Title: Nonlinear impulse response function for dichotomous models
Authors: Quentin Lajaunie - Université Paris Dauphine (France) [presenting]
Abstract: A generalized impulse response function (GI) for dichotomous models is proposed. Building on previous work, we develop the exact form of the response functions for each specification of their binary model. Using a block-bootstrap method, we compute robust confidence intervals for these response functions. We illustrate the usefulness of this analytical result for static and dynamic dichotomous models of U.S. recessions. According to the different specifications, we empirically find that the persistence of an impact of an exogenous shock to the U.S. economy is between one to five quarters.