DSGE Model Estimation on the Basis of Second-Order Approximation
This article compares the properties of different non-linear Kalman filters: the well-known Unscented Kalman filter (UKF), the Central Difference Kalman Filter (CDKF) and an unknown Quadratic Kalman filter (QKF). A small financial DSGE model is repeatedly estimated by several quasilikelihood methods with different filters for data generated by the model. Errors in parameters estimation are a measure of the filters’ quality. The result shows that the QKF has a reasonable advantage in terms of quality over the CDKF and the UKF, albeit with some loss in speed.