Directorate General Statistics

STD/STESEG(2002)9

ECB Requirements for Shortterm economic Statistics

Contribution to the OECD Expert Group Meeting on Shortterm Economic Statistics, 2425 June 2002

1.The ECB requirements for short-term economic statistics

Price stability in the euro area is the primary objective of the ECB’s monetary policy. Price stability defined as an annual increase in the Harmonised Index of Consumer Prices (HICP) for the euro area of below 2% is to be maintained over the medium term. The ECB strategy is based on two pillars: (1) the analysis of monetary growth, signalled by a reference value for the growth of a broad monetary aggregate (M3), and, (2) the analysis of a wide range of other economic and financial variables in order to form a broadly based assessment of the outlook for price developments and the risks to price stability[1].

The second pillar includes the analysis of a set of indicators which influence price developments in the shorter term. Such shortterm developments are relevant for monetary policy since their effects may spillover and determine the price developments in the medium term as well. The ECB therefore continuously reviews developments in the goods, services and labour markets.

Thorough and regular analysis requires sound and reliable statistics. Policy mistakes due to incomplete or unreliable statistics can be costly for the economy in terms of higher inflation and lower output and employment. The statistical requirements of the ECB on general economic statistics have been published[2] and can be summarised as follows:

the statistical needs of the Eurosystem (i.e. the ECB and the NCBs of participating Member States) are similar to those of other major monetary policy institutions;

the ECB’s main focus is on euro area aggregates, so the availability of statistics for all euro area countries, or at least a sufficient share, is essential;

a sufficient degree of harmonisation is necessary for building euro area aggregates.

Among the indicators with the highest priority are indicators on output and demand (e.g. industrial production, output in the service sector, retail trade turnover) and price and cost indicators (e.g. CPI, PPIr prices, labour costs). Many of these indicators coincide with the activity of the OECD Expert Group.

A recent assessment from the ECB concluded that significant shortcomings in the provision of comprehensive and reliable general economic statistics for the euro area exist[3]. The shortcomings concern the availability and timeliness of euro area aggregates in important areas such as the national accounts, labour market statistics and other shortterm statistics. In order to achieve progress, the ECOFIN Council[4] invited the European Commission (Eurostat), in close collaboration with the ECB, to establish an Action Plan on EMU Statistical Requirements[5] identifying for each Member State of the European Union (EU) where urgent progress should be made as well as possible modifications to existing statistical regulations. While not all countries will meet the EMU Action Plan for all indicators, it is hoped that the objective, namely 80% coverage of the euro area by the end of 2002, will be met. Annual progress reports asses the improvements achieved. Work towards better euro area statistics is also carried out in newly created “Friends of the Chair Group on key EU/EMU shortterm economic indicators” which prepares a proposal for a list of Principal European Economic Indicators similar to an existing standard in the United States.

The following discussions distinguish between euro area “domestic” issues and international aspects, the latter being the main focus of the new OECD group.

2.Statistics for the euro area – some issues from the ECB’s point of view

Timeliness and reliability

The timeliness[6] of euro area general economic statistics is – besides the nonavailability of many required harmonised variables – a key issue in the present discussion on euro area macroeconomic statistics. The situation has been criticised not only by the press and financial market participants; increasing pressure also stems from governments of the EU Member States, from the European Commission and from the ECB[7]. In the discussion, producers of general economic statistics (i.e. national statistical institutes) stress the tradeoff between timeliness and accuracy. The ECB, while aware of this tradeoff, argued that the existing large differences in the production schedules for national statistics in EU countries point to a significant potential to improve timeliness for areawide data without undermining the accuracy of the results and without replacing “actual” statistics by rough estimates.

Several different benchmarks may be used for assessing the timeliness of euro area statistics, e.g. the Special Data Dissemination Standard (SDDS) of the IMF, or the frequently quoted timeliness of statistics in the United States. A useful benchmark for defining the required timeliness for euro area statistics is the timeliness achieved in those EU countries with the best record today. This is the natural yardstick of the ECB and it is consistent with the EMU Action Plan.

Revisions in euro area aggregates

The EMU Action Plan put emphasis on the timeliness of economic statistics. However, improved timeliness of euro area data is only of benefit when it does not cause significantly higher revisions. Timeliness and reliability of statistical data are integral parts of the quality of the data.

Eurostat’s publication dates for euro area data depend on the timetables in the EU countries, which vary considerably. In the domain of shortterm indicators, Eurostat produces a first estimate when available national data cover at least 60% of the euro area (or EU). Revisions of euro area results may be mainly triggered by more countries reporting data for a certain period and national data being revised by NSIs.

