Bayesian Analysis of Count Variables

Abstract

The existing econometrics literature on count data models has largely ignored the Bayesian paradigm of inference. Likewise, in Zellner’s (1971) influential book on Bayesian inference in econometrics, the Poisson regression model is not mentioned. The probable reasons for this neglect are computational complexities that in the past made the Bayesian analysis of count data models appear unattractive. However, increased computer power now allows for fast evaluation of posterior distributions by simulation methods. The basic approaches to Bayesian inference by simulation are discussed in this chapter.