32000Y1127(01)

Special Report No 17/2000

on the Commission's control of the reliability and comparability of the Member States' GNP data together with the Commission's replies

(pursuant to Article 248(4), second subparagraph (EC))

(2000/C 336/01)

CONTENTS

>TABLE>

EXECUTIVE SUMMARY

GNP, which is considered to be representative of the Member States' ability to contribute to the Community budget, plays a much more important role than that of merely financing the budget. This aggregate is used as a reference indicator in the field of structural measures as well as in the context of economic and monetary union. Community control procedures are the only control procedures capable of guaranteeing the reliability and comparability of GNP data established by the Member States using their own statistical procedures. On the Commission's initiative, extensive harmonisation work has been carried out since 1989, resulting in substantial changes in the national GNPs. The Commission should, however, improve and extend the controls, provide better monitoring and introduce an adequate documentary framework. The Court recommends that the efforts to improve data quality continue, so that GNP can validly perform the role assigned to it.

INTRODUCTION

1. Gross national product (GNP)(1) gives a measurement of a quantity of production, the real value of which is and remains unknown. It is calculated on the basis of estimates, information from various sources, economic models and expert estimates.

2. At Community level, GNP is considered to be the indicator which best represents the Member States' ability to contribute to the budget. That is why, particularly since 1988, it has come to play a growing and decisive role in the funding of the Community budget. In March 1999 the Berlin European Council decided to increase the weight of the GNP resource in the funding of the budget, with effect from 2002(2).

3. Considerable work on harmonising national statistical systems has been carried out since 1989 under the aegis of the Commission. The Court of Auditors has examined the extent to which the Member States' calculations of GNP can now be considered comparable and reliable for use at Community level(3).

4. The audit, which was carried out at the Commission, aimed to establish the level of assurance of the system, at Community level, as regards the reliability and comparability of the statistical data in question. It considered the Commission's procedures in respect of Member States' compliance with the rules for establishing GNP and examined the underlying statistical data.

5. The comparability of GNP between Member States largely depends on taking proper account of two particular aspects of the national accounts, namely their exhaustiveness and the distinction between intermediate consumption and final uses(4). The Court analysed information and data for four Member States(5), whose share of Community GNP is about 70 %. In addition, in order to consider more fully the reasons for certain anomalies that had been discovered during previous Court audits, the specific question of how VAT evasion is included in the national accounts was also examined in two other Member States(6).

THE USE OF GNP AT COMMUNITY LEVEL

Own resources

6. The ceiling for the own resources contribution to the Community budget is 1,27 % of total national GNP(7).

7. Contributions to the budget via the GNP resource are calculated as a percentage of each national GNP. This resource forms part of the budget funding and is used to make up the difference between planned expenditure and the other resources available (customs duties and agricultural duties and the VAT resource).

8. Calculation of the VAT resource is affected by statistical data in several respects. The basis of assessment for this resource is limited to the equivalent of 50 % of GNP where necessary.

9. The statistical data on which GNP is based also have a significant influence on the actual calculation of this assessment base(8). This is the base that is used to establish the weighted average rate. It also applies for several compensations or corrections that are designed to take account of national characteristics that affect the application of VAT legislation.

Other uses of GDP/GNP

10. In the Community context, the role of GDP/GNP(9) is not restricted to the financing of the budget. This aggregate is also taken into account in decisions on structural aid policies for regions whose development is lagging behind. It is also considered in setting the maximum amount of expenditure for the EAGGF Guarantee Section, when allocating the European Central Bank's capital between the Member States and in determining the financial participation of non-member States which are members of the European Economic Area.

11. This aggregate is also used in the context of the Stability Pact provided for in the context of economic and monetary union, particularly for monitoring government deficits and public indebtedness ratios. On a more general level, GDP/GNP is used as a benchmark for the coordination of national economic policies and for the policies concerned with the completion of the single market.

12. The correct establishment of GDP/GNP is, therefore, not just about determining own resources and ensuring equality of treatment for the Member States in the funding of the budget. It is a requirement that has much wider repercussions.

CHECKING GNP

At Community level

13. The Commission is responsible for validating data which are initially drawn up at national level, but may subsequently become Community statistics(10).

14. Directive 89/130/EEC, Euratom provided for the setting up of a committee(11) to assist the Commission in reviewing the calculation of GNP, notably with regard to:

- compliance with the European system of integrated economic accounts (ESA)(12),

- examining data forwarded by the Member States, statistical sources and calculation procedures,

- verifying and assessing the practical comparability and exhaustiveness of GNP.

15. The provisions on the determination of the GNP resource lay down that every year the Commission, together with the Member State concerned, is to check that the aggregates sent to it have been correctly taken into account, especially in the context of cases notified by the GNP management committee. In individual cases, the Commission may "examine calculations and basic statistics (...), where no proper assessment would otherwise be possible"(13).

16. When checking the GNP calculations, the Commission is empowered under Article 6 of Directive 89/130/EC, Euratom to take the measures it judges necessary(14). In order to safeguard the Community's financial interests, the Member State concerned is sent notice of a reservation. Under the provisions on own resources(15), this notification makes it possible to require the Member State concerned to make the appropriate adjustments. As long as the reservations have not been lifted, the Commission does not consider that the GNP for a given financial year is definitive.

At national level

17. In the Member States, own resources based on GNP are entered in the accounts as national expenditure paid over to the Community budget. They may therefore be audited by the national authorities according to their own rules. In practice, the payments to the Community budget in connection with the GNP resource are not audited until the statistical data underlying the assessment base have been obtained. The only control procedures that apply to the establishment of these statistics and to the working methods and concepts used are those that exist within the national statistical offices. Mainly because of their independent status, it is the latter that compile the data, on their own responsibility.

THE RELIABILITY AND COMPARABILITY OF THE DATA

18. The legitimacy of GNP as a benchmark aggregate at Community level is closely dependent on the possibility of checking that the data in question fulfil the quality conditions. This concept, which is based on the two criteria of the reliability and the comparability of the data, is summarised in the following figure(16).

>PIC FILE= "C_2000336EN.000501.EPS">

19.

The Court's checks focused on the factors that may make compliance with these criteria difficult, namely:

(a) the process of establishing GNP (paragraphs 20 to 26);

(b) the control procedures (paragraphs 27 to 62);

(c) the possibility of measuring the quality of the data (paragraphs 63 to 79).

The process of establishing GNP

20. The components of GNP are defined by the European system of integrated economic accounts (ESA), which provides a framework of statistical standards, concepts, definitions and rules(17). It is obligatory for the Member States to use the ESA. Under the provisions applicable, and Directive 89/130/EEC, Euratom in particular, they must provide statistics of GNP.

21. Although, by means of the ESA, Directive 89/130/EEC, Euratom established standard principles for the definition of the aggregates, it did not impose a common method of calculation on the Member States. This is due to different national circumstances with regard to the availability and timeliness of the data sources, the calculation and assessment methods used and the organisation of the statistical systems.

22. In general, the Member States use at least two approaches to calculating GNP, but sometimes the estimates involved are not entirely independent of each other. In most cases, it is the output and expenditure approaches that are used, the former being the dominant one(18).