The ECB recently compared revisions of euro area data and revisions of data for the EU countries for four key series of general economic statistics (industrial production, industrial producer prices, retail trade and production in construction) with the timeliness of the release of first results[8]. Some results were:

A general correlation between the timeliness of the data releases and the revisions is – in a comparison between countries – not evident. A similar performance in timeliness was achieved with different level of revisions.

Among the indicators analysed, the size of revisions was lowest (and often negligible) for industrial producer prices, more significant for industrial production and high or very high for retail trade and construction data.

Most EU country data show revisions of the latest values as well as regular and irregular revisions further back. As a result, the euro area aggregates were revised continuously over long time spans due to the uncoordinated revision practices in EU Member States – a disadvantage for economic analysis.

Furthermore, the punctual transmission of data by national producer, the timely dissemination of the data by the databank provider and a thorough data checking in all institutions involved appears to be a trivial request but is a prerequisite for highquality statistics which is not always met.

Countrystratified euro area sampling

Considering that even after full implementation of existing regulations and the EMU Action Plan a considerable gap between the releases of, for example, the United States, the most timely EU countries and the release of euro area statistics will remain, further efforts to reduce reporting lags are under way.

Following a proposal of the Eurostat Task Force on Benchmarking in infraannual economic statistics, the approach to compile euro area data by means of countrystratified sampling has been studied. The basic idea of the countrystratified euro area sample is to split the treatment of national sample data into two steps: first, the treatment of the data needed to compile a contribution for euro area aggregates; second, the treatment of the data needed to compile results for national aggregates and detailed sectoral results. This appeared to be the most promising way to combine the improvement of timeliness and the use of national expertise and experience, without overburdening national statistical systems.

Eurostat has set up an Expert Group which studied the approach on the example of the retail trade index; It concluded that the countrystratified European sample for retail trade results in a significant reduction in the sample size compared to the current total of all national samples. The drastic reduction in the sample size (by ensuring the requested degree of precision) may result also in an improved timeliness (from t+6070 days to t+30). The release of the results could be in two steps: (i) a new advance release would contain only results for the euro area as a whole, and (ii) the publication of results based on the complete, national samples would follow later, including the breakdown by national results. The implementation of the countrystratified sampling is planned by end2004.

These concepts are not specific to the retail trade index. It can be considered for other surveybased shortterm indicators as well and might offer possibilities for further improvements in other statistical domains.

Service statistics

Almost 70% of the GDP for both the euro area and the EU is produced by activities classified as services (NACE GQ). Although some provisions for services statistics exist for all domains, the coverage of service statistics in EU statistical legislation is not complete. The ECB proposed recently a list of actions in order to overcome this unsatisfactory situation[9].The proposals cover:

increase of the number of indicators; highest priority for more information on producer prices, output and employment in the services sector;

improvement in coverage of in particular market service activities for turnover and prices as well as employment for market and nonmarket activities;

improvement in the level of detail in order to allow for the analysis of the contributions from subsectors to the services aggregates;

improvement in timeliness and frequency of indicators (towards more quarterly and monthly data).

3.Statistics for the euro area in an international context

The ECB regularly analyses economic developments outside the euro area[10]; mainly in the US and Japan, but also in emerging economies in Asia and Latin America and in the central and eastern European countries. Although the euro area as a whole depends less than the economies of the individual euro area countries on external developments, the global economy has an important impact on the euro area.

Statistical requirements for economic data are dependent from the purpose of the macroeconomic analysis: for most current ECB purposes the (monetary) policy relevant key figures for noneuro area countries are used, regardless of their comparability with euro area data. This is the obvious choice for the analysis of individual economies. However, the usefulness of national figures must be questioned when comparisons are drawn between economic areas. The main statistical factors which can cause differences in the results have to be known and distinguished in the analysis. For example the US CPI includes an estimate for the expenditure of owner occupiers, but not the HICP for the euro area.

Another criterion is the use of either shortterm or structural analysis. Shortterm analysis for noneuro area countries by the ECB mainly aims at “mirroring” countryspecific, national policy considerations and, therefore, uses data according to national definitions. Structural analysis compares areas outside the euro area with the euro area – the requirement for comparable data is higher for these kinds of analysis.

