Forecasting mortality rates of elderly in Indonesia using the first generalized Cairns-Blake-Dowd model

Abstract

The insurance companies in Indonesia is always aware of the importance of forecasting future mortality rates. If information about mortality rates for several future periods can be obtained in the present time, then the financial planning and policy in determining the amount of premium are expected to be better and more directed. This paper proposed the First Generalized Cairns-Blake-Dowd (CBD) model to forecast the mortality rates of the Indonesian population. The CBD model contains time series parameters. The first stage is to use the Least Square and Newton-Raphson methods to estimate these parameters. The accuracy of the estimation results is verified by the Mean Squared Error (MSE) value. The second stage is to use Holt's linear trend method to forecast the values of estimated parameters aims to calculate the future mortality rates. The level of forecasting accuracy is verified by Mean Absolute Percentage Error (MAPE) value. The MSE and MAPE for each age group are less than 10-4 and 5%, respectively. The forecasted mortality rates for the next periods are presented in the tabular and graphical form. The trend of the projected mortality rate shows a downward trend for each age group.