Inference about mortality improvements in life annuity portfolios

Recent mortality trends lead to the use of projected mortality tables when pricing and reserving for life annuities (as well as for other living benefits, such as Long Term Care benefits, whole life sickness benefits, etc.). However mortality patterns continuously evolve along time, so that any projection might reveal weak when used for pricing new annuities and reserving for in-force business. Hence, adjustments must be made in pricing and reserving bases. Monitoring mortality provides data, while an appropriate inferential model should constitute the structure underpinning the adjustment procedure. In this paper, inference about portfolio mortality trends is first focussed. Then, a Bayesian inferential model is proposed, aiming at mortality adjustments based on prior information and statistical evidence. Numerical examples illustrate the inferential mechanism. Finally, some actuarial applications are proposed.