Short-Term Statistics

STD/STESEG(2002)18

Shortterm StatisticsEurostat's role in the production of data

Gunter Schäfer

Eurostat D3

1.The role of Eurostat

The Eurostat unit D3 collects from the National Statistical Institutes (NSIs) of the European Union, EFTA and Candidate Countries for the EU shortterm statistics (STS) according to the Council Regulation1165/98 and calculates European aggregates based on this data. The STS Regulation came into force in June 1998.

The indicators covered by STS are:

For industry: Production, Turnover (total, domestic, nondomestic), New orders received (total, domestic, nondomestic), Number of persons employed, Hours worked, Gross wages and salaries, Output prices (total, domestic, nondomestic).

For construction: Production (total, buildings, civil engineering), New orders received (total, buildings, civil engineering), Number of persons employed, Hours worked, Gross wages and salaries, Construction costs (total, material costs, labour costs), Building permits (number of dwellings, square metres of useful floor area or alternative size measure).

For retail trade and repair: Turnover, Number of persons employed, Deflator of sales

For other services: Turnover, Number of persons employed

Most of the indicators are supplied on a monthly basis but some are quarterly although some Member states also supply monthly data for indicators for which there is an obligation to supply quarterly data. The level of detail may reach up to 4 digits of NACE for certain indicators.

The data supplied from Member States and/or generated within Eurostat derived from Member States data may have also different representations (unadjusted, working day adjusted, seasonally adjusted and trend cycle).

2.The calculation of European aggregates

Eurostat's prime objective is to calculate European aggregates for the EU and the Eurozone The principle user is the European Central Bank (ECB) but also other institutions that observe the development of the economic cycle.

Eurostat collects national data either in absolute values or in form of indices and calculates European figures through a weighted aggregation of the national data. It gives priority to the socalled direct method. Gross data or working day adjusted data is aggregated to a European time series, which is subsequently seasonally adjusted. National data that is missing for the aggregation is estimated using ARIMA forecasting.

Eurostat also applies a number of quality checks to the data and verifies the consistency between national results and European figures and between aggregates and details.

Eurostat publishes the European aggregates as well as the national series. In case working day and seasonal adjustments have been applied by the Member States, these figures are shown in publications. Only if the Member States do not apply the adjustments, Eurostat uses its own methods for data adjustments.

Eurostat is neither be seen as a pure producer, nor as a pure user of data although it takes part in the production process. It functions as an intermediary between the NSIs that produce national data and data users who need European aggregates for economic or monetary policy making.

All variables of STS have been collected in one form of another in most Member States even before the creation of the STS Council Regulation. Two main challenges were faced by Eurostat in 1998 when the Regulation went into force:

To complete the implementation of all STS variables in all Member States;
To harmonise the variables so that an aggregation of data with sufficiently similar meaning was possible.

For most variables, an implementation period of up to 5 years is foreseen starting in June1998. Today that about 4 years of these 5 years period has passed roughly 50% of the variables can confidently be judged operational in terms of data quality and coverage. Eurostat sees a good coverage achieved if roughly 80% of the total European weight is available in form of national data. For the remaining variables that cannot yet be judged operational, the implementation process is well under way in the in the Member States or concrete implementation plans exist.

In the last phase of the implementation period, the variables that concern nondomestic markets, industrial new orders, labour input, and the entire service sector need most attention.

The task of achieving a degree of harmonisation sufficiently high for good quality indicators is more challenging than the pure availability of data. Eurostat sees the harmonisation as a continuous process over a longer period of time.

A first major step in the harmonisation was achieved by a set of common definitions for the variables (Commission Regulation588/2001). These definitions clarified major aspects to be taken into account, such as handling of taxes, fees, reference periods, etc. without going deeper into methodological details. Further work on harmonisation is done in the form of a common methodological manual and supplementary Eurostat recommendations. These documents are currently discussed with the Member States.

