The role of data revisions and disagreement in professional forecasts

This paper aims at evaluating individual expectation accuracy of professional forecasters for 57 U.S., European, and German macroeconomic indicators over the period 1999-2010. The empirical analysis shows that initial announcements are partly considerably revised, and that some revisions occur systematically. Taking into account whether announcements are revised systematically and whether economists (assumingly) aim at forecasting the initial release or the latest revision, significant differences can be observed with regard to forecasters’ expectation errors. In general, forecasters that are (assumingly) aiming to predict the latest revisions of German indicators are able to form better forecasts if these indicators are revised systematically. Though to a lower extent, this relationship is also observable regarding U.S. indicators. Forecasters’ disagreement about fundamentals is higher during recessions and when stock markets are volatile.