From academic papers to mainstream media, reports on changes in life expectancy abound.1 While the U.S. population has experienced overall positive mortality improvements over time, disparities remain evident among different sub-groups.
The lifespans of defined benefit (DB) pensioners in particular may deviate from those of the general population, with distinct characteristics influencing their mortality. Digging even deeper, different pensioner groups may have their own specific mortality-influencing characteristics.
When considering baseline mortality assumptions of DB pensioners, insurers often distinguish between the general population and various socioeconomic groups or industry segments. This differentiation is less common for mortality improvement assumptions. However, evidence suggests that distinguishing between various groups should, at least to some extent, be applied to these assumptions as well. This tailored approach allows for more accurate modeling of the mortality improvements and liabilities of DB pension plans.
Often derived from overall U.S. population data, general mortality improvement assumptions may not capture the nuanced characteristics of DB pensioners.
69É«ÇéƬ research indicates that adjusting population mortality improvements when projecting pension liabilities can help ensure a more accurate model and reduce the risk of understating pension liabilities.
This paper explores the Society of Actuaries (SOA) research on life expectancy across socioeconomic groups2and examines the potential implications for the pension industry.
The pension and insurance industry has traditionally relied on U.S. general population-level mortality improvements. For example, a commonly used scale for projecting future mortality rates is the Society of Actuaries’ (SOA) Mortality Improvement Scale MP-2021, which is calibrated to the U.S. population.
However, U.S. population data may not fully represent DB pensioners. Consider this: Of the U.S. population age 60 and older, working less than 30 hours a week, less than half received an income source from a DB pension in 2013.3
DB pensioners, characterized by distinct employment histories and healthcare access, may have different mortality improvement dynamics than the average U.S. person, challenging the validity of a one-size-fits-all approach. For instance, DB pensioners were healthy enough to earn a DB pension through years of sustained, regular employment and likely enjoyed employer-provided healthcare.
The individual characteristics of different pension groups can also influence life expectancy. In private-sector retirement plans, for example, participants in white collar jobs tend to have lower rates of mortality than those in blue collar positions.4
It is therefore reasonable to assume that it is not always appropriate to use general population mortality improvements to project pension liabilities.
The SOA sponsored research to explore differences in life expectancy, and thus implicitly mortality improvements, by looking at different socioeconomic groups in the U.S, using a Socioeconomic Index Score (SIS) for 3,000+ counties across the country. The mortality analysis by socioeconomic quintile and decile covers 1982 –2019, and therefore does not include COVID-19-related mortality.
The research reveals that, while life expectancy has been increasing, mortality improvements have varied for each quintile, resulting in an increasing inequality in life expectancy within the total U.S. population.
Figure 1 shows that over the last three decades, the range in life expectancy from the lowest to the highest socioeconomic groups has grown materially. In 1983, the difference was relatively modest at 0.6 years and 0.3 years for males and females, respectively, whereas by 2018 it had increased to 2.7 years and 2.4 years, respectively. Over the study’s period, 65-year-old males and females in the highest socioeconomic group have realized more than two more years of life than those in the lowest socioeconomic group.
Figure 2 illustrates how the mortality improvement rates of different socioeconomic groups compare to the aggregate national data over a 30-year period, 1988-2018.
Similar to the life expectancy data, the research reveals that historical mortality improvements have differed by socioeconomic group. The extent of the difference depends upon the age and sex of the individuals and the time period being considered. While not always reflected in all combinations, in general, mortality improvement rates have been higher for higher socioeconomic groups and lower for lower socioeconomic groups.
While the SOA research is important for the U.S. pension industry, highlighting mortality differences by socioeconomic group, it should be noted that it analyzes the general U.S. population rather than focusing specifically on DB pensioners. Relying on general population improvements runs the risk of overstating or understating the improvements for a specific pensioner group.
This could lead to potential mispricing of pensioner liabilities and have a significant financial impact when assessing a transaction.
Figure 3 illustrates the difference in the liabilities5 when using improvements from different socioeconomic groups compared to the MP-2021 scale calibrated to U.S. population data.
The liability for a pensioner aged 65 is over 1% more when using the improvements from the highest socioeconomic group (quintile 5), compared to the MP-2021 scale.
Using population improvements without adjustments may lead to inaccurate assumptions for specific pensioner groups, such as overestimating the improvements of lower socioeconomic groups and, perhaps more importantly, underestimating those of higher socioeconomic groups. If the MP-2021 scale is used for DB pensioners from high socioeconomic groups, the liabilities will likely be underestimated.
Considering the characteristics of a specific