Stochastic mortality model in a state-space framework

Abstract

The incorporation of the time-varying parameter in the mortality model has become one the main contributions in the actuarial field since it al- lows for the stochastic nature of the mortality rates. However, it has also become a growing concern among the researchers since the residu- als of the proposed model are evaluated independently. In this study, we extended the existing leading independent stochastic mortality model which is the O0Hare mortality model into the state-space representation of the O0Hare mortality model. The parameters of the extended model are estimated using the Expectation-Maximization algorithm of the max- imum likelihood estimation method. Using the Malaysian mortality data, we have found that our proposed model significantly improves the accu- racy of the in-sample fitting and the seven-year out-sample forecast as compared to the existing model considered.