This paper deals with effect of different real-time data vintages (recent and first outturn) on accuracy of real GDP growth forecasts produced by main Czech and Slovak public authorities (ministries of finance, central banks). Firstly, variation in the real-time data itself was analysed, along with of multidimensional forecasting error evaluation (MAE, RMSE, MASE measures). Then, battery of statistical tests was applied in order to determine, whether the switch from first to recent real-time data affects forecasts´ accuracy in a significant manner (Wilcoxon Signed Rank test, Sign test) and whether it affects relative accuracy between individual institutions (Kruskal-Wallis test, Mann-Whitney U test). Our results show that while the change in underlying data affects forecasting accuracy in our sample (using recent data lead to higher errors), the changes were neither found statistically significant in strong majority of surveyed cases, nor affected the relative accuracy of involved institutions.
JavaScript is turned off in your web browser. Turn it on to take full advantage of this site, then refresh the page.