For some purposes the use of the earliest available economic statistics is essential, in particular when assessing the impact of the external environment on the euro area. For other purposes timeliness of the data is less of a concern and more emphasis is put on other qualities of statistics. This aspect is important, because comparable indicators may become available only with some delay, in particular if they are compiled as supplementary, additional information to existing national data. Wellknown examples are the standardised unemployment rates and national unemployment rates.

Though full comparability among international indicators is probably not possible, the use of very similar or common statistical classifications for products, sectors and branches (e.g. ISIC, COICOP) are important for users. Current joint work of experts from the US, Canada and the EU on the convergence between the General Industrial Classification of Economic Activities within the European Communities (NACE) and the North American Industry Classification System (NAICS) is one example for this.

As almost all shortterm statistics are used in seasonally adjusted format, differences due to the seasonal adjustment of the data may have an effect that might be as relevant for the interpretation of the results as differences in other statistical methods. Though they cancel out in the longer term, shortterm changes crucially depend on the adjustment method used and the assumptions made in their use[11].

Easy access to the international databanks is a prerequisite for daily work. Currently, a range of international databanks providing shortterm economic indicators is available to ECB users. The main databanks used are (besides some commercial sources) the OECD’s Main Economic Indicators database, the IMF’s International Financial Statistics and the databank of the Bank of International Settlements (BIS) (access confined to central banks). As time series in these databanks are not always identical, the identification and selection of indicators need thorough analysis. A lack of good timeliness is sometimes another shortcoming, and a main reason for economists to use commercial sources or seek direct access to data producers outside Europe via Internet.

As regards the metadata information, the Special Data Dissemination Standard (SDDS) of the IMF has proved to be particularly helpful, as much of the information can be accessed in a user friendly manner.

4.Issues for discussion by the OECD shortterm economic statistics group

From the topics outlined in the previous chapters the following points (ordered by priority for the ECB) might be further considered by the OECD Expert Group:

1.International databanks

Which databanks are considered the most appropriate for shortterm economic indicators? What are their source, how are they updated and how should existing databanks be improved to increase their use? These points are important prerequisites for the use of existing databanks and good statistics.

2.Quality/ timeliness/ accuracy

The discussion about improvements of timeliness in euro area statistics and the quality tradeoff between timeliness and reliability often lacked a sound empirical basis. Which statistical measures are recommended for assessing accuracy and revisions? How should information on experienced or expected revisions be made available to the public (e.g. in press releases)? A revision analysis for one or few shortterm indicators may shed further light on this and provide guidance for future decisions in the EU.

3.Services statistics

The development of shortterm demand and output indicators for the services sector is a challenge for all countries. An exchange of experiences and plans would be helpful, also in the light of the forthcoming decisions on developing service statistics in the EU.

4.International comparability

For the comparison (as well as the further development and harmonisation) of international statistics it is important to know their main conceptual differences. Whilst some conceptual information is available in a countrybycountry presentation e.g. from the IMF SDDS templates, a crosscountry comparison for key indicators would be beneficial.

5.Practices for early estimates

The practices for releasing first results vary greatly between EU countries and even more between EU and nonEU countries. Exchange of information on this may improve the knowledge of established practices and useful input to further discussions in the EU.

[1] See ‘The two pillars of the ECB’s monetary policy strategy’, ECB Monthly Bulletin, November 2000.

[2] “Statistical requirements of the European Central Bank in the field of general economic statistics”, ECB, August 2000.

[3] “Assessment of general economic statistics for the euro area”, ECB Monthly Bulletin, April 2001.

[4] Council of Economic and Finance Ministers (ECOFIN Council).

[5] Action Plan on Economic and Monetary Union (EMU) Statistical Requirements (EMU Action Plan), September 2000 (

[6] The time taken to compile and publish any statistical indicator.

[7] “Trade-off between timeliness and accuracy – ECB requirements for general economic statistics” by W. Bier and H. Ahnert, ECB, March 2001; (

[8] “Revision analysis for selected euro area and EU country short-term indicators”, ECB, March 2001; published in: “European Commission: Short-term statistics - Improving timeliness and cooperation”, Luxembourg, 2001.

[9] “Short-term statistics on services for the euro area”; March 2002; document produced for the Committee of Monetary, Financial and Balance of Payments Statistics (CMFB).

[10] See, for example, ECB Annual Report 2001, Chapter I, Section 6 “The global macroeconomic environment, exchange rates and the balance of payments”.

[11] In the field of quarterly national accounts proposals have been made by the IMF and recently by a Eurostat/ECB Task Force.