Multivariate Zero-Inflated INAR(1) Model with an Application in Automobile Insurance (Q6953014)

In this paper, the authors introduce an elegant solution for studying multidimensional non-life claim count data. Specifically, the data exhibit time-dependence, cross-dependence and zero inflation. To this end, the authors propose a multivariate INAR(1) model, where the innovations are described by either a multivariate zero-inflated Poisson distribution or a multivariate zero-inflated hurdle Poisson distribution. In addition, the proposed model also features the impact of individual and coverage-specific covariates on the mean parameters. Consequently, it is possible to evaluate customized insurance premiums in the light of changing risk profiles. The authors adopt the expectation-maximization (EM) algorithm to estimate the model parameters. Using the European motor third-party liability claim count data, the authors illustrate the effectiveness of the proposed model. In Section 2 of the paper, the authors discuss the multivariate zero-inflated Poisson distribution and the multivariate zero-inflated hurdle Poisson distribution. Section 3 introduces the multivariate INAR(1) model. Specifically, the multivariate INAR(1) model with three types of innovations are described. The estimation method based on the EM algorithm is also considered. Section 4 presents a simulation study. Section 5 provides the estimation results, a study of the predictive performance, and an application to ratemaking.