23. Before GNP is established, the results of these different approaches are integrated. The breakdown level at which integration takes place varies from one Member State to another. For most of them, this is done at product level, but some make adjustments at aggregate level(19).

24. Notwithstanding the national characteristics that lie behind the differences in sources and methods of producing GNP, the Court, like the Commission, considers that the real question is whether, above and beyond the purely formal harmonisation resulting from the ESA, Member States' GNP figures are comparable on both the conceptual and the practical level(20).

25. Furthermore, as the Court has already pointed out in previous observations(21), the principle of the impartiality of Community statistics(22) dictates that the accounting framework used should be the most powerful one available. A new version of the ESA (ESA 95) took effect in April 1999. However, the Regulation setting up the system(23) provided that the old ESA (ESA 79), which is considered less reliable, remains in force, for own resources purposes only, as long as the current financing system has not been changed.

26. As a consequence of retaining ESA 79 the accounts, which are currently drawn up according to ESA 95, are subject to a backwards conversion, purely for own resources purposes. This operation not only duplicates work that has already been done, it also presents a real risk of additional error. Even if the new Decision on own resources that is in the process of being adopted by the Council provides for ESA 95 to be applied as from 2002, the effects of this double system will nevertheless be felt beyond that date(24).

The control procedures

Reservations concerning GNP

27. Following the entry into force of Directive 89/130/EEC, Euratom the Commission set itself the objective of harmonising methods of calculating Member States' GNP. To do this, it applied itself to examining with the GNP Committee inventories of the procedures and statistical bases used by the Member States for calculating GNP and its components.

28. Following this analysis, the Commission notified the Member States of a certain number of reservations, mainly in 1992(25). Technically, the points in question concerned the methodologies and sources used in drawing up the national accounts, where the Commission felt that one or other of the Member States had not respected the Community's conceptual framework. One reservation that applied to all the Member States concerned the exhaustiveness of their national accounts.

29. In order for these reservations to be lifted, Member States were expected, by improving their statistical tools, to provide evidence of the "comparability, reliability and exhaustiveness" of GNP(26). After several deferrals, the deadline for aligning the statistical systems was set at 1 October 1998. Austria, Finland and Sweden were given an extra year.

30. At the end of 1998, there were 85 reservations on the GNPs of the 12 "old" Member States. One year later, only 20 were still outstanding, concerning three in particular(27). It is clear, therefore, that the Commission has lifted most of these reservations.

Observations on the reservations

31. For the national GNPs referred to in paragraph 5, the Court's examination focused on two reservations out of those mentioned above, namely, the exhaustiveness of GNP and the distinction between final uses and intermediate consumption(28). Examination of these reservations, since lifted by the Commission, revealed several problems, which are discussed below.

EXHAUSTIVENESS OF GNP

Transparency of adjustments

Lack of satisfactory justification for certain adjustments

32. The number of adjustments for exhaustiveness was high. It was not possible for the Court to judge the validity of some of the adjustments, for want of relevant explanations, or explanation of any kind, or, even, because of lack of data.

33. This was particularly the case when exhaustiveness adjustments were justified by means of a technical coefficient or the term "best estimate", without any further information. On occasion, the major part of the exhaustiveness adjustment was based on residual items (other, miscellaneous), without further explanation. Up to 80 % of the correction relating to final consumption may be justified by an item of this type. In another case inconsistencies were revealed by comparing changes in an adjustment for exhaustiveness and VAT evasion. It was not possible to clarify this question, because there are no figures for some of the exhaustiveness adjustments, especially those that are implicit in the national accounts.

34. For example, in one case where the Commission did not feel able to verify the accuracy of a classification, it chose not to follow up the matter. In another, the Commission accepted an exhaustiveness adjustment which was admitted to have no reliable statistical base. This last question now appears to have been resolved.

Comparability bias

Comparability of adjustments between Member States

35. The Court took note of a study which sought to compare a Member State's own estimate of the income under-declared by taxpayers with the estimate obtained for the same Member State by means of a method used by another Member State. The exercise produced very different results, which could not be explained. It thus raises the question of whether it is really possible to be sure that exhaustiveness adjustments are comparable.

Treatment of illegal activities

36. Illegal activities are included in the GNP field(29). However, because of practical difficulties associated with the measurement of them, Decision 94/168/EC, Euratom excluded them from the scope of exhaustiveness adjustments. The Court considers that in consequence their inclusion by Member States is not homogeneous and there is still no standard definition of what constitutes an illegal activity.

Outdated sources

37. The exhaustiveness adjustments are often based on technical coefficients calculated in terms of a base year, or on surveys or ad hoc estimates which, in turn, are sometimes outdated. This raises the question of how up to date the sources are, especially where the dynamics of the unobserved economy differ from those recorded directly by the statistical offices. In that case, the estimated adjustments will be equally biased.

38. For example, there is no guarantee that bases that are 20 years old are still relevant or that the structural changes that have affected fraud since then (volume, sectors of activity and type) have been taken into consideration.

Procedures for revising adjustments

39. Some Member States prefer to make substantial adjustments(30) for the base years only, in order to have series that are homogeneous over time. This meets national requirements, which aim to have series that are homogeneous in terms of growth rates. The inherent risks are, among other things, the absence of comparability between Member States' adjustment figures, for example between Member States which frequently update their adjustments and those which do so only when the base is altered.

40. Evaluation of these problems is difficult, because Decision 94/168/EC, Euratom only provides for a "description (referring to) a recent year for which final estimates are available"(31), and does not require a description of the procedures for revising exhaustiveness adjustments.

Taking account of VAT evasion

41. The concept of GNP at market prices is such that the value of purchases of goods and services includes the taxes paid, particularly VAT. In case of fraud or tax evasion(32), the GNP calculation takes these purchases into account at the price actually paid, i.e. excluding VAT. Where the purchaser has paid the tax and it has not subsequently been paid to the tax authorities by the supplier, Member States are, however, required to make adjustments under the heading of tax evasion "without complicity".

42. Decision 98/527/EC, Euratom provides an ad hoc method for calculating tax evasion "without complicity"(33). However, it allows Member States the possibility of reaching the same result by other means.

Equivalence of the methods used by the Member States

43. Although the Decision lays down precise rules, it was not always possible to assess the extent to which the method recommended by the Commission had been applied or to establish the equivalence with other methods used by the Member States. In some cases the supporting documentation was missing, while in others the definitions laid down had not been applied.

44. For some of the national GNP examined it therefore remains to be established whether VAT evasion has been correctly taken into account.

Anomalies in the results of the calculation of theoretical VAT

45. The calculation of theoretical VAT through the national accounts includes the factors that may explain a VAT underpayment(34). For that reason theoretical VAT can be expected to be higher than the VAT actually collected. In the GNP Committee's view, this is the only way of ensuring that the national accounts have at least taken account of some of the transactions that are not visible to the tax authorities.

46. However, the Court detected cases where, contrary to the above principle, the VAT collected was higher than the amount theoretically due. This finding may be symptomatic of an underestimation in the national accounts. This problem has already been highlighted by the Commission, following work carried out in 1980.

