In this paper we provide a comprehensive evaluation of the euro area GDP growth and unemployment rate forecasts collected in the quarterly ECB Survey of Professional Forecasters (SPF) over the period 1999Q1–2008Q4. Our results suggest that while SPF forecasts generally appear to be slightly superior to naive and purely backwardlooking benchmarks, forecast errors nonetheless exhibit a high degree of persistence. In addition, our analysis of the heterogeneity across individual SPF replies suggests that the broad pattern of the individual forecasts is essentially the same as that of the aggregate SPF results. This may refl ect a high degree of commonality in the information available (and not available) to panel members, thus leading them to get it wrong (or right) not only in the aggregate, but also individually. In particular, although a small number of forecasters perform substantially above average for some variables and horizons, none does so systematically for all variables and all horizons. Lastly, we have presented and assessed the information about forecast uncertainty provided by the SPF. In line with other studies based on the US SPF, disagreement among panel members does not appear to be a good proxy for overall macroeconomic uncertainty, i.e ., a high degree of consensus is not necessarily an indication of a low level of forecast uncertainty. Our analysis also suggests that, at the individual level, panel members may not fully internalise the overall level of macroeconomic uncertainty. For example, compared with the level of uncertainty indicated by the historical volatility of actual GDP growth and the unemployment rate, the perceptions of individual panel members about uncertainty appear quite low. This possible underestimation of overall uncertainty is much less severe when densities are aggregated across forecasters.
Read full abstract