Mean reversion in stochastic mortality: why and how?

Life insurance companies use stochastic models to forecast mortality. According to the literature, non-mean reversion models are more suitable for mortality modelling than mean reversion models with a fixed long-term target. In this paper, we adopt stochastic affine processes for the force of mortality and study the impact of adding a time-dependent long-term mean reversion level to two non-mean-reverting processes. We calibrate the models to different generations of the Belgian population and assess these models’ abilities to predict mortality using different statistical methodologies. The backtest shows that the survival curves provided by the mean-reverting processes are closer to reality. Thus, we conclude that incorporating a time-dependent target into these considered models improves their performance significantly.