52006DC0350

[pic] | COMMISSION OF THE EUROPEAN COMMUNITIES |

Brussels, 29.6.2006

COM(2006) 350 final

REPORT FROM THE COMMISSION TO THE EUROPEAN PARLIAMENT AND THE COUNCIL

Quality report according to Regulation (EC) No 1221/2002 of the European Parliament and of the Council of 10 June 2002 on quarterly non-financial accounts for general government

REPORT FROM THE COMMISSION TO THE EUROPEAN PARLIAMENT AND THE COUNCIL

Quality report according to Regulation (EC) No 1221/2002 of the European Parliament and of the Council of 10 June 2002 on quarterly non-financial accounts for general government

Regulation (EC) No 1221/2002 of the European Parliament and of the Council of 10 June 2002 [1], states, in Article 8, that the Commission (Eurostat) shall submit to the European Parliament and the Council a report containing an assessment of the reliability of quarterly data delivered by Member States.

The report by the Monetary Committee on information requirements, endorsed by the Ecofin Council of 18 January 1999, underlined the need for a comprehensive statistical information system to support policy-makers’ decisions. In this context, high priority was given to short-term public finance statistics with the aim of compiling a simplified version of quarterly non-financial accounts for the general government sector according to ESA 95 rules. A two-stage approach was followed. The first stage was the adoption of Commission Regulation (EC) No 264/2000 of 3 February 2000[2]. In a second stage, it was established in Regulation (EC) No 1221/2002 that data relating to the remaining ESA 95 transactions were to be transmitted[3].

This Quality Report includes a general section, and also provides a summary of conclusions and recommendations by country. The general section explains the underlying concepts and technical issues and summarizes the country section. It should be noted that a more extensive review will be available on the Eurostat Website in the course of 2006. The aim of the report is to capture the multiple dimensions of quality in statistics, and it follows the main criteria commonly used to assess quality[4]. Its structure was discussed and agreed in broad terms with the members of the Working Group on Short-Term Public Finance Statistics.

Structure of the Quality Report

1. Institutional arrangements

2. Accessibility and clarity

3. Timeliness and coverage of data transmission

4. Coherence

4.1. Coherence between quarterly and annual data

4.2. Coherence between provisional and final statistics

5. Comparability over time

6. Accuracy and reliability

6.1. Coverage of data sources

6.2. Methodologies and assumptions used in the estimates of statistics

6.3. ESA 95 conceptual adjustments

6.4. Revisions of statistics

7. Conclusions and recommendations

Quality Report

1. INSTITUTIONAL ARRANGEMENTS

Quarterly non-financial accounts for general government are compiled by the National Statistical Institutes in all Member States, except for Belgium (National Bank of Belgium on behalf of the National Accounts Institute) and Spain (Ministry for Economic Affairs and Finance). In a number of Member States, working groups made up of representatives of the National Statistical Institute, National Central Bank and Ministry of Finance deal with quarterly methodological issues (a similar approach to that adopted for annual accounts). Moreover, these working groups often analyse consistency between quarterly non-financial and financial accounts[5].

2. ACCESSIBILITY AND CLARITY

Accessibility refers to the physical conditions in which users can obtain data, i.e. where to go, how to order, delivery time, and formats for accessing the data. It also indicates whether data are accompanied by appropriate metadata and methodological manuals. Eurostat is currently publishing Member States’ data on taxes and social payments, which are covered by Commission Regulation (EC) No 264/2000, on its website. Remaining quarterly ESA 95 transactions will be available to the public at the beginning of 2006.

As for dissemination policies at national level, the situation is as follows: 12 Member States - Czech Republic, Denmark, Spain, Italy, Cyprus, Malta, the Netherlands, Austria, Slovenia, Finland, Sweden and United Kingdom - currently publish quarterly non-financial accounts for general government, albeit only partially in some countries (i.e. mainly quarterly taxes and social payments). A number of other Member States - Lithuania, Luxembourg, the Netherlands, Poland and Sweden – plan to publish or to further enlarge the published data set in 2006. Published statistics are usually available on the websites of the National Statistics Institutes, mainly in their national languages. Data are sometimes accompanied by metadata or brief methodological notes.

