Quelques constats sur les prévisions conjoncturelles de la croissance française

We study the quality of the French growth forecasts from the first decade of the century. An optimism bias of the quarterly Consensus growth forecasts can be asserted right from a graphic analysis and its existence is confirmed by the calculation of the mean errors which are systematically negative. The Theil index reveals that Insee forecasts are superior to the Consensus ones. In addition, the Consensus forecasts are superior to a naïve forecast. Proper tests seem to confirm that an optimism bias exists ; this bias could arise from the combination of several forecasts. The study of the Consensus Economics fixed event forecasts regarding the next coming year requires a preliminary analysis due to missing data. We thus only retain the forecasts of nine institutions that we compare to those of the Government, the IMF, the OECD, and the European Commission. It appears that the forecasts are fairly close to the Consensus forecast and that the optimism bias is still observable. Finally, the disagreement between the forecasters increases towards a recession and, then, decreases.