For cases with sufficient mortality experience, a robust estimate of base mortality can be produced, again with an appropriate measure of uncertainty.
We can model mortality experience for a portfolio relative to a standard mortality table or the postcode model – we default to the latter because this automatically captures heterogeneity, without the need for credible evidence of heterogeneity in the experience data itself (which requires very large datasets). In a mortality context, persistent heterogeneity is known as ‘frailty’ and tends to increase present value (because, as the liabilities are run off, the average mortality becomes lighter).
This matters because not capturing frailty makes you more likely to under-estimate liabilities. At Aon, we regard the above approach as a key safe-guard:
- for cases where the experience data is credible in overall terms, but insufficiently so to fully capture heterogeneity, then our client is protected, and
- for cases where the experience data is strongly credible the postcode model will be overridden by the experience data.