Allowing for shocks in portfolio mortality models

by Stephen Richards

The COVID-19 pandemic creates a challenge for actuaries analysing experience data that includes mortality shocks. Without sufficient local exibility in the time dimension, any analysis based on the most recent data will be biased by the temporarily higher mortality. Also, depending on where the shocks sit in the exposure period, any attempt to identify mortality trends will be distorted. We present a methodology for analysing portfolio mortality data that offers local flexibility in the time dimension. The approach permits the identification of seasonal variation, mortality shocks and late-reported deaths (OBNR). The methodology also allows actuaries to measure portfolio-specific mortality improvements. Finally, the method assists actuaries in determining a representative mortality level for long-term applications like reserving and pricing, even in the presence of mortality shocks. Results are given for a mature annuity portfolio in the UK, which suggest that the Bayesian Information Criterion (BIC) is better for actuarial model selection in this application than Akaike's Information Criterion (AIC).