Eurostat: why its data is dodgy too

Date: 02/10/03

The European Commission's statistical agency, Eurostat, deserves closer attention - and not only because of the ongoing allegations of fraud at the top, argue Daniel Franklin and Laza Kekic

WHEN it comes to Eurostat, the sharpest focus, because the biggest sums of money are involved, should perhaps be on an area that has so far attracted surprisingly little scrutiny: the statistics on which key EU decisions may be based.

Dull as they are, these statistics matter. They can shape perceptions and policy on important issues. Billions of euro of EU aid may depend on them, since Brussels distributes its so-called structural funds to regions that qualify under certain statistical criteria of relative poverty compared with average levels in the Union.

The importance of these numbers will only grow as the EU expands to include a batch of new members in May next year. Eight of these newcomers are former communist countries in central and eastern Europe (the other two are the Mediterranean islands of Cyprus and Malta). Two other eastern European countries, Bulgaria and Romania, are hoping to join the EU in 2007.

Eurostat's numbers are important in influencing views on enlargement, as well as in determining aid flows.

The countries set to join the EU are much poorer than the current 15 member states. How much poorer? This is where Eurostat's calculations come into play.

One way of measuring relative wealth is by looking at gross domestic product (GDP) per head at market exchange rates. On this basis, living standards in the eight central European countries average just 24% of the EU-15.

The range is from as low as 15% of the EU average in Latvia to 46% in Slovenia. It is lower yet, at 9%, in future entrants, Bulgaria and Romania.

But, for many well-known reasons, incomes converted into a common currency at market exchange rates cannot properly measure real differences in living standards. So economists estimate 'purchasing-power parity' (PPP) exchange rates, which eliminate the differences in relative price levels between countries. But estimating PPPs may be more of an art than a science. The results are highly sensitive to weighting methods and the treatment of 'non-market' services (such as health and education) that are hard to compare.

Frustratingly, the results for different benchmark years are rarely compatible with national-accounts growth rates. The margins of error in these calculations are therefore large.

In the case of the eight central European countries set to join the EU next year, extrapolated PPPs from a 1993 benchmark study (by Eurostat and the Organization for Economic Cooperation and Development, the OECD) show their average GDP per head in 2002 at 35% of the EU-15's level: still poor, but less so than at market rates.

For the next (1996) benchmark study and all subsequent ones (1999 and 2000) Eurostat brought in various revisions to the methodology - most notably the practice of making extensive adjustments for quality, especially in comparisons of non-market services, was discontinued.

This greatly improves the picture for most central European countries: to an average of 46% of the EU-15 level in 2002.

It is natural for PPP rates to nudge poorer countries' income per head upwards, since non-tradeable services (such as haircuts or gardening) are relatively cheap there. And there may be reasons why the true numbers deserve to be higher than those based on the 1993 benchmark study: for example, the central European countries may have particularly large black economies, or their actual growth rates in recent years may have been understated because GDP numbers have failed to account properly for the improving quality of goods and services.

But the more you look into Eurostat's numbers, the harder it is to escape the conclusion that they are upwardly biased.

Extrapolated back to 1990, the numbers show the central Europeans to be much richer than in previous studies, which might suggest that communism was not so bad after all. Against all common sense, the Prague region is far above the EU average and not far behind Vienna.

Wages in US dollars correspond to those found in countries with far lower incomes per head - or indicate that labour is priced at implausibly low levels relative to productivity (measured at the recent PPPs). Compared with similar economies, the share of trade in GDP appears too low and the implied degree of undervaluation of currencies implausibly high.

A comparison of a range of physical indicators that are highly correlated with GDP also suggests that the latest Eurostat numbers overestimate GDP per head in the region.

Eurostat itself warns that “these indices are not intended to rank countries strictly. In fact, they only provide an indication of the comparative order of magnitude of the per capita GDP volume of one country in relation to another”.

Yet Eurostat also notes that one of the particularly important uses of PPPs is for the European Commission “to establish both the regions that could benefit from the EU structural funds as well as the amount of funds to be allocated to each region. One criterion for allocating these funds is based on PPP-converted GDP per capita.” Translation: the numbers are imprecise, but precision matters.

For now, numbers that show the central Europeans to be not quite so poor suit both the accession countries and the current EU members: admitting downright paupers to a rich-people's club might be more embarrassing.

Yet the case for EU enlargement should be strong enough without needing to rely on dodgy data.

The numbers may receive closer scrutiny when they show countries approaching levels at which they are in danger of losing EU aid. It is time to take a closer look not just at Eurostat's top officials, but at the numbers the agency produces.

• Daniel Franklin is editorial director and Laza Kekic is director of country forecasting at the Economist Intelligence Unit. The statistics in this article are taken from a recently released report entitled Europe enlarged: Understanding the impact.

At a time when Eurostat, the European Commission's statistical agency is under scrutiny, this article argues that the way in which statistics are produced on GDP in applicant countries could be regarded as suspect.