A dynamic hurdle model for zeroinflated panel count data

Abstract

This article proposes an approximate conditional dynamic finite mixture hurdle model for panel count data with excess of zeros and endogenous initial conditions. We provide parameter estimates by using the Expectation-Maximization (EM) algorithm in a Nonparametric Maximum Likelihood (NPML) framework. An application to a unique data set on traffic violation counts of a subpopulation of Italian drivers is given.

Keywords:

• hurdle model
• dynamic models
• finite mixture
• approximate conditional model
• illegal behaviour

JEL Classification:

• C14
• C33
• K42

Acknowledgement

We thank Antonio Nicita for sharing with us the traffic violations data used in this article.

Notes

¹ We have also fitted a negative binomial hurdle model, whose results are not reported here for sake of brevity, and because they are similar to those obtained with the standard hurdle model.