47. In 1991 the Commission did actually establish, in the case of one Member State, that this anomaly in the difference between actual and theoretical VAT revenue was due to underestimation of the final consumption by households. Whilst the Court does not have a complete set of evidence, it notes that in the Commission's view the problem mentioned has now been resolved.

Business register

48. In order to ensure that the national accounts are exhaustive it is essential to have the most detailed knowledge possible of the units whose output is to be taken into account when the statistics are compiled. The purpose of such a tool, processing information from several sources, is to avoid omission or double-counting of producer units. In 1992, the Commission stated that the potential impact of the business register on GNP measurement could be as great as taking the underground economy into account.

49. Member States are required to establish such a register(35). The Court's audit nevertheless revealed that the reservation concerning exhaustiveness in respect of one Member State had been withdrawn, even though the latter still had no such register. In such cases the Commission should state the reasons that enabled it to obtain assurance that this deficiency does not affect the exhaustiveness of the national accounts.

DISTINCTION BETWEEN FINAL USES AND INTERMEDIATE CONSUMPTION

50. Unlike the question of GNP exhaustiveness, there are detailed definitions at Community level of the difference between final uses and intermediate consumption. The Court's audit revealed several cases where the definitions specified in the ESA had not been respected. In some cases the Commission itself had established a lack of compliance, in some instances for all the Member States. In the Commission's view these cases did not have a significant impact on the calculation of GNP.

Balancing procedures

51. When confronted with lack of compliance with the ESA conventions some statistical offices say that balancing-out the accounts provides a solution to the problems. The Court's checks did not provide any assurance that balancing out the uses and resources for a product automatically prevents an implicit bias in the data.

52. In order to validate such an assumption, proof of independence, in relation to the problem of non-compliance with the ESA, needs to be provided for least one of the sources used in the procedure. It is also important to evaluate the efficiency of the procedure itself.

The limited nature of the checks

53. With regard to the exhaustiveness of the national accounts, the Commission stated in July 1999 that the matter had been subject to in-depth checks and that the underground economy was taken into account in the national accounts. As of May 2000, however, there was still a reservation concerning this point in the case of five Member States(36).

54. The Commission said that the lifting of this reservation was the result of a global assessment of all the explicit and implicit adjustments made. The assessment was based on the study of sources and statistical methods which the Commission defines as being its primary source for auditing purposes(37).

55. The Commission admits that this source alone is not sufficient to guarantee the quality of the data and that, since the level of detail in the inventories is not sufficient, it is sometimes necessary to obtain additional information from the Member States.

56. As regards the two reservations covered by the inquiry, the documents supplied by the Commission reported specific checks. The Commission did not carry out any tests that could provide global quality assurance, for example examining Member States' statistical organisation, the conditions under which the basic data were produced, the compatibility of the various data sources used, the balancing procedures and the methods used to establish the level of assurance.

57. The checks made are not sufficiently well-documented for it to be possible to ascertain their extent or to determine their coverage of the various GNP components. As regards the reservation concerning the exhaustiveness of the national accounts in particular, a detailed programme of checks based on the criteria of reliability and comparability of data should have been drawn up(38).

58. Moreover, the Commission judges the accuracy of the data, not on the basis of a strict application of the concepts or exhaustive practical measurement, but on whether a procedure is unaffected by substantial errors. The Court is aware that in the very nature of GNP the quality of the estimates is also subject to cost-benefit considerations. The fact that such a principle applies presupposes that criteria exist and that justification is adequate. However, the reply that was given in several cases merely described the problems raised as negligible, without providing any information that could be used to assess the reasons underlying this opinion.

59. There are also grounds for wondering how the Commission, in certain cases, arrived at the view that the effects of several types of non-compliant treatment cancel each other out and therefore have no impact on the GNP calculation, or, even, that the use of different definitions by the statistical offices produces congruent and comparable results.

THE ROLE OF THE GNP COMMITTEE

60. The GNP Committee delivers an annual opinion on the basis of information supplied by the Member States and the Commission. Every year since 1990 the GNP Committee has asserted in this document that "considerable improvements have been made" and "notes with satisfaction the progress made and being made in the Member States in harmonising GNP data". As regards compliance with quality concepts such as those mentioned in paragraph 18, the Committee's opinion offers no benchmark against which the levels achieved in each Member State and the progress yet to be made may be registered.

61. In 1992, in its report on the application of Directive 89/130/EEC, Euratom, the Commission asserted that as far as examination of the data was concerned, the Committee had confined itself to establishing their availability. The results of the analysis of the detailed inventories and the findings that might result were unavailable.

62. This argument is no longer valid today. The Committee, comprising experts in this field, is supplied with information by its members, on the one hand, and obtains data from the Commission's analyses, on the other. In principle, it has all the information it needs to form an overall opinion concerning the comparability and reliability of the data.

The possibility of measuring the quality of the data

63. The production of the national accounts may be divided into three main stages. The first uses a complex information system, made up of sources which are purely statistical, fiscal, administrative, exhaustive or the result of sampling and which make it possible to approach GDP, and then GNP, from three angles (demand, output and income)(39) as well. During this phase every source, and its processing, gives rise to errors. Quality measurement involves modelling the error of each of the basic components and then deriving from it an indicator of the quality of the overall aggregate.

64. At the second stage information obtained on the basis of different approaches or independent sources is compared and balanced. Comparative differences are another indicator of quality, in addition to those mentioned above (comparisons based on the various approaches).

65. Lastly, as information gradually becomes available, the economic aggregates produced undergo some revision. The final stage of the quality measurement approach is to study the profiles of the revisions, more specifically to measure any bias in the early version by reference to the final version (analysis of changes in the revisions).

66. Given the heterogeneous nature of the systems for producing GNP figures, it is essential that the level reached by each Member State is evaluated and compared with a quality concept such as that described in paragraph 18.

67. In the past(40) the Court has recommended the introduction of a means of measuring the comparability (on the basis of indicators) and the reliability (by means of confidence intervals) of GNP data.

68. The Commission entrusted the examination of this matter to two national statistical institutes, which have just completed their work. It showed that progress in this area cannot be achieved unless a satisfactory solution to the problems described below is found.

Comparison on the basis of the approaches to GNP

69. In order to improve the quality of the GNP calculated from the various approaches, use is occasionally made of "account balancing". In some cases quality cannot be measured unless there is a trail through the intermediate stages of construction, especially of the aggregates at the various stages of comparison. However, this is not the case in all the Member States. The Commission and the Member States should, therefore, make specific provision to this effect.

70. It is difficult to assess the quality of derived aggregates when account balancing is also used to evaluate certain variables, such as, for example, variations in inventories. A country which compares two approaches in order to estimate variations in inventories will, in fact, automatically record fewer balancing errors than a country which estimates inventories more directly and, paradoxically, collects more information.

71. At branches of activity level, the refinement of the calculation exercise depends on a basic choice by the national accounts authorities; it therefore differs markedly from one country to another and produces, ex hypothesi, a lack of comparability.

