU and not of the economic operator. And money which is at the disposal of one party rather than another is available to bear interest for the former, but not the latter.

128.

In that situation, what matters is not so much the potential undue enrichment of the party in possession of the sum as the impossibility of normal enrichment for the party deprived of possession. That was the logic underlying paragraphs 82 to 89 of the Court’s judgment in Metallgesellschaft, ( 48 ) concluding that the premature levy of a tax, contrary to EU law, entitled the taxpayer to claim ‘the amount of interest which would have been generated by the sum, use of which was lost as a result of the premature levy of the tax’. In accordance with Masdar, that principle must apply also when the EU itself, rather than the Member State, is responsible for the error in collecting a levy.

129.

Consequently, I consider that the general principle precluding undue enrichment requires it to be possible for a sugar producer, entitled to repayment of a levy collected by the competent national agency and paid into the EU budget, to claim interest on the sum concerned in respect of the period during which that sum was not available to it for use.

130.

Moreover, to the extent that the sum concerned was available to the EU budget itself rather than to the Member State, the national agency whose duty it is to pay out the interest must be able to recover that interest, as well as the principal sum, from the EU budget.

131.

In the latter regard, I am not persuaded by the arguments to the effect that nothing in the own resources legislation provides for such interest. No explicit provision is needed in order to allow the application of a general principle such as that in issue, and no provision has been cited which might preclude the payment of interest. By contrast, the legislation in question ( 49 ) does make explicit provision for the payment of interest by Member States where there is a delay in crediting sums due. It would be inequitable if there were no possibility of obtaining interest in the reverse situation, where sums have been unduly credited.

132.

The Commission contends, however, that the 25% of the amount of production levies retained by the Member States ‘by way of collection costs’ should serve to cover any interest which the national agency is bound to pay in accordance with national law. It starts from the premiss that, where sums are levied by Member States and paid into the EU’s own resources, the relationship between the taxpayer and the Member State cannot be put on the same footing as the relationship between the Member State and the EU. Failure by a national agency to collect a levy which is due does not exonerate the Member State from its obligation to pay the sum concerned into the EU budget, together with any attendant interest. ( 50 ) By the same token, the fact that a Member State may be obliged to pay interest on repayments of amounts of own resources unduly levied does not imply a corresponding right for it to claim interest from the EU in the context of correcting the accounts. The 25% retained by the Member State is intended to cover all costs associated with collection, including those arising from national litigation concerning the consequences of over-payments. Since interest payments are ordered pursuant to national law in the framework of such national litigation, they should be seen as collection costs.

133.

I am not convinced. If failure by a national agency to collect a levy which is due does not exonerate the Member State from its obligation to pay the sum concerned into the EU budget, together with any attendant interest, then the logical conclusion is that, when the EU legislature unlawfully imposes a levy which is collected and paid into the EU budget by the Member States, which are then required to repay the sums concerned to the taxpayers with interest, the EU is not exonerated from its obligation to repay those sums to the Member States, together with corresponding interest.

134.

As regards the 25% retained by the Member States, it is true that such a percentage must contemplate a surprising level of inefficiency if it is intended to cover no more than actual collection costs. It is to be hoped that only a portion will in fact serve that purpose. There is, however, no indication in the own resources legislation that the remainder is intended to cover interest on repayment of sums unduly levied on the basis of EU regulations which have been declared invalid – which does not fall within the normal meaning of ‘collection’.

135.

In this regard, it seems to me that the most obvious approach is that taken by the French Government in its observations, viewing the total amount of the levy as being simply split between the EU and the Member State. 75% of the levy was paid into the EU budget and 25% retained by the Member State. Consequently, 75% of the amount to be repaid, together with any interest pertaining to that proportion, is to be charged to the EU budget, and the remainder to the Member State.

136.

The first two parts of the High Court’s fourth question should therefore be answered to the effect that EU law does not preclude economic operators from recovering interest on sums overpaid under an invalid regulation from the relevant national agency, or that agency from recovering equivalent interest from the EU budget.

137.

There remains the third part of that question: whether EU law precludes the national court from exercising any discretion it may have under national law to refrain from awarding any interest at all.

138.

As regards the referring court, such discretion is provided for in section 35A(1) of the Senior Courts Act 1981: ‘… in proceedings … for the recovery of a debt … there may be included in any sum for which judgment is given simple interest, at such rate as the court thinks fit, on all or any part of the debt … for all or any part of the period between the date when the cause of action arose and [payment or judgment]’.

