Unmasking Mortality Trends
Modelling Mortality under Anti-selective Lapse in Universal Life Insurance
Will my life business benefit from mortality improvements, or is mortality in my portfolio deteriorating?
This important question life insurers and reinsurers seek to answer when analysing their experience data. The good news and the bad news is “both”.
In a regular book of universal life insurance, there are multiple forces at play at the same time:
- mortality improvements
- deterioration of mortality caused by anti-selection.
Without the right tools, it is difficult to separate the two competing trends. That’s why it is important to use a multivariate time-varying approach to analysing mortality.
In a recent client project on universal life insurance, we were able to unmask the anti-selective deterioration of mortality by policy year. To do so, it is necessary to first control for several other important factors:
- Get the age shape of the mortality table right. Don’t rely on an industry table but fit a mortality model which properly describes portfolio-specific mortality.
- Identify all relevant risk factors and quantify their impact. Properly modelling mortality differentials by gender and smoking status, as well as all other important risk categories is necessary to control for changes in the composition of the portfolio over time.
- Select the right selection shape. As new business is put on the books, it masks any deterioration because freshly underwritten business has lower mortality. Using the data to calibrate a selection pattern for the portfolio is important.
- Capture year-on-year mortality trends in the model explicitly.
- With the right baseline, we can now isolate the anti-selective deterioration of mortality by policy duration, and even show that it is different for level cost of insurance and increasing cost of insurance business.
These five steps are mapped out in detail in our Research Report.