72. Similarly, a country which uses only one approach for its calculation cannot, therefore, take advantage of the possibility of comparisons between approaches and will not be troubled by comparative differences. It will inevitably record less variation, but the quality of the data will not necessarily be any better.

73. Furthermore, as the approaches used in the GNP calculation are never totally independent, balancing two approaches with the same shortcomings would not show up those shortcomings and might have an adverse effect on the quality of the derived aggregate.

74. In the case of sampling errors, there is some consensus on the statistical techniques. For other types of error(41), on the other hand, information is either unavailable or subjective. However, subjectivity in error measurement also has an impact on the validity of indicators of comparability between Member States.

Analysing changes in revisions

75. The bias in the early versions of GNP is measured against the final version. In order to draw conclusions from such an exercise, the quality of the final version must therefore be evaluated before the revision changes are analysed. In effect, a rudimentary final account that does not use much supplementary information relative to the early versions would indicate a lack of bias in the early versions. Excessive automatic use of indicators based on changes in revisions would lead to underevaluation of the quality of the GNP of a Member State with an information system which is complex and therefore slow to mobilise.

Comparability between Member States

76. National experiments on measuring the quality of economic aggregates are currently pursuing the Member States' own objectives. The primary aim of the latter is to ensure that the aggregates are good indicators of economic trends (growth or recession) and that these indicators provide relevant information on any divergence between growth and a long-term trend or an indication of the economy's position within a cycle.

77. The first obstacle that must be overcome before the quality of GNP can be measured comparatively is the diversity of the statistical systems themselves. It is a question of reaching a consensus on how various types of error are to be evaluated and aggregated in order to give the overall error.

78. The Commission's work on defining the quality of the macroeconomic aggregates and the classification of errors is helping to establish a common basis on which all the Member States can begin to measure quality.

79. However, the work done so far has not succeeded in making it possible for quality measurement to be considered from the viewpoint of its comparability between Member States, even though that is a prerequisite for using GNP as a reference aggregate in the Community context. A certain amount of work therefore needs to be devoted to both this aspect and the problems mentioned in paragraphs 69 to 75, for example in the context of the quality-measurement task-force which the Commission has decided to create.

CONCLUSION

An aggregate that is widely used in the Community context

80. GNP is considered to be the best indicator of a Member State's ability to contribute (see paragraph 2). To a large extent it determines the Member States' contributions to the financing of the Community budget and directly affects certain items of expenditure (paragraphs 6 to 10). At the same time GNP is the basis for important decisions, such as those taken in the context of economic and monetary union (see paragraph 11).

Guaranteeing legitimacy

81. Essentially GNP measures by estimation a value of production that is, and remains, unknown (see paragraph 1). Added to the uncertainty associated with the statistical nature of this aggregate is the fact that each Member State applies its own procedures for producing statistics. It is therefore necessary to ensure that this aggregate is estimated with the requisite precision (see paragraphs 20 to 24).

82. In order to guarantee the legitimacy of GNP as a reference aggregate, it is important in the first instance for it to be established transparently on the basis of explicit, well-documented methods. There must be adequate control procedures, to ensure that the data are reliable and comparable at Member State level (see paragraphs 53 to 62).

83. The Commission's role in this context is not just important, it is also unique: at the national level, the only existing controls on the procedures for producing statistics are those set up within the statistical offices. The Commission is thus the only authority that has both an overview of the situation and the necessary analytical and legislative resources (see paragraphs 13 to 17).

84. Any advance in measuring the quality of GNP must, inevitably, enhance its legitimacy (see paragraphs 63 to 79).

Completing harmonisation

85. Harmonisation of GNP has been a main objective since the adoption of Directive 89/130/EEC, Euratom in 1989. In 1996 the Commission acknowledged, however, that complete harmonisation of GNP had not been achieved and that much work remained to be done.

86. For example, at the beginning of 1999 several members of the GNP Committee still considered that the work of improving GNP data was unfinished. Most of the reservations have nevertheless been lifted since then (paragraph 30). The scope and volume of the work that is still necessary should be examined in relation to the present and future role of GNP.

Effective controls at Community level

87. The existence of a European system of economic accounts and of several other legislative provisions cannot, in itself, guarantee that data are reliable and comparable (see paragraph 50).

88. Tests on the chosen data must cover the intrinsic reliability of each statistical system, the conceptual comparability (compliance with the ESA) and the practical comparability (sources and methods used). This entails examining the statistical organisation of the Member States, the conditions under which the basic data are produced, the compatibility and independence of the various sources, balancing procedures and confidence limits.

89. The Court's checks revealed the existence of several problems that affect the procedures for checking and validating GNP. The shortcomings noted in terms of compliance with the ESA, transparency of the adjustments, bias of comparability and outdated sources could affect the quality of the data relative to the applicable provisions (see paragraphs 32 to 52).

90. Despite certain improvements that have been made in recent years, the Commission has not yet adopted sufficiently transparent analytical procedures for validating the data used in calculating GNP. The objectives and content of its inspection activities should be better defined and documented (see paragraphs 53 to 59).

91. It appears essential for the Commission to quantify the impact of its reservations on GNP. This factor should be one of the criteria for exercising its prerogatives and referring matters to the Court of Justice where necessary, as provided in Articles 226 and 228 of the EC Treaty.

92. Finally, provided that the best available data are used, the application of ESA 95 to own resources could improve and rationalise the system for drawing up the national accounts (see paragraphs 25 to 26).

This report was adopted by the Court of Auditors in Luxembourg, at the Court meeting of 14 September 2000.

For the Court of Auditors

Jan O. Karlsson

President

(1) In accordance with Council Directive 89/130/EEC, Euratom of 13 February 1989 on the harmonisation of the compilation of GNP at market prices (OJ L 49, 21.2.1989, p. 26), GNP, of which the version at market prices is applicable for calculating own resources, is calculated on the basis of gross domestic product (GDP). GNP equals GDP plus compensation of employees and the property and entrepreneurial income received from the rest of the world minus the corresponding flows paid to the rest of the world. For a given period, GDP measures, in monetary and gross value terms (that is to say, without deducting consumption of fixed capital), the final result (goods and services) of the production activity by resident producer units in an economic territory. GDP may be presented from three different points of view (the output of resident producer units, the final uses of goods and services by these units or the income generated by these units). The market prices applied to evaluate, in monetary terms, the goods and services produced include all the taxes linked to the production or importation of these goods and services net of subsidies. The GNP and GDP aggregates form part of the definitions in the European system of integrated economic accounts (ESA), the last version of which (ESA 95) was adopted by Council Regulation (EC) No 2223/96 of 25 June 1996 (OJ L 310, 30.11.1996). In ESA 95, the concept of GNP that existed in the previous ESA (ESA 79) was replaced for most practical purposes by gross national income (GNI).

(2) European Council meeting in Berlin, 24 and 25 March 1999, Presidency conclusions, III. Own resources and budgetary imbalances, paragraphs 67 to 76. This agreement, which was reached by the Heads of State or Government, called for the adoption of a new decision on the system of own resources. The entry into force of such a decision is subject to national ratification procedures. A proposal for a new decision was presented by the Commission in July 1999 (COM(1999) 333 final of 8.7.1999). The Court has delivered an opinion on the proposed decision (see Opinion No 8/99 (OJ C 310, 28.10.1999, p. 1)).