139.

According to the case-law, such national rules are applicable to determination of any award of interest in cases such as the present.

140.

However, that is true ‘in the absence of provisions of [EU] law regarding reimbursement of charges improperly levied on the basis of [EU] regulations that have been declared invalid’. ( 51 )

141.

In the present cases, there may be no such ‘provisions’ of EU law but it seems to me that the primacy of individual national rules must yield to a general principle which is common to the legal systems of the Member States and which is recognised in the application of EU law. Under that principle, interest is recoverable to compensate for deprivation of enjoyment of the use of the amounts improperly levied. If there is any good cause to reduce the amount of interest awarded (as a result, for example, of the applicant’s conduct), the national court may exercise any discretion it has under national law to do so. However, the mere fact that there may be a loss to the EU budget cannot be regarded as good cause for that purpose by virtue of EU law.

142.

The final question to be addressed is relevant only when repayments are to be made in a currency other than the euro. Of the three sets of main proceedings, therefore, it concerns only British Sugar, though it may be of relevance in other Member States. Given that (i) the overpaid sums were expressed originally in euro then at various times converted into sterling for collection, (ii) any repayments to the national agency from the EU budget will have to be converted between euro and sterling at a later date, and (iii) there have been exchange rate fluctuations over the period concerned, which date or dates should determine the calculation of any necessary conversions?

143.

Under Article 6 of the contested regulation, Articles 1 to 4 apply from the dates of entry into force of the respective regulations whose provisions they replace. In addition, point 5 of the Commission’s guidance to Member States for the application of the contested regulation ( 52 ) states: ‘In the case of sugar levies collected, Member States which have not adopted the euro have converted the amounts fixed in euro in the relevant sugar regulations in order to enter the amounts in the own resources account. For consistency reasons and since the correcting regulation applies retroactively, the same exchange rate used by the Member States at the time the original sugar levies were calculated shall also be used to convert the corrected levies. Although the reimbursement of the unduly paid levies is governed by national law, the use of the historic exchange rate appears to be consistent with the inherent nature and objectives of the reimbursement and with the aim of avoiding distortions in the implementation of reimbursements across different Member States.’ The guidance then reminds Member States of the yearly regulations which fixed for each of the marketing years concerned the specific exchange rate applicable to, in particular, the production levies, for the currencies of those Member States which have not adopted the single currency.

144.

In England and Wales, a claimant may, on giving reasons, claim a sum of money expressed in a foreign currency. If he is successful, the award is for the sum in that currency ‘or the sterling equivalent at the time of payment’. ( 53 ) On that basis, British Sugar is seeking, before the national court, repayment of the overpaid amount, established in euro and converted into sterling at the rate applicable at the time of payment.

145.

The referring court asks whether the applicable rate is determined by EU law; if so whether Article 6 of the contested regulation requires application of the rate applied to the calculation of the original levy; and, if so, whether Article 6 is valid.

146.

I have already reached the conclusion that the contested regulation is formally invalid as a whole, and that it is substantively invalid in so far as it is based on a particular method of calculation. In that light, it seems irrelevant to examine further the meaning and validity of Article 6, which concerns only the dates of entry into force and application of provisions which I consider invalid.

147.

Since, however, the same issue will arise on the occasion of any subsequent regulation adopted to replace the contested regulation, it is relevant to ask whether the date for determining the applicable exchange rate is to any extent a matter of EU law.

148.

The applicable exchange rate is, in principle, an ‘ancillary matter relating to reimbursement’ within the meaning of the case-law ( 54 ) and, as such, to be determined in accordance with national rules.

149.

However, as the Commission points out, those rules do not operate in a vacuum, and the national court cannot ignore aspects which are governed by EU law – in particular the fact that specific exchange rates, fixed for each marketing year, were applied when the production levies were originally collected. I agree with the Commission that it would seem logical, if a producer’s claim is to be determined on the basis of restitutio in integrum, to take those exchange rates into account. If a more recent exchange rate were none the less to be applied as a result of national rules, it would further seem appropriate for the national court to bear any windfall gain (or loss) in mind when awarding interest.

150.

In the light of all the foregoing considerations, I am of the opinion that the Court should rule as follows in response to the questions raised by the Finanzgericht Düsseldorf, the High Court of Justice of England and Wales and the Tribunal de grande instance, Nanterre: