Matthieu Cornec, Fanny Mikol
ENSAE, Amphi 1
Jeudi 9 Février 2012
"Nowcasting GDP directional change with an application to French business survey data".
Despite a rich litterature on GDP growth level forecasting, few studies
have focused on forecasting GDP directional change. This is all the more
suprising that economic outlook analysis is mainly explained in terms of
acceleration or deceleration. We conduct a comparative study between
di erent class of models ranging from econometrics to machine learning.
Empirical investigations on French economy suggest that classi cation
models slightly outperform their level-based counterpart, with the respect
to the sign forecast exercise. We also provide analytical properties for
future model-based predictions. Eventually, we construct a directional
risk index which describes in probability terms the risk of an upcoming
deceleration. Applied to the balances of di erent business surveys, it
appears as a useful tool for economic forecasters