Publication details
States Estimation in Baseline New Keynesian DSGE Model: Kalman or Bootstrap Filter?
| Authors | |
|---|---|
| Year of publication | 2008 |
| Type | Article in Proceedings |
| Conference | Mathematical Methods in Economics 2008 |
| MU Faculty or unit | |
| Citation | |
| Field | Economy |
| Keywords | DSGE model; Kalman filter; Bootstrap filter; |
| Description | The aim of this article is to compare the ability of estimation and filtering methods. Under assumptions of normality and linear structure, Kalman filter seems to be more accurate in comparison with Bootstrap filter that cooperates with empirical distributions. But we can show that 'chances' are approximately similar if we set for Bootstrap filter the same variances of structural shocks as we obtain from Kalman smoothed structural shocks. A construction of Kalman filter enables to flexibly change estimated covariance matrix of observation vector estimation error during filtering and smoothing procedure. But this 'advantage' is not implemented in Bootstrap filter since the bootstrap filter measurement error variances and structural shocks variances must be constant during filtering and smoothing procedure. The results will be shown on the DSGE baseline new Keynesian model of Czech economy. The model will be filtered and smoothed by above mentioned filters and a comparison study will be carried out. Practical experiences will be used for time - varying parameters estimation of DSGE models by Bootstrap filter since time - varying models are nowadays modern and very popular tool for analyzing changes in economic environment. |
| Related projects: |