Eurostat has published two manuals on the methodology of quarterly government accounts. The first edition, Manual on Compilation of Taxes and Social Payments on a Quarterly Basis[6] was published in 2002 and concentrated on the transactions included in Commission Regulation (EC) No 264/2000. This Manual was supplemented by a new Manual on quarterly non-financial accounts for general government that covers all of the remaining ESA 95 transactions. It has a similar structure, providing methodological guidance and describing national practices for each of the sub-sectors of general government. This Manual will be available to the general public on the Eurostat website at the beginning of 2006.

It should be noted that under the terms of the two Regulations on short-term public finance statistics, Member States are required to provide Eurostat with a description of the sources and methods used to compile quarterly data and to provide details of any revision of the initial description when communicating the revised data. The only Member State that has still to provide a description of sources and methods for transactions covered by Regulation (EC) No 1221/2002 is Ireland.

3. TIMELINESS AND COVERAGE OF DATA TRANSMISSION

Timeliness reflects the time gap between data availability and the event it describes. According to the Regulations on short-term public finance statistics, data shall be delivered to Eurostat not later than 3 months from the end of the quarter to which the data relate.

Examining the quarterly transmissions from December 2004 to September 2005, fifteen Member States reported quarterly non-financial accounts for general government without delays: they were Denmark[7], Germany, Greece, Ireland, Italy, Latvia, Luxembourg, Malta, Netherlands, Portugal, Slovenia, Slovakia, Finland, Sweden and United Kingdom. Other Member States (Belgium[8], Czech Republic, Spain, France, Cyprus, Hungary, Austria and Poland) had a delay of 3 to 5 days in at least one of their transmissions. Lithuania had an average delay of 5 days in all four transmissions. Eurostat is currently analysing data very recently received from Estonia. It should be pointed out that some of the countries (Germany, Slovenia and Sweden) transmit their data up to 20 days in advance, depending on the quarters.

Regarding the coverage of data transmission, for taxes and social payments (Article 5 of Regulation (EC) No 264/2000) Member States are required to deliver quarterly data to Eurostat starting from the first quarter of 1991. For the new Member States, data starting in 1995 are requested, in the context of the revised ESA 95 Transmission Programme[9]. For the remaining ESA 95 transactions (Regulation (EC) No 1221/2002, Article 6), quarterly data should be transmitted starting from the first quarter of 1999.

Greece, Spain, Ireland, Luxembourg and Sweden still need to report back series. For the other ESA 95 transactions (Regulation (EC) 1221/2002), most countries comply with the requests. However, for Cyprus, Hungary and Malta, the missing time series concern the period 1999-2003 (for Malta, only 1999 is missing). For Slovenia, the coverage of a number of ESA 95 transactions is low. The most serious gaps in coverage are found in Cyprus and Hungary.

4. COHERENCE

Coherence of statistics relates to their suitability to be reliably combined in different ways and for various uses.

4.1. Coherence between quarterly and annual data

In both Regulations it is stated that the quarterly data and the corresponding annual data shall be consistent (Articles 3 and 4 respectively). When analysing the consistency between quarterly and annual figures, it is possible to conclude that some inconsistencies remain in certain specific transactions for Denmark, Luxembourg, Poland, Sweden and the United Kingdom before 1999.

In the data for 1999 and after, inconsistencies are found for the following Member States: Czech Republic for several ESA 95 transactions (for 2003 and 2004), Denmark (for D211 in 2003), Ireland (mostly for P2 and D4, for all the time-series), Poland (several inconsistencies before 2001), and the United Kingdom (inconsistencies mostly for D4, D5 and D7 for all the time-series). Slovakia is the Member State with the most inconsistencies between quarterly and annual data; this can be largely explained by the fact that is not consolidating D4, D7 and D9 on a quarterly basis for the period 2001-2004.

4.2. Coherence between provisional and final statistics

The average revision of a value after two years, i.e. after eight quarters, was analysed in both absolute and nominal values to assess whether the final statistics differ significantly from the provisional ones. It should be noted that this calculation was possible only for those Member States that have reported at least eight quarterly transmissions for each of the ESA 95 transactions[10].

For those Member States that show significant revisions in this period, reference is made to the ESA 95 transactions that have significantly influenced either total expenditure or total revenue (or both). This indicator helps in assessing the impact of revisions, particularly the volatility of net lending/borrowing B.9, which is a prominent balancing item[11]. Several factors affect revisions; for quarterly statistics these are related not only to the availability of final annual accounts but also to other factors such as updating, quality reviews, changes in data sources or methods, ESA 95 conceptual adjustments, etc. One important aspect of both Regulations is that direct sources must be used for compiling non-financial accounts for general government and no seasonally adjusted figures are to be transmitted to the Commission ( raw data ).