Nevertheless, the level of harmonisation achieved until now on a European scale is still far away from the detailed guidelines of procedures manuals for the data collection and data processing for a certain indicator used in NSIs. The common European methodological manual and the supplementary recommendations still concentrate on rather fundamental aspects leaving the details to be determined by the NSIs.

3.Important aspects in the production of data

Besides focusing on the availability of data and the achievement of a base level of harmonisation, Eurostat has particular requirements before indicators can be considered operational.

3.1 Common and reliable release calendar

The recent work on the definition of principle European economic indicators (PEEI) carried out in the name of the Statistical Program Committee (SPC) of the European Statistical System has once again drawn the attention on timeliness as a major area in which improvements are urgently needed.

The calculation of timely European aggregates fully depends on the availability of the national data. A reliable publication planning for Eurostat that aims at a good timeliness is only possible, if a reliable publication planning for the Member States exists. The more accurate the planned release dates are, the smaller safety margins can be kept in the calculation planning of European aggregates. The creation of a European wide release calendar that is sufficiently reliable is one of the major achievements of STS in recent years.

Nevertheless, further progress is needed, in particular with respect to a coordination between the release dates of Member States. It is of little value if some Member States make particular efforts to reduce the delays for a certain indicators while others do not. The timeliness of European data can only be improved in a concerted effort by all or most Member States. Until now, publications by NSIs are determined mainly by national factors with little consideration for European objectives. The efforts to establish common timeliness targets for the PEEIs with shorter delays than foreseen in the STS Regulation, is suitable to improve this situation significantly.

3.2Long time series

Economic cycle analysts need for their models a period that includes at least 23 complete previous economic cycles, thus 10 to 15 years. Longer periods would be desirable.

Unfortunately, European aggregates are often very short. For STS, most time series start in 1990 or 1995 because of a number of reasons. The most important one is certainly the adaptation of a previously national indicator with a definition and methodology different from the European one. Some indicators have newly been created in the majority of Member States. The German unification is also an event that interrupted the continuity of existing series.

The need for specific efforts to backcast national time series, for example by using other information is not well recognised by the Member States.

Eurostat is facing the very difficult situation that it either has only few national time series available that extend backwards sufficiently long or that it performs the backcasting itself. Both options have serious disadvantages. The attempt to create long European time series already for a period for which only a few national series are available has the consequence of a very low coverage for a European aggregate at the start of the series. Moreover, level shifts in the series are inevitable when Member States data with a shorter length is aggregated into the longer European series. For the other option to do the backcasting itself, Eurostat is certainly in a worse position compared with the NSIs to backcast data by using other sources of information. In general, the NSI itself has the highest economic competence to carry out such operations.

3.3 Common quality criteria

Eurostat has until now very little information on the quality of the national data that is used to calculate European STS data. Only for the retail trade index, Eurostat and the NSIs have now finished a detailed study on the underlying quality of the national data and got a good idea on the quality level of the European data. In addition, a recent quality report showed the issue of the missing common quality concept for STS.

Eurostat is well aware, that the quality is a very comprehensive concept and cannot be reduced to just a few indicators. Nevertheless, such a common framework of quality criteria is necessary in order to provide the users at least with a minimal framework for judging quality.

Of particular concern for quality are revisions of data. Although, revisions are not an absolute measure of the data quality, they constitute a frequently used criterion for judging the reliability of early estimate. With the increasing pressure to improve the timeliness of STS data, the analysis of revisions will become increasingly important.

3.4 Availability of metadata

Users request from Eurostat the methodological background of the national data used in the integration. Ideally, there is a complete set of methodological information provided for each set of data. The problem of Eurostat and the Member States lies in the sheer volume of the necessary information involved and the frequent changes such methodological information undergoes. The daily practice of different ways meta data is presented, different languages, different styles, etc. makes the service of providing good meta data to the users extremely difficult to deliver on a regular and reliable basis.

