Head of stats office resigns over GDP miscalculation
The head of the Czech Statistics Office (CSU) Marie Bohata resigned last Friday. The reason was a scandal that had shaken the credibility of the institution - the publishing of substantially wrong GDP figures for the third quarter of 2002 which influenced strategic decisions by the government and Czech National Bank as well as investors and ordinary people.
In order to establish the cause and responsibility for the accident and to prevent its recurrence in the future, Mrs. Bohata asked Eurostat, the European statistics office, to carry out an audit at the Czech Statistics Office.
Eurostat's director for foreign trade statistics, Daniel Byk, explained the core of the problem:
"On July 1, an amendment to the Customs Act and a new common decree of the Ministry of Finace and the Czech Statistics Office entered into force. For CSU, it involved a significant change, in particular in the customs procedure in order to comply with EU legislation. These new rules made it necessary to adapt an existing decision table on how to select and process each customs procedure for statistical purposes. But the customs did not implement the table sent by the statistics office. In Eurostat's opinion, there was a failure of communication between the statistics office and the Customs Authority with respect to this conversion table."
One of the biggest questions now is why the development did not strike economic analysts as suspicious. Their comments suggest that they had been surprised but none of them dared to protest. Observers say the reason is simple: until January 16, data from the Czech Statistics Office simply were there to be interpreted and commented on, not for their validity to be questioned.
Daniel Byk of Eurostat says one of the factors that contributed to the problem was the August floods, and not only in the economic respect.
The level of imports and exports recorded over the period was not exception at the first glance. Only the level of August export was particularly low - 80.5 million CZK, the lowest figure since January 2000. CSU pointed out the low the low number of customs declarations for that month, explaining it by the effect of the floods on export, combined with the usual seasonal effect.
The floods contributed also to the disorganisation of CSU: computer resources were disrupted, staff were dispersed between at least eight sites. They were only able to regroup part of the staff in a single new building at the end of October. The identification of a possible problem required comparison with other sources of data, such as household consumption, gross capital formation and output. These sources were not available when figures July to September were not available."
The statistics office admitted it was disturbed by the discrepancy in data but went ahead with publishing the result, while secretly looking for possible errors in cooperation with the customs authority.
While mistakes in statistics occur everywhere in the world and macroeconomic data revisions are not uncommon, it is the Statistics Office's secretive approach that many criticise. Now, what Eurostat recommends Czech statisticians to do to avoid such situations in the future?
"The existing rules for processing values of exports for statistical purposes should be reassessed in order to avoid misinterpretation and facilitate data processing. The current methodology for compiling trade statistics defined in national legislation should be carefully examined in the light of the EU regulation and in the context of enlargement. Guidelines on the implementation of the statistical rules in the processing system should be specified by the statistics office. Cooperation between CSU and the Customs Authority should be reinforced and better structured with contact persons clearly identified clearly on both sides and at each level of responsibility. In case of suspicious data, communication to users on potential risk of data should be more transparent and explicit. There is a need to develop a culture of analysis and economic validation of statistical data. In the view of the new responsibilities and challenges for CSU in the future, appropriate resources should be secured."
What happened in the Czech Republic is not unique. For example, in the United States, an early GDP growth estimate for the fourth quarter of 2001 was 0.2 percent. A month later, the figure was corrected to 1.4 percent, then to 1.7 percent, rising eventually to 2.7 percent. In Mr. Byk's opinion, statisticians are in an awkward position everywhere in the world, having to respond to various - and often contrary - demands from data recipients:
"We statisticians are under schizophrenic pressure - timeliness on the one hand, quality on the other one. The only way to solve it is to use quick estimation methods but together with strong economic analysis of the data. This is something which is to be developed Europe, in the European Union. The top statisticians in the EU say we have to improve in that line, and this is valid, too, for the Czech Republic. It is not an exception, it is a common situation."