In summary, where data availability allowed this analysis to be carried out, it was observed that the absolute average revision of B.9 varies between 0.2% and 0.4% of Gross Domestic Product (GDP) for Belgium, Germany, France and United Kingdom. A second group of Member States - Denmark, Spain, Netherlands, Portugal and Finland - shows an absolute average revision over two years of between 0.6% and 0.9% of GDP. A third group shows an absolute average revision higher than 1% of GDP; this group includes Greece (3.2%), Italy (1.5%), Latvia (2.1%), Luxembourg (8.1%), Austria (1.1%) and Sweden (1.2%). The arithmetic average revisions of the values two years later tend to show a lower impact of revisions. The Member States presenting the most significant average revisions over this period have been Greece (-3.2%)[12], Luxembourg (5.7%) and Sweden (1.2%).

The aim in compiling those statistics would be to minimize, for each quarter, the differences between the initial estimates and final figures. Although no similar quality assessment is being made for annual statistics, since 1999 the non-weighted average of absolute revisions in the annual deficit ratios of Member States has been 0.27% of GDP after six months, 0.34% after one year and 0.43% after 18 months.

5. COMPARABILITY OVER TIME

The purpose of comparability is to measure the impact of differences in applied statistical concepts and measurement tools/procedures when statistics are compared between geographical areas, non-geographical domains, or over time. Regulation (EC) 1221/2002 provided a transitional period for those Member States which were not able to use the sources and methods and/or to follow the timetable laid down in the Regulation. “Best quarterly estimates” were to be transmitted initially, gradually incorporating new information that became available during the process of compiling an improved system. The transitional period ended on 31 March 2005 (Article 9(4)).

Most Member States have not reported breaks[13] in the time-series during the transmission period. Exceptions were the Czech Republic (new quarterly data sources were incorporated starting in 2003), Spain (new benchmark year starting in 2000), the Netherlands (benchmark revision between 2000 and 2001; incorporation of new data sources), Poland (new methodology applied in 2004) and Portugal (new benchmark year starting in 2000).

6. ACCURACY AND RELIABILITY

6.1. Coverage of data sources

Regarding the use of data sources, Article 3 of Commission Regulation (EC) No 264/2000 states, on the subject of taxes and social payments, that quarterly data shall be based on direct information available from basic sources, such as for example from public accounts or administrative sources, representing, for each category, at least 90% of the amount of the category. Regulation (EC) 1221/2002, which covers all the remaining ESA 95 transactions, specifies that quarterly data shall be based as far as possible on direct information from basic sources, with the objective of minimising, for each quarter, differences between the initial estimates and the final figures, when relevant.

Most Member States meet the 90% criterion for taxes and social payments, particularly for taxes (D2, D5, D91). The social benefit category (D62) causes more difficulties, in particular for Belgium (87%), France (85%) and Ireland (75%).

The direct information requirements in Regulation (EC) 1221/2002 are not as strict, and it is stated that quarterly data shall be based as much as possible on direct information. The situation is different from one Member State to another, and some of them - Belgium, France, Ireland, Latvia, Lithuania, Luxembourg, Hungary, Portugal and Sweden - point to difficulties in using direct basic sources for compiling quarterly ESA 95 transactions.

6.2. Methodologies and assumptions used in the estimates of statistics

Both Regulations specify that direct information shall be completed by coverage adjustments, if needed, and by conceptual adjustments in order to bring quarterly data in line with ESA 95 concepts. Adjustments are needed in all Member States for estimating quarterly non-financial accounts for the general government sector, especially for those ESA 95 transactions covered by Regulation (EC) 1221/2002. The problem of data availability is mostly related to local government (for example, in Belgium, Luxembourg, Austria, Portugal, United Kingdom) and/or to specific units in general government.

The extent to which statistical models are used for estimating missing figures varies across Member States. These models have been developed in Belgium, the Czech Republic, France and Italy using basic sources as indicators and/or forecasts. Some countries adopt a mixed approach based on the codification of quarterly basic sources into ESA 95 transactions, similar to the methodology followed in annual accounts, with missing figures being estimated.

6.3. ESA 95 conceptual adjustments

As stated above, conceptual adjustments are to be made in order to bring quarterly data in line with ESA 95 concepts. Most of the adjustments made in quarterly accounts of general government are similar to those made in annual accounts. They follow ESA 95 principles as regards time of recording, but whenever basic sources are on a cash basis, accrual adjustments are implemented (mainly for taxes and social contributions).