The collection and dissemination of methodological data that is sufficiently complete and uptodate can only be achieved if the methodological descriptions are sufficiently harmonised, Eurostat develops adequate tools and the Member States provide the data regularly and report important events, such as changes in the methodology when they occur. The SDDS format for metadata developed by the IMF is a major step forward in the handling of metadata. Further efforts will be necessary to create a reliable flow of information on methodological metadata.

3.5 Reliable system of weights

Weights play an enormously important role for the aggregation of national data to European aggregates. Only when a reliable system that correctly reflects economic reality is available, can European aggregates be correctly calculated. Ideally, the weights used by Eurostat should be identical to the weights used on the national level.

Eurostat has invested a very strong effort to obtain a good system of weights. It consulted other available statistics, such as national accounts, structural business statistics and foreign trade statistics, in order to propose an initial system of weights to the Member States. In STS, the current weights system uses the reference year of 1995. At that time, most indicators were either not at all or differently calculated by the Member States. Wherever possible, the alignment with the national weights system was achieved in bilateral contacts between the NSIs and Eurostat.

In the forthcoming change of the base year to 2000, the STS system of weights will need to be revised substantially. Fortunately, the data basis for the base year 2000 is now much better developed that it was for the year 1995.

4.Challenges for the future

The European Statistical System is currently in the process of reviewing the STS Council Regulation. A number of important challenges must be met:

4.1 European data first

The introduction of the Euro and the awareness that a considerable part of the economic and monetary policy is now done on a European or Eurozone level has created a change of mind. The NSIs and Eurostat are aware that European data needs a higher priority for good timeliness than national data. The demanding timeliness requirements defined in relation to the PEEIs show the scope of the issue. The necessary changes in the way data is collected and aggregated are rather fundamental in nature.

The ESS needs to develop concepts of how to cope with this challenge. A number of potential ways exist, such as early estimates of national aggregates or a sampling that takes into account European requirements. A recent study on the retail trade turnover index showed the feasibility of such a concept. It defines a countrystratified European sample. Each Member State contributes to a sample that is optimised on a European scale. The delays for the European aggregates for retail trade can be reduced from 60 days to 30days using the countrystratified European sample. A specific Task Force of Eurostat and the NSIs aims of exploring the general principle further.

4.2 Complete set of indicators

Although STS is a significant step forward, it is for many economic analysts falling short of requirements in terms of available indicators. Already the ECOFIN Action Plan of 2000 identified separate Eurozone indicators and import prices as essential extensions to the list of indicators named in the STS Regulation.

Further indicators have been identified by Eurostat in cooperation with the NSIs. Of particular importance are improved statistics on services. To develop service output prices for corporate customers will be a major longterm project not only for the ESS but also for other countries. Eurostat will create a joint taskforce with the OECD in order to define the corresponding research project and principle methodological guidelines. Other indicators that dearly lack for the service sectors are labour input indicators that are essential, among other purposes, for productivity measures.

4.3 Harmonised European indicators

The gradual harmonisation that many indicators are currently undergoing will have to lead to a full harmonisation of certain indicators in the longerterm future. The example that due to its success will inspire similar efforts, is the HICP, the harmonised index of consumer prices. Together with the ESA 95, it shows both the benefits and the feasibility of fully harmonised indicators.

The European Statistical System will have to lead a discussion about the degree of harmonisation the various indicators will need in order to be most effective. The methodology of some indicators will certainly remain fully in the responsibility of the Member States while others will need a high degree of methodological harmonisation.

5.Conclusions

The STS Council Regulation of 1998 has been a major step forward in the process of providing European data on the economic cycle. The full implementation of the Regulation is the prime objective at this moment. Nevertheless, the discussion on the future evolution of the Regulation and the medium to longterm requirements has started. It is conducted in a cooperative spirit between Eurostat and the NSIs.

The task of Eurostat to collect national data and to calculate European aggregates induces special requirements on the provision of data and the cooperation with the Member States.