A1092
Title: Dynamic Poisson state space prediction model for automobile insurance
Authors: Jiakun Jiang - Beijing Normal University (China) [presenting]
Abstract: Prediction modelling of claim frequency is important for pricing and risk management in nonlife insurance. It needs to be updated frequently with the insured population and technology changes. Existing methods are either done in an ad hoc fashion, such as parametric model calibration or less so for the purpose of prediction. A Dynamic Poisson state space (DPSS) model is developed, which can continuously update the parameters whenever new information becomes available. DPSS model allows for both time-varying and time-invariant coefficients. To account for smoothness trends of time-varying coefficients over time, smoothing splines are used to model time-varying coefficients. The smoothing parameters are objectively chosen by maximum likelihood. The model is updated using batch data accumulated at prespecified time intervals, which allows for a better approximation of the underlying Poisson density function. The simulation shows that the new model has significantly higher prediction accuracy compared to existing methods. This methodology has been applied to real-world automobile insurance claim data sets in China over six years. It demonstrates its superiority by comparing it with the results of competing models from the literature.