(3) The Court reported on its work on GNP in its Annual Reports concerning the financial years 1991 (paragraphs 1.119 to 1.131), 1992 (paragraphs 1.80 to 1.104), 1993 (paragraphs 1.68 to 1.101), 1994 (paragraphs 1.112 to 1.131) and 1995 (paragraphs 1.104 to 1.128).

(4) For a fuller definition of the concepts of exhaustiveness, intermediate consumption and final consumption, see Annex I.

(5) Germany, France, Italy, United Kingdom.

(6) Ireland, Luxembourg.

(7) In 1999, own resources amounted to EUR 85555 million in total, of which 37 % was provided by the VAT resource and 44 % by the GNP resource.

(8) Subject to any specific adjustments that may be necessary for calculations to be comparable between the Member States, the assessment base for a Member State's VAT resource is obtained by dividing the revenue collected during a financial year by the weighted average rate. This rate is supposed to represent the weighting of the various rates of VAT applied to the various taxable operations, reconstituted statistically.

(9) The difference between GDP and GNP is described in footnote 1, page 3.

(10) At the Commission, Eurostat is responsible for carrying out work on the production of these statistics (Commission Decision No 97/281/EC of 21 April 1997 (OJ L 112, 29.4.1997, p. 56)). For the purposes of the GNP resource, Eurostat approves the data used for the assessment base and communicates them to the authorising department for own resources. Apart from specific provisions applicable to the calculation of GNP, Council Regulation (EC) No 322/97 of 17 February 1997 on Community statistics provided a definition of the basic principles with which Community statistics are to comply (OJ L 52, 22.2.1997, p. 1). These include the principles of impartiality, reliability, relevance, cost-effectiveness, statistical confidentiality and transparency. These same principles have also been introduced into the EC Treaty recently (Article 285). Under that provision the Commission may propose to the Council measures for the production of statistics where necessary for the performance of the activities of the Community.

(11) The GNP Committee is composed of representatives of the Member States and is chaired by the Commission.

(12) The function of the ESA is described at paragraph 20.

(13) Article 19 of Council Regulation (EEC, Euratom) No 1552/89 of 29 May 1989, (OJ L 155, 7.6.1989).

(14) See also footnote 3, page 4. regarding the fundamental principles governing Community statistics and the production of statistics needed for the performance of the activities of the Community in the light of Article 285 of the EC Treaty.

(15) In particular, Article 10(8) of Council Regulation (EEC, Euratom) No 1552/89 of 29 May 1989.

(16) Directive 89/130/EC, Euratom uses numerous terms and concepts which contribute to the quality of GNP, but without defining or ranking them. It mentions the need to "reinforce comparability and reliability" (first recital) and specifies that "GNPmp data must be comparable in concept and in practice and representative of the Member States' economies" (third recital). These same concepts are found in Article 7, which defines the role of the GNP Committee (see paragraph 14).

(17) The ESA is an accounting framework adapted to the structures of the Member States' economies. The fact that it is compatible with the United Nations system of national accounts (SNA) means that it can be used to draw international comparisons. The ESA provides a detailed and systematic description of a "total economy" (a region, a country or a group of countries), its components, its structure, its relations with other economies and the way it changes over time.

(18) The approaches to establishing GDP, from which GNP is derived, are discussed in footnote 1, page 1. An indication of the variety of approaches used is given in Annex II, which summarises the way in which each Member State calculates its GNP.

(19) The main characteristics of the integration methods are described in Annex II.

(20) Annual report of the Court of Auditors concerning the financial year 1991, Commission's reply to paragraph 1.129.

(21) Special Report No 6/98 concerning the assessment of the systems of own resources based on VAT and GNP, paragraphs 3.17 to 3.22.

(22) See footnote 3, page 4.

(23) Council Regulation (EC) No 2223/96 of 25 June 1996 on the European system of national and regional accounts in the Community (OJ L 310, 30.11.1996, p. 243).

(24) Due, in particular, to the fact that the assessment base for any given financial year gives rise to adjustments in subsequent budget years.

(25) See paragraph 16. These points, of which there were 157, are listed in Commission Decision 97/619/EC, Euratom of 3 September 1997 (OJ L 252, 16.9.1997, p. 33). Some Member States have also used the right to notify reservations in order to provide better quality data on specific points at a later date.

(26) Decision 97/619/EC, Euratom, first recital.

(27) At the end of 1999 the total number of reservations outstanding was 69. To the 20 reservations mentioned must be added a further 34 concerning Austria, Finland and Sweden, as well as one reservation for each Member State concerning the transition between ESA 95 and ESA 79.

(28) See Annex I for a detailed definition.

(29) National accounts are, by definition, concerned with economic flows, not with their legality. Accordingly, illegal activities are entered in the accounts as output, provided that the transaction is undertaken by agreement. This principle is found in the 1993 United Nations' system of national accounts (SNA) (paragraphs 3.54 to 3.56 and 6.30 to 6.33) and in ESA 95 (paragraphs 1.13 and 1.42). ESA 79 does not refer explicitly to these activities but, in the absence of any statement to the contrary, it may be interpreted as including them. SNA 1993 (paragraph 6.35) states that the demarcation between the underground economy (that is, the economy that is not observed directly) and illegal activities cannot be precisely determined.

(30) For example, changes in concepts, the substantial improvement or, possibly, creation of a source.

(31) Article 3 of Decision 94/168/EC, Euratom.

(32) In this case, the supplier, with the agreement of the purchaser of a good or service, does not declare a taxable transaction to the tax authorities. The price paid is therefore, by definition, net of tax. In such cases there is no need to adjust the GNP for VAT. This form of evasion is classed as tax evasion with complicity.

(33) OJ L 234, 21.8.1998, p. 39. According to this Decision, tax evasion "without complicity" is calculated by subtracting from the theoretical VAT revenue (the amount that would be collected if units subject to VAT paid it as required by law) the following factors:

- actual VAT revenue,

- time differences between treasury data and the national accounts,

- insolvency-related cancellations of claims,

- missing revenue (evasion "with complicity").

(34) These include, inter alia, time differences in treasury data and cancelled claims (see footnote 1, page 8).

(35) Council Regulation (EEC) No 2186/93 of 22 July 1993 on Community coordination in drawing up business registers for statistical purposes (OJ L 196, 5.8.1993, p. 1).

(36) Including Austria, Finland and Sweden. For these Member States, a deadline of 1 October 1999 had been set for transmission of the supporting documents.

(37) See paragraph 27.

(38) See, for example, the figure illustrating paragraph 18.

(39) See footnote 1, page 3.

(40) Annual report of the Court of Auditors concerning the financial year 1995, paragraph 1.126.

(41) For example errors concerning the parent population or the base, measuring errors, errors in methods, errors in the treatment of failures to reply, errors due to the modelling of economic phenomena.

Annex I

The exhaustiveness of GNP

The conceptual framework

1. Exhaustiveness reflects the need to include in the calculation of GNP all activities which produce goods or services. It is one of the factors that have considerable influence on practical comparability.

