论文信息 - Poisson Hidden Markov models for time series of overdispersed insurance counts

Poisson Hidden Markov models for time series of overdispersed insurance counts

We suggest the use of Poisson hidden Markov models (PHMMs) in non life insurance. PHMMs are an extension of the well-known mixture models and we use them to model the dynamics of overdispersed data, in particular of the claim number. PHMMs allow us to explicitly consider unobserved factors influencing the dynamics of the claim number. This has an immediate impact on the value of the pure risk premium: the expected claim number is given by a weighted average of the intensity parameters of a PHMM. We show how the maximum likelihood estimators of the parameters of PHMMs may be suitably obtained using the EM algorithm and apply PHMMs t,o model the daily frequencies of injuries in the work place in Lombardia (Italy).

Roberta Paroli | Luigi Spezia | Giovanna Luisa Maria Redaelli

[1] Lain L. MacDonald,et al. Hidden Markov and Other Models for Discrete- valued Time Series , 1997 .

[2] C. B. Bell. Statistical Inference for Some Special Families of Stochastic Processes , 1975 .

[3] P S Albert,et al. A two-state Markov mixture model for a time series of epileptic seizure counts. , 1991, Biometrics.

[4] C. D. Daykin,et al. Practical Risk Theory for Actuaries. , 1994 .

[5] M. Puterman,et al. Maximum-penalized-likelihood estimation for independent and Markov-dependent mixture models. , 1992, Biometrics.

[6] Jdh. Statistical Inference for Stochastic Processes. By I. V. Basawa and B. L. S. Prakasa Rao. Pp. 435. Academic Press, London. £26.80. , 1981 .

[7] P. Bickel,et al. Asymptotic normality of the maximum-likelihood estimator for general hidden Markov models , 1998 .

[8] D. Oakes. Direct calculation of the information matrix via the EM , 1999 .

[9] P. Guttorp. Stochastic modeling of scientific data , 1995 .

[10] S. P. Pederson,et al. Hidden Markov and Other Models for Discrete-Valued Time Series , 1998 .

[11] J. Corcoran. Modelling Extremal Events for Insurance and Finance , 2002 .

[12] Kenneth Lange,et al. Numerical analysis for statisticians , 1999 .

[13] G. McLachlan,et al. The EM algorithm and extensions , 1996 .

[14] Paolo Giudici,et al. Likelihood‐Ratio Tests for Hidden Markov Models , 2000, Biometrics.