6.4. Revisions of statistics

In order to provide a quantitative analysis on revisions, the indicator used was the revision in the fourth quarter of the three preceding quarters from 2002 to 2004, for each of the ESA 95 transactions. The findings below describe the impact of revisions in B.9 (Net lending/net borrowing).

For Belgium, the Czech Republic, Denmark, Germany, Spain, France, Cyprus, Latvia, Lithuania, the Netherlands and the United Kingdom, the absolute average revision of the three preceding quarters conducted in the fourth quarter changed B.9 by between 0.1% and 0.5% of GDP. For Italy, Portugal and Finland these revisions accounted for 0.7%, 0.6% and 0.9% of GDP respectively. The countries with the highest average revisions on B.9 were Greece (1.9% of GDP), Luxembourg (3.4%), Hungary (2.8%), Austria (1.1%) and Poland (1.0%)

As B.9 is a balancing item, an assessment was also made as to whether revisions in the fourth quarter affected total expenditure and revenues of the three preceding quarters differently. For Greece both total expenditure and revenue were on average significantly revised (3.3% and 2.8 % of GDP respectively). For Luxembourg and Austria total expenditure was revised more significantly than revenue, while the opposite was the case for Hungary. In Poland, the scale of revisions made in expenditure and revenue was almost the same (2.4% and 2.6 % of GDP respectively). The bias when analysing the arithmetic average values is less than 1% for almost all countries, except Hungary (2.3%).

The magnitude of these revisions is not very significant, given that eleven Member States have revised the three preceding quarters by between 0.1% and 0.5% of GDP on average. The experience of compiling these statistics will probably lead to lower revisions over time.

7. CONCLUSIONS AND RECOMMENDATIONS

SIGNIFICANT PROGRESS HAS been achieved and quarterly non-financial accounts for general government have been further developed since the first transmissions. Data for most countries are of appropriate quality and useful for analytical purposes, and they should accordingly be disseminated. Some Member States must implement specific measures for improving the quality of the reported data. In some cases, the improvement could be achieved by complying with the Community obligation of reporting time series to Eurostat. This would, on the other hand, prevent the Commission from initiating an infringement procedure under the Treaty in order to ensure full respect of Community regulations.

The observed revisions are similar to those of other short-term (quarterly) statistics. The publication of quarterly non-financial accounts for general government should be promoted, with appropriate guidance to users. This will mean providing appropriate metadata to reflect national practices and to warn users of the volatility of these quarterly figures, particularly for certain Member States. Publication will be a stimulus for quality, and Eurostat plans to disseminate these statistics at the beginning of 2006 in consultation with Member States. Experience shows that, at national level, users are becoming more familiar with these statistics, more accustomed to dealing with the volatility of quarterly statistics and are developing a better understanding of them over time.

It is recommended that this assessment be continued over time in order to monitor further progress achieved by Member States and to improve quality. Some more specific recommendations by Member State are described below:

- Belgium: further progress is needed as regards coverage of direct sources, particularly for D62, for which the 90% criterion is not fulfilled (87%).

- Czech Republic: consistency between quarterly and annual data must be ensured. Interest is to be consolidated on a quarterly basis over all time-series. The progress in gradually incorporating improved quarterly basic sources is welcomed.

- Denmark: no specific remarks, other than the need to ensure consistency between quarterly and annual statistics over time.

- Estonia: data have been transmitted very recently. Eurostat is currently analysing the information provided and will revert if quality problems are identified.

- Greece: it is recommended that basic sources be further developed and quality improved. The impact of revisions has been significant after two years, and also in the fourth quarter for the three preceding quarters.

- Spain: missing time-series must be reported to Eurostat, and an action plan needs to be drawn up in this respect.

- France: further progress is expected so as to increase the coverage of direct sources. For D.62, there is 85% coverage of direct information. For the other ESA 95 transactions covered by Regulation 1221/2002, direct sources are mostly used, among others, for D.1, D.4 (expenditure) and P.1.

- Ireland: the methodological description of the national practices for the ESA 95 transactions covered by Regulation 1221/2002 should be supplied to Eurostat. Further progress is expected in widening the coverage of direct sources. The 90% criterion of Regulation 264/2000 is not met for the social benefit category, D62 (75%). Consistency between quarterly and annual data must be ensured.