2. A proportion of economic activity escapes direct observation by means of methods such as questionnaires, periodic censuses and sampling. The fact that this part of the economy is not directly observed is essentially a consequence of an intrinsic weakness in the measuring capacity of the statistical apparatus(1). Alternatively, it may be attributable to economic activities that are carried out in breach of the law, in particular for the purpose of tax evasion or social-security fraud(2). The term "underground economy" is often used in this context(3).

3. For all these reasons, the statistical offices must make appropriate adjustments to the various components of GDP/GNP. The part of the economy that is not directly observed varies from one Member State to another. It is linked to the nature and quality of the national statistical apparatus.

Exhaustiveness: the necessity of compromise

4. By enhancing methods and sources it is possible to improve the coverage, i.e. exhaustiveness of GNP. Shortcomings in this area lead to bias in GNP measurement and, in practice, to poor comparability between Member States' GNPs, thereby reducing the validity of GNP as a reference aggregate.

5. The search for exhaustiveness does, however, encourage the use of non-statistical sources which are not primarily designed for national accounts purposes(4). The result is, firstly, difficulty in reconciling these external sources with the existing statistical system and, secondly, inaccuracy, which is all the greater because the populations studied are unusual and hidden and hence the information is biased(5).

6. There are thus limits to the possibility of making GNP exhaustive. The process entails a compromise between reducing bias (considering a more comprehensive field of domestic production) and increasing imprecision (using sources and methods which are necessarily imperfect). From the Community point of view, the success of this compromise also depends on the comparability of Member States' statistics.

Decision 94/168/EC, Euratom

7. When Directive 89/130/EEC, Euratom was adopted, the Council and the Commission highlighted the question of increasing the coverage of the 'parallel economy' within the national accounts framework. The Commission noted that, in this respect, the situation differed from one Member State to another, "even for countries which have comparable levels of statistical development"(6).

8. The aim of Commission Decision 94/168/EC, Euratom of 22 February 1994(7) was to improve the exhaustiveness of GNP. The terms of reference, which are to be taken into consideration by the Member States(8), are shown in the following figure.

>PIC FILE= "C_2000336EN.001401.EPS">

9. This decision simultaneously aims to make the calculation of GNP more transparent while providing audit trails which make it possible to appraise the plausibility of certain components of GNP(9).

The distinction between final use and intermediate consumption

10. The added value of an economy, from which GDP is derived, is equal to the value of goods and services produced, minus intermediate consumption. It is therefore necessary to distinguish the final use of goods and services from uses included in the production process.

11. This distinction is a basic operation in the production of national accounts. Incorrectly allocating the value of goods and services has a direct effect on the value of GDP. For example, depending on whether intermediate consumption is under- or over-valued, GDP will be over- or under-valued.

12. The concepts of intermediate consumption and final use are defined by the ESA(10). Where it thought it necessary, the GNP Committee(11) introduced supplementary classification criteria.

(1) For example, small producer units are sometimes exempt from the obligation to supply information to the tax and social security authorities, or the information in question is not sent to the statistical offices automatically. In this case, the statistical offices' information on registered and economically-active units may be incomplete or obsolete. It may be noted in this connection that the proportion of enterprises with fewer than 10 employees and, at the same time, the parallel risk of not being taken into account by the statistical offices, may vary considerably from one country to another.

(2) This is true of clandestine production units, or registered taxpayers who declare to the tax authorities only part of the taxable added value.

(3) The concept of the underground economy is defined in many ways. At international level, one definition is given by the United Nations' system of national accounts (paragraphs 6.34 to 6.36). The ESA has not defined the term. The methods used by the various countries rely on both the statistical data available and the degree of exhaustiveness of the measurement under investigation. Several Member States have opted for an approach based on differences in recording error between sources (consumption or production data/fiscal data), but they have added a plethora of ad hoc corrections by means of specific enquiries that are subject to statistical secrecy, the use of tax-adjustment statistics and model-based estimates of production (e.g. by using yields in the agricultural sector).

(4) For example fiscal sources, periodic enquiries, police enquiries and administrative sources.

(5) By way of illustration, the accounts which followed the earlier standards of the United Nations' system of national accounts concentrated on the recorded economy. Although not exhaustive, they were reliable. If various sources are used, reconciliation becomes necessary, resulting in additional use of models, which in turn leads to loss of precision, inadequately controlled reduction of bias and, ultimately, lower quality.

(6) First report on the implementation of Directive 89/130/EEC, Euratom, SEC(92) 588 final of 31.3.1992, p. 3. Commission report on the follow-up to the comments accompanying the Council's recommendation concerning the discharge to be granted to the Commission in respect of the implementation of the general budget of the European Communities for the financial year 1992, 23.1.1995, Annex I, pp. 1 and 3.

(7) OJ L 77, 19.3.1994, p. 51.

(8) See Article 2 of this Decision. "Primary income", ESA 95 (paragraph 8.22) "is the income which resident units receive by virtue of their direct participation in the production process, and the income receivable by the owner of a financial asset or a tangible non-produced asset in return for providing funds to, or putting the tangible non-produced asset at the disposal of, another institutional unit".

(9) The Member States were required to produce a description of all calculations and adjustments deemed necessary in order to take account, depending on the approach used in calculating GNP, of factors that cannot be directly observed. They were also to perform a validation test by means of employment data, carrying out a survey of the components of income in kind and of tips or gratuities, and a report on the feasibility of using information from fiscal audits to improve the exhaustiveness of the national accounts. A fifth factor, specifically linked to the treatment of VAT fraud in the national accounts, was provided for by Commission Decision 98/527/EC, Euratom of 24 July 1998 (OJ L 234, 21.8.1998, p. 39).

(10) ESA 79, paragraphs 320 to 335. ESA 95, paragraphs 3.69 to 3.99.

(11) See paragraph 14 in the text of the report.

Annex II

Principal characteristics of the calculation of GNP by the Member States

>TABLE>

The Commission's replies

EXECUTIVE SUMMARY

The Commission is fully aware of the need to guarantee the quality of the Member States' gross national product (GNP) data, bearing in mind the vital role assigned to this aggregate in the Community context. It is for this reason that, under the auspices of Eurostat (Statistical Office of the European Communities) and in conjunction with the representatives of the Member States in the GNP Committee, it has made strenuous efforts for more than 10 years to improve the reliability, comparability and exhaustiveness of GNP data.

The main stages of this action were as follows:

- preparation of the GNP Directive as the legal framework for GNP harmonisation, involving the setting up of the GNP Committee of the Member States' most experienced national accountants and statisticians under the chairmanship of Eurostat,

- preparation by the Member States, with financial backing from the Commission, of an inventory of their GNP sources and calculation procedures,

- examination of these inventories by Eurostat and the GNP Committee. In conjunction with Eurostat evaluation missions, this examination led the Commission, with the Committee's approval, to seek to clarify the relevant European system of integrated economic accounts (ESA 79) and ensure the improvement of the GNP data. It also prompted the Commission to inform the Member States of the problems to be resolved in order to guarantee the reliability, comparability and exhaustiveness of their GNP data (GNP reservations). These reservations are designed to protect Community interests and ensure impartiality among the Member States,

- Commission financial support for, and Eurostat/GNP Committee monitoring of, the improvements made to GNP data by the Member States in response to the reservations. In an attempt to make this follow-up even more transparent, representatives of certain Member States participated in the information missions conducted in most of the others on the finalisation of the principal reservations, particularly that concerning the exhaustiveness of GNP data.

