Creating portfolio-specific mortality tables: a case study

Abstract

Effective risk management of a portfolio demands accurate and succinct models which explain the main risk factors. Since portfolios have detailed individual records, an ideal approach is to use survival models. We look at a case study of how the administrator of a large multi-employer pension scheme created its own mortality tables. In addition to looking at statistical tests of fit, we consider a process for checking the suitability of a model for financial purposes. We also illustrate how a given scheme can test whether its experience is significantly different from other schemes, even after allowing for various known risk factors.