- Italy: further progress is expected as regards the coherence between provisional and final statistics.

- Cyprus: missing time-series should be supplied to Eurostat and an action plan must be drawn up for this purpose.. Further progress is expected in developing the availability of quarterly basic sources for local government.

- Latvia: further progress is to be achieved as regards coherence between provisional and final statistics.

- Lithuania: missing time-series have to be provided to Eurostat, and plans must be drawn up accordingly. The timeliness of transmissions should be further improved.

- Luxembourg: missing time-series have to be provided to Eurostat, and an action plan needs to be established. Further progress is expected in order to increase the collection of quarterly basic sources for local government. The impact of revisions has been significant after two years, and also in the fourth quarter for the three preceding quarters. Quality should be stepped up in this area.

- Hungary: missing time-series must be provided to Eurostat, and an action plan needs to be established. Timeliness in transmission should be further improved. Further developments are to be expected in the collection of quarterly basic sources from local government. The impact of revisions has been significant when analysing revisions made in the fourth quarter of the three preceding quarters and quality in this area should be reinforced.

- Malta: missing time-series must be provided to Eurostat, and an action plan needs to be drawn up.

- Netherlands: further progress is expected in order to gradually incorporate improved quarterly basic sources.

- Austria: the impact of revisions has been significant after two years and also in the fourth quarter for the three preceding quarters. Quality should be improved in this respect. Further progress is expected in order to gradually incorporate improved quarterly basic data sources.

- Poland: there is a need to ensure consistency between quarterly and annual data over time. There are significant inconsistencies for almost all variables for the years 1995-2000. Further improvement is expected in increasing the collection of quarterly basic sources. The impact of revisions has been significant in the fourth quarter for the three preceding quarters, and quality should be improved in this respect.

- Portugal: further improvement is needed in collecting quarterly data from regional and local government, and particularly in respect of collection for P.51.

- Slovenia: missing ESA 95 transactions have to be provided to Eurostat and plans to transmit this information in 2006 should be implemented.

- Slovakia: consistency should be ensured between quarterly and annual data and in particular D4, D7 and D9 should be consolidated. Further improvements are expected in estimating accrual data on a quarterly basis.

- Sweden: missing time-series have to be reported to Eurostat and an action plan needs to be drawn up. Further progress is to be achieved in collecting quarterly basic sources, particularly for social security funds. Further improvement is needed in consolidating D.4 both on an annual and on a quarterly basis. Efforts should be made to ensure coherence between provisional and final statistics.

- United Kingdom: there is a need to ensure coherence between quarterly and annual data over time, and to further develop the availability of quarterly basic sources for local government.

[1] In http://europa.eu.int/eur-lex/pri/en/oj/dat/2002/l_179/l_17920020709en00010005.pdf

[2] In http://europa.eu.int/eur-lex/pri/en/oj/dat/2000/l_029/l_02920000204en00040006.pdf

[3] The content of the quarterly non-financial accounts for general government was defined by reference to the list of ESA 95 transactions constituting general government expenditure and revenue.

[4] see Eurostat CIRCA Interest Group “Quality in Statistics”,http://forum.europa.eu.int/Public/irc/dsis/Home/main.

[5] In some Member States such as Belgium, Lithuania, the Netherlands or Luxembourg the same institution compiles both quarterly non-financial and financial accounts for general government.

[6] European Communities (2002), Manual on Compilation of Taxes and Social Payments on a Quarterly Basis, Luxembourg: Office for Official Publications of the European Communities.

[7] Denmark reported to Eurostat in advance both in December 2004 and March 2005.

[8] In the last three transmissions.

[9] The revised ESA 95 Transmission Programme is due to come into force in 2006.

[10] Average difference of the following: second quarter 2005 minus second quarter 2003, first quarter 2005 minus first quarter 2003, fourth quarter 2004 minus the fourth quarter 2002, third quarter 2004 minus the third quarter 2002, second quarter 2004 minus the second quarter 2002 and first quarter 2004 minus the first quarter 2002.

[11] The question of revisions is further analysed under item 6, accuracy and reliability.

[12] A minus sign means increasing deficit figures.

[13] Breaks in statistical time series occur when there is a change in the standards for defining and observing a variable over time. Such changes may be the result of a single change or the combination of multiple changes at any point in time of observation of the variable. The specific causes of breaks in a statistical time series include changes in classifications used, definitions of the variable, coverage, etc. ( in OECD, Glossary of Statistical Terms ).