Is there a profile for  » good » chief forecasters?

Mardi | 2015-02-17
Sully 5, 16h30-18h

Jean-Charles BRICONGNE – Irène GRIALOU

 » Does the profile of chief economists in charge of forecasting in public and private institutions matter for the quality of their forecasting? Different arguments may be invoked in favor or against this argument: chief economists do not always influence directly forecasting, and it may be rather due to the quality of the staff, but on the other hand, the staff may be chosen, reorganized and influenced by the chief economist, whose final opinion should be decisive.Forecasting is indeed a technical matter which requires skills, but opportunism may also play a role, by considering other institutions’ forecasts, and this is not always easy to take into account in a profile.We thus use an agnostic approach, starting from the principle that if a characteristic leads to better or worse forecasting quality, then characteristics matter and a profile for  » good » forecasters may emerge.Among the main stylized facts, we find that the horizon of the forecasting and the variable forecasted matter: performance is not the same whatever the horizon of the indicator. We also find that, for the chief economists, former experience matters, among others former experience in the public sector (especially for French civil servants as regards budget balance forecasts). Lastly, when using the concept developed in Tetlock (2005), we find that  » hedgehogs » rather over perform compared to  » foxes » and that this characteristic may occult other explanatory variables. »