The fifth edition of WTW’s life insurance industry mortality study (TOAMS 5) covers calendar years 2015 through 2018. Aggregate results experience was consistent with TOAMS 4 (2011 – 2014) on a claims count basis but appreciably lower on a benefits paid basis (accounting for historical mortality improvement).
Under-reporting of claims at highest ages has been observed in past studies, but TOAMS 5 is the first time we have studied the trend in attained age by issue year. The study shows a statistically significant interaction on experience between the attained age of the customer and the issue year of the policy. The under-reporting effect diminishes for newer policies with no significant difference for policies issued after 1960 (Figure 1).
While Social Security sweeps have improved experience study data quality, underlying experience in TOAMS 5 shows that claims counts decrease at the highest ages for these older policies. This suggests that the sweeps are unable to identify all deaths, leading to under-reporting of claims above age 100, with the severity of under-reporting increasing with policy age. As a result, TOAMS 5 experience is credible up to an attained age of 95.
Contrary to prior TOAMS analyses, mortality experience between 2015 and 2018 was highest for universal life products without guarantees (ULNG). Notably, ULNG exhibited higher experience than whole life policies, with the gap increasing between ages 55 and 75 before reversing and exhibiting wear-off. Another distinct finding was the parabolic shape to Term and universal life with secondary guarantees (ULSG) curves (decreases then increases) (Figure 2).
Indexed universal life (IUL), variable universal life (VUL) and ULSG products exhibited lower relative mortality than whole life at younger attained ages, but differences diminished at older attained ages.
These examples illustrate differences in mortality by product, with some differences persisting at high attained ages. And while product differentials for level term still exist above age 95, there was limited exposure at these older attained ages.
TOAMS 5 experience demonstrated decreasing mortality experience by face amount and with further distinctions between term and permanent products (Figure 3).
Mortality decreases were most prominent below $250,000 before leveling off (a cohort dominated by permanent products). Exposure above face amounts of $250,000 showed higher concentrations of level term, which exhibited lower experience than permanent products; however, the discount on level term started reversing above $1 million, resulting in higher mortality than permanent products above $4 million. This suggests that some degree of anti-selection may exist for extremely large policies, as level term requires lower premiums to sustain the death benefit.
Where customers extended a fixed-term policy, experience shows significant anti-selective mortality in the early years after extension, which is most likely associated with people in poor health. These findings match expectations and other industry data points.
In these initial years, mortality jumps by a factor of two to nine depending on the ratio of premium increase. While over time, the mortality experience does converge, it doesn’t do so completely. Even after 11 years, some level of premium remains (Figure 4).
depending on the ratio of premium increase. While over time, the mortality experience does converge, it doesn’t do so completely.
It is generally acknowledged that the impact of preferred underwriting wears off over time, but limited experience exists to support assumptions for 3- and 4-class nonsmokers due to the more recent introduction of these policies.
TOAMS 5 findings suggest that the wear-off of the preferred risk discount appears to be age-related as opposed to duration-based; however, the pattern varies materially by the preferred risk class (Figure 5). As available experience does not demonstrate that the impact is fully worn off at a certain age, there is significant judgment involved in setting the assumptions.
| 2-class | 3-class | 4-class | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Preferred | Residual standard | Best preferred | 2nd preferred | Residual standard | Best preferred | 2nd preferred | 3rd preferred | Residual standard | |||
| Initial discount | -35% | 0% | -45% | -36% | 0% | -45% | -35% | -21% | 0% | ||
| Final discount (@AA95+) | -12% | -25% | -13% | -25% | -12% | -2% | |||||
| % discount worn off | 66% | 44% | 64% | 44% | 66% | 91% | |||||
| % discount remaining | 34% | 56% | 36% | 56% | 34% | 9% | |||||
Most classes show similar results to prior TOAMS studies except for 2-class structure (i.e., one preferred risk class) where the percentage discount is more significantly worn off by attained age 95 in TOAMS 5.
The mortality trends observed in TOAMS 5 should be taken as descriptive rather than prescriptive, even where they largely confirm and quantify prevailing viewpoints. While predictive modeling allows for the control of distributional differences for the variables included in our model (e.g., product, underwriting class), we still observed unexplained differences by company, which are likely due to factors not available in our data (e.g., underwriting criteria, adherence to underwriting criteria, distribution channel, target market).
The full version of TOAMS 5, which includes many more findings in addition to those discussed in this article, is now available to non-participants for purchase. Further work on TOAMS 5 will look to investigate additional topics, such as the impact of lapses and surrenders as well as joint life mortality.
