J. Japan Statist. Soc., Vol. 32 (No. 2), pp. 141-154, 2002
Abstract. This paper considers the count model with endogenous switching proposed by Terza (1998) from a Bayesian point of view. We consider Markov chain Monte Carlo methods to estimate the parameters of the model. Furthermore, an extension is made to handle the case of non-normality and model determination is discussed. Our approach is illustrated with both simulated and real data sets.
Key words and phrases: Bivariate t distribution; Count data; Endogenous switching; Markov chain Monte Carlo; Poisson model; Unobserved heterogeneity.