Metoda największej wiarygodności a regresja przełącznikowa

Abstract

The article deals with the problem of switching regression. The model of switching regression together with the maximum likelihood method and an equivalent Kiefer’s method meant as such model’ s parameters estimation method are presented here. One of the examples of the switching regression models are the econometric disequilibrium models describing demand and supply in the non-balanced markets. The article contains results of parameters estimation for two disequilibrium markets: fish and their products market and the labour market in the petroleum and power industry by means of Kiefer’s method. The Monte Carlo analys is of estimators’ properties obtained by means of Kiefer’s method has been presented ?s well. In the view of the performed experiments, estimators for two assumed specifications of disequilibrium models were characterized with statistically significant mean biases and deviations from the normal distributions when the sample size was small, as it was assumed in the experiments (23 and 24). It results from the above that the Kiefer’s method does not give good enough estimates of the switching regression models’ parameters.