Multi-population mortality modeling with Lévy processes

This paper constructs a theoretical framework for multi-population mortality modeling via generalized linear models and Lévy stochastic perturbations driven by a common Brownian motion and idiosyncratic factors to capture the mortality shocks. By having Lévy stochastic perturbations, our model admits various jump types, which is increasingly important for capturing mortality shocks such as pandemics, particularly when they affect various populations differently. At the same time, the proposed model allows a novel dependence structure of multiple populations, which is essential when it comes to the development of multi-population or joint-life products in the context of mortality shocks. In our empirical investigations, the mortality experiences of male and female lives in the UK and Japan are used. Compared with pure Poisson-generalized linear models, the proposed multi-population model shows superiority in predicting future mortality rates.