Fiscal data revisions in Europe

Public deficit figures are subject to revisions, as most macroeconomic aggregates are. Nevertheless, in the case of Europe, the latter could be particularly worrisome given the role of fiscal data in the functioning of EU’s multilateral surveillance rules. Adherence to such rules is judged upon initial releases of data, in the framework of the so-called Excessive Deficit Procedure (EDP) Notifications. In addition, the lack of reliability of fiscal data may hinder the credibility of fiscal consolidation plans. In this paper we document the empirical properties of revisions to annual government deficit figures in Europe by exploiting the information contained in a pool of real-time vintages of data pertaining to fifteen EU countries over the period 1995-2008. We build up such real-time dataset from official publications. Our main findings are as follows: (i) preliminary deficit data releases are biased and non-efficient predictors of subsequent releases, with later vintages of data tending to show larger deficits on average; (ii) such systematic bias in deficit revisions is a general feature of the sample, and cannot solely be attributed to the behaviour of a small number of countries, even though the Greek case is clearly an outlier; (iii) Methodological improvements and clarifications stemming from Eurostat’s decisions that may lead to data revisions explain a significant share of the bias, providing some evidence of window dressing on the side of individual countries; (iv) expected real GDP growth, political cycles and the strength of fiscal rules also contribute to explain revision patterns; (v) nevertheless, if the systematic bias is excluded, revisions can be considered rational after two years