An Analysis of Revisions to Quarterly National Accounts Since the Introduction of Flash Estimates in 2005

Quarterly national account estimates are subject to revisions over time, to account for newly available data and new methods of computation, as well as a revision of their underlying data source. In economically turbulent periods, such as the financial and economic crisis of 2008-09, revisions can be expected to be more pronounced. This is true in particular for the first publication of quarterly national accounts (the so-called Flash Estimates). This analysis finds that revisions of the Flash Estimates compared to their final value were neither biased in any direction nor large in their scope. Moreover, the direction of revisions cannot be forecast. Accordingly, the first publication shows a high degree of reliability in terms of expected revisions, in spite of the reduction in its publication date (from 70 days to 45 days after the end of a given quarter). The crisis years of 2008 and 2009 had the expected impact in the revisions: during the crisis, the downward course of the GDP was significantly underestimated. Nevertheless the upper turning point was correctly dated and did not have to be corrected even after several revision runs. In a similar vein, the duration of the contraction and the start of the recovery were found to be quite robust vis-à-vis new calculations.