Modelling and Forecasting the Mortality Risks to Investigate its Impact on Insurance Companies in Malaysia: Poisson Lee-Carter Model and Cairns-Blake-Dowd Model
1Ong Ying Ming, Raja Rajeswari, Hazlina Darman
The objective of this paper is to model and forecast the future mortality rates based on the data of insurance companies in Malaysia. This paper compares and applies three stochastic mortality models which are Lee-Carter, Poisson Lee-Carter and Cairns-Blake-Dowd models to forecast mortality rates for population of Malaysia. All genders and all single age data are fitted to all the models from year 2000 to year 2015. Based on the results, it shows that the fitted and forecasted values are having the different pattern of time index for Lee-Carter of family model and Cairns-Blake-Dowd model. To estimate the appropriate model among these models used in this study, the maximum log likelihood, AIC and BIC are used. According to the results of goodness of fit, Poisson Lee-Carter model has the best and the lowest values compared to other two models. The results from MAE and MAPE indicate that the forecast values are closer to the actual value. Therefore, Poisson Lee-Carter model is a good fit model to forecast Malaysia mortality rate.
Forecasting, Mortality rate, Lee-Carter model, Poisson Lee-Carter model, Cairns-Blake-Dowd model