Life Insurance Solvency Assessment Using Stochastic Simulation Techniques

Abstract

While most UK life insurance companies employ asset/liability models to project their emerging cash flows, few employ stochastic simulation methods. The preferred approach is to project the business deterministically, using sensitivity tests to assess any insolvency risk. This method has been notably advocated by Brender(1988) in his work on a solvency standard for the Canadian industry. But with many insurers being forced to slim their reserves in order to remain competitive, is this deterministic, generally rather ad hoc approach adequate? In contrast with Brender’s conclusions, the Faculty of Actuaries Solvency Working Party (1986) recommended strongly that insurers should, once techniques were suitably developed, be required to demonstrate to the supervisory authorities that their estimated probability of ruin is acceptably low, using stochastic simulation. In this paper a model office is used to demonstrate the difference in the quality of information available from a set of stochastic simulations compared with a traditional deterministic approach.