The Commission thinks that the regular checks conducted by its departments and the aforementioned activities have made it possible to improve the quality and achieve a sufficient degree of comparability of the GNP data of the Member States.

Whilst significant progress has already been made, improving the quality of national statistics and accounts is, by its very nature, an ongoing process. Adequate resources must be made available to enable the relevant activities and research to continue.

The Commission recognises that further efforts will be required to improve both the description of the checks conducted and the documentation of the corresponding dossiers.

Nevertheless, it feels that the specific comments of the Court of Auditors on the Member States investigated do not call into question the validity of the GNP data and their evaluation by the Commission. The following detailed responses reflect the importance of these comments.

THE USE OF GNP AT COMMUNITY LEVEL

6 to 12. The Commission is aware of the importance of the reliability of GNP and GDP data for different Community policies. It was for this reason that it initiated the introduction of machinery which would allow it to monitor these aggregates. The Member States verify the efficiency and transparency of this machinery through their participation in the GNP Committee.

THE RELIABILITY AND COMPARABILITY OF THE DATA

The control procedures

Observations on the reservations

Exhaustiveness of GNP

Transparency of adjustments

32 and 33. Most of the economy is covered by reliable statistical or administrative sources. In addition, the Member States carry out numerous adjustments, which take account of the underground economy, in an effort to ensure the exhaustiveness of their GNP data. It is also true that the explanations given for these adjustments are sometimes short on detail. This is primarily because, by definition, statistical sources in respect of the hidden economy are not directly accessible. As a result, expert assessments are sometimes the only source available to the different national statistics institutes in making adjustments. It must, however, be stressed that only a very limited proportion of the GNP is subject to this approach.

The Commission will continue to urge the national statistics institutes both to replace expert assessments by improved statistical sources wherever possible (especially where statistical information is not used on grounds of unreliability or inconsistency) and to provide better information on adjustments.

Failure to follow up a classification problem and acceptance of an adjustment to ensure exhaustiveness lacking a reliable statistical basis

34. The enquiry and discussions conducted by Eurostat revealed that the four Member States investigated (Germany, France, Italy and the United Kingdom) apply procedures or adjustments which make possible the satisfactory resolution of the classification problem raised by the Court.

In 1988, i.e. prior to the Decision on exhaustiveness, Germany in fact conducted an adjustment for which there was no reliable statistical basis. Such a situation inevitably led the Commission to enter a reservation with regard to exhaustiveness. Since Germany took the action required by the Decision on exhaustiveness and extrapolated the results back to 1988, the adjustment in question no longer plays any part in GNP assessment.

Comparability bias

Comparability of adjustments between Member States

35. The individual Member States have always adopted their own procedures for including the underground economy in the calculation of GNP. This is because of differences in their statistical sources and political and administrative systems, which mean that certain approaches can be adopted in some countries but not in others.

In this context, the Commission sees dangers in comparing the result produced by a particular national method with its putative impact in another country. In accordance with the Decision on exhaustiveness, the Commission checks the comparability of recognition of the underground economy in the different Member States from various standpoints including, in particular, application of the common methodology provided for by the Decision.

Treatment of illegal activities

36. Illegal activities were excluded from the scope of adjustments to ensure exhaustiveness and the GNP calculations because, during the discussion of Decision 94/168/EC, Euratom, the Member States instructed their GNP Committee representatives to point out that they lacked the statistical and methodological resources required for the assessment of such activities. Furthermore, the limited experience available in this area proved to be very piecemeal and to vary greatly from one country to another. The Committee therefore decided to focus its efforts on the definition and implementation of national procedures designed to ensure the exhaustiveness of the GNP data by including activities which, whilst in themselves legal, were neither declared nor directly observed (e.g. undeclared employment).

The essential problem from the standpoint of the GNP is that the principal activities customarily referred to as "illegal" are extremely difficult to quantify in practice, regardless of their actual legal status in any one country.

With a view to promoting the development of reliable methods of evaluating these activities, the Commission co-financed certain activities of the United Kingdom Office for National Statistics (ONS) in 1998 together with a seminar on this topic for the Member States, which was held in the Netherlands in September 1999. The initial results of the ONS activities were presented at the seminar.

The conclusions of the seminar, which were announced at the meeting of the Working Group on national accounts(1) in December 1999, represent the first contribution of Eurostat and the Member States to the manual on the underground economy which is being prepared by the OECD. Once this manual and the results of current research in certain Member States become available, Eurostat will submit proposals for the evaluation of illegal activities.

Outdated sources

37 to 40. Some of the coefficients used to calculate the adjustments to ensure exhaustiveness are undoubtedly based on outdated sources.

Eurostat is conscious of the problem which this state of affairs creates for the goal of achieving exhaustiveness. That is why, during its evaluation visits, it has stressed the need to use the most up-to-date sources.

Taking account of VAT evasion

Absence of proof of taking account of VAT fraud or of the equivalence of the methods used by the Member States

41 to 44. VAT fraud in the targeted Member States can be implicitly taken into account through the use and cross-checking of expenditure data.

Anomalies in the results of the calculation of theoretical VAT

45 to 47. The Court notes that, in Luxembourg, theoretical VAT (i.e. the product of VAT rates and taxable transactions) fell short of VAT revenue, suggesting that certain transactions had not been taken into account in the national accounts.

The Commission believes that the auditing of the Luxembourg national accounts has resolved the problem of negative VAT discrepancies. Indeed, the VAT discrepancies calculated on the basis of the national accounts transmitted in October 1999 were positive.

Business register

48 and 49. In Germany, the reservation in respect of exhaustiveness has been lifted, even though this country still has no integrated business register, which is a particularly important instrument from the standpoint of exhaustiveness.

A reliable register is clearly an extremely valuable tool in the context of measures to improve exhaustiveness. Germany is now compiling such a register pursuant to the deadline laid down in the derogation granted by the Statistical Programme Committee.

The exhaustiveness exercise conducted in Germany was essentially based on a comparison of employment data. This exercise is specifically designed to offset any shortcomings in exhaustiveness arising from the lack of a reliable register, since the demographic data used to validate the GNP assessments are not derived from existing registers.

Distinction between final uses and intermediate consumption

Balancing procedures

51 and 52. Eurostat accepts the Court's argument that balancing a product's uses and resources does not, in itself, allow the automatic correction of a bias affecting the two data sources simultaneously. Nevertheless, Eurostat believes that, if sufficiently detailed and based on reliable reference sources, such a balancing exercise would provide an excellent foundation for the formulation of valid estimates. It is for this reason that Eurostat stresses, and will continue to stress, these two factors in the context of its evaluation missions.

