A0464
Title: Nowcasting French GDP with macroeconomic random forests
Authors: Julien Andre - University Paris Dauphine/Banque de France (France) [presenting]
Abstract: The performance of three types of random forests is compared and evaluated in a recent period. Random forests with and without predictors are reviewed in a setting with mixed frequencies and missing values. The primary focus is on their ability to nowcast French GDP growth during the Covid crisis. Monte Carlo simulations are used to validate estimations, and variable contributions are assessed using variable importance tools.
