Dynamic filtering estimation of Markov regime-switching cointegrating regression model
马尔可夫协整回归模型的动态滤波估计

A dynamic filtering method is proposed to estimate the parameters of Markov regime-switching cointegrating regression model. In order to eliminating both the serial and contemporaneous correlation between the regressors and errors, an auxiliary dynamic regression model is developed by using a leads-and-lags approach. The maximum likelihood estimation (MLE) is performed on the auxiliary model by employing the Hamilton filter. Simulation experiments show that the method reduces the bias of the parameter estimator. A Markov cointegrating model is estimated between the export and import trade in China from January 1990 to October 2011.