Furthermore, in cases where both sources are simultaneously subject to bias, Eurostat asks the national accountants to investigate alternative approaches, such as the comparison of employment data, which often provide a satisfactory solution.

The limited nature of the checks

54 to 56. For more than 10 years, all the work carried out by the Commission in conjunction with the Member States in this field has been aimed at ensuring the quality of GNP data.

The inventories first transmitted by the Member States in 1991, on which the Commission and the GNP Committee base their activities, describe the national statistical systems and the statistical sources and accounting procedures used in each country. In the course of its checks, Eurostat assesses the extent to which these elements contribute to the reliability of the GNP estimates.

Whilst these inventories are extremely detailed in certain respects, they have been found to be insufficiently detailed in others.

For example, they were drawn up before the adoption of the Decision on exhaustiveness and the recommendations of the GNP Committee on the distinction between intermediate consumption and final use. As a result, investigation of these two factors revealed a need for the Member States to take further action and provide additional information. The information requested was transmitted with the national reports on exhaustiveness and in documents and explanations provided by the national statistics institutes during and after the Eurostat evaluation missions, particularly from 1995.

By its very nature, the evaluation of national accounts and the GNP involves continuous contacts with the national statistics institutes, which implies a need for prior analysis of the available documentation followed by requests for, and consideration of, additional information on specific points.

In this connection (paragraph 56 also), reference should be made to Eurostat's comments on the "Measurement of the quality of national GNP data" (paragraphs 63 to 79).

57. Pursuant to the GNP Directive, Eurostat conducts its GNP assessments with reference to the reliability of the available statistical sources, the comparability of the rules applied (based on ESA compliance) and the results obtained and exhaustiveness (cross-checking of different sources, existence of a set of adjustments considered satisfactory).

These factors are analysed in Eurostat's reports to the GNP Committee, with reservations being notified and followed up if Eurostat considers that any of the three criteria have not been satisfied. Thereafter, in the light of the action taken and corrections made to their GNP data by the Member States, Eurostat decides whether one or more reservations may be lifted. At its request, the reports on these activities and corrections supplied by the Member States in October 1998 were sent to the Court of Auditors. Eurostat's analysis of the possibility of lifting or maintaining the reservations applicable to the four target Member States was also transmitted to the Court.

Nevertheless, the Commission agrees with the Court that further efforts will be required to improve both the description of the type of checks conducted and the documentation of the evaluation dossiers.

58 and 59. As regards the significance or irrelevance of any factor used in the calculation of GNP, Eurostat did not think it realistic to issue an a priori definition of what should be considered significant, for example by specifying a percentage GNP figure. It adopts its position on a case-by-case basis by making a professional assessment of the relative importance for statistical calculation of the economic phenomena in question.

From the cost-effectiveness standpoint, Eurostat thinks that an a priori definition of what should be regarded as significant for the GNP evaluation would require the calculation of every small item involved, thereby necessitating the commitment of disproportionate resources. It believes that, in this particular case, such action would infringe the cost-effectiveness principle.

The role of the GNP Committee

60 to 62. The Commission notes the Court's observation that the GNP Committee is now in possession of all the necessary information (supplied by the Member States and Eurostat) to make an overall assessment of the comparability and reliability of the GNP data in its opinion on the GNP.

In its earlier opinions, the GNP Committee confined itself to noting the availability of these data. In its current annual GNP opinion, it expresses the view that considerable progress has been made in harmonising GNP data, that is in implementing the GNP Directive.

In reality, the GNP Committee receives and analyses documentation from the Member States together with Eurostat reports and the latter's proposals for the improvement of GNP data.

The GNP Committee has discussed the type of opinion it should deliver on several occasions. So far, it has taken the view that it is for the Commission to comment on the quality of each Member State's GNP data. Nevertheless, the Commission intends to propose that the Committee re-examine this question in the near future.

The possibility of measuring the quality of the data

63 to 79. The Commission believes that, in conjunction with the activities of the GNP Committee, the regular checks of compliance with accounting rules and the improvement of statistical sources conducted by its departments are sufficient to ensure the correct and satisfactory application of the GNP Directive.

The Commission is open to proposals regarding new methods of checking the quality of GNP data. Nevertheless, both the Commission and the GNP Committee have already expressed reservations with regard to certain conceptual aspects and the viability of mathematical comparability indicators.

The Commission is aware of the theoretical importance of the measurement of reliability by confidence intervals, even though this approach might have no impact on budgetary practice. After reporting to the GNP Committee in July 1999 on the results of the studies it had assigned to Istat and the ONS (CPNB/272), the Commission set up a quality-assessment task force comprising representatives of six Member States. A notable feature of this task force is the presence of highly experienced national experts responsible for evaluating the relevance and feasibility of the proposed approaches.

The Commission recognises the importance of some of the issues raised by the Court, especially the difficulties associated with error measurement and the comparability of the solutions examined. The task force will consider these questions.

The Commission will put forward proposals in the light of the task force report, which is due in the first half of 2001.

CONCLUSION

An aggregate that is widely used in the Community context

Guaranteeing legitimacy

80 and 81. The Commission is fully aware of the need to guarantee the quality of the Member States' national accounts and GNPs given the crucial role assigned to the GDP and GNP in the Community context. To this end, the Commission, working in close collaboration with the Member States, has made strenuous efforts for more than 10 years to improve the reliability, comparability and exhaustiveness of this aggregate.

82 to 84. The transparency of the GNP evaluation machinery is ensured through the activities of the GNP Committee, which is made up of representatives of the Member States under the chairmanship of Eurostat. Inventories of national statistical sources and procedures, Eurostat reports (on the inventories and on missions analysing the reliability, comparability and exhaustiveness of individual national accounts) and the results of Member States' action on GNP reservations are transmitted to the GNP Committee for discussion.

Completing harmonisation

Effective controls at Community level

85 to 91. The Commission considers that, in conjunction with the activities of the GNP Committee, the regular checks conducted by its departments into the reliability and suitability of statistical sources, the comparability of definitions, rules (based on ESA compatibility) and results and the exhaustiveness of national accounts (cross-checking of different sources, existence of a set of adjustments considered satisfactory) are sufficient to guarantee the correct and satisfactory application of the GNP Directive.

The improvements made to their GNP data by the Member States in the wake of these checks and activities have justified the lifting of GNP reservations. The document referred to by the Court in paragraph 86 (CPNB/267, "Minutes of the meeting of the GNP Committee held in February 1999", point 11.2) sets out the response of the GNP Committee to a draft statement by the Statistical Programme Committee concerning the quality of national accounting data. The GNP Committee was anxious to emphasise the ongoing nature of the process of improving the quality of national statistics and accounts. To this end, it stressed the need to employ measured language so as not to create the impression that action to improve GNP data had been completed. On the other hand, the Committee did not question the validity of the GNP data on which the lifting of the reservations was based.

The Commission acknowledges that further efforts will be required to improve both the description of the checks conducted and the documentation of the corresponding dossiers.

92. The Commission can only endorse the Court's call for own resources to be based on ESA 95, particularly as it proposed this solution at the time of the adoption of ESA 95 in 1996.

(1) The Working Group on national accounts is made up of experts from Eurostat and the Member States.