Estimating Nonlinear Approximation of DSGE Models: the Case of Czech Economy

The aim of this paper is to estimate nonlinear approximation of a New Keynesian DSGE model of Czech economy and to compare the results with the results obtained from standard estimation procedure which uses (log)linear approximation of the model. Several authors, e.g. An and Schorfheide (2006), Fernandez-Villaverde and Rubio-Ramirez (2005), have used baseline models and/or simulated data to show that the nonlinear approach provides more accurate estimates and better data fit. In our application we use slightly modified Liu's (2006) open economy model. We solve the model for second-order approximation and estimate it by Bayesian techniques, particularly evaluate its likelihood function with particle filter. For computations we employed standard MATLAB functions and also developed our own procedures.