Analysis of revisions to general economic statistics

The preparations for the introduction of the euro in 1999 involved the need for a new set of statistics for the euro area. Since then, significant progress has been made with regard to the coverage, timeliness and accuracy of these statistics. The reliability of the first releases – i.e. their stability in the process of later revisions – is an important quality-related feature. New data releases for the euro area have generally shown a very small or no bias, i.e. data revisions have been very modest and comparable with those of, for example, the United States or Japan. Despite the relatively small size of revisions, however, their combination with the low growth of the euro area economy may have drawn attention to such revisions of economic data for the euro area. This paper quantifies the revisions to selected key indicators in the period from the start of Monetary Union in 1999 to July 2007 and compares them with the corresponding medium term averages (1999-2006). The analysis covers the euro area, its six largest member countries, the United Kingdom, the United States and Japan. For this purpose, available time series for the various periods involved are used, series that record all revisions to published statistical data releases. The analysis is carried out separately for GDP growth and its expenditure components, for employment, unemployment rates, compensation per employee, labour cost indicators, industrial production, retail trade turnover and consumer prices. Overall, the evidence presented in this paper suggests that euro area data releases have generally shown a very small or no bias and have been more stable than those for individual euro area countries. Furthermore, recent euro area data show levels of revisions similar to those of the past, or levels of revisions that stabilised after the implementation of harmonised statistical concepts had largely been completed.