Title: Scenario response distributions
Authors: Caroline Jardet - Banque de France DGEI-DCPM (France)
Alain Monfort - ENSAE Paris (France)
Fulvio Pegoraro - European Central Bank and CREST (France) [presenting]
Abstract: A statistical methodology generalizing the standard Impulse Response Function (IRF) theory in two directions is proposed. First, we consider the case where the new information may be not only about the present value of the multivariate process of interest, but also about future values. Second, we show how to compute not only the average responses of the variables but the whole distribution of these responses and, therefore, several features like variances, quantiles or prediction intervals. Our methodology, called Scenario Response Distribution (SRD) methodology, can be used not only in the context of standard Gaussian VAR models, but also in the context of nonlinear models, in particular censored or regime switching models. An empirical exercise illustrating the SRD methodology will be provided.