Title: Expectations correction and DSGE model selection
Authors: Giovanni Angelini - University of Bologna (Italy) [presenting]
Abstract: The focus is on a bootstrap refinement of the Expectations Correction idea already proposed in the literature. Expectations Correction is a feasible remedy to the dynamic misspecification that characterizes the small-scale New-Keynesian monetary policy models. These models are generally not able to take into account all the dynamic correlation structure present in quarterly data hence the empirical performance of these models could be problematic. To solve this problem a pseudo-structural model is built from the baseline of a DSGE model by adding a number of lags present in the statistical model for the data. We refine this idea using bootstrap techniques for both the lag structure selection and both in the empirical evaluation. The selection of the true number of lags is a crucial point in the definition of the pseudo-structural model and it is particularly important in the forecasting performance. In this sense bootstrap techniques are particularly useful to identify the true lag structure. Moreover, the addition of new lags in the structural model produces a higher number of non-linear cross equation restrictions and this is the cause of a high rejection rate of the theoretical model and a bootstrap approach can reduce this overejection. To evaluate the performances of our method we provide a Monte Carlo simulation study and an empirical illustration based on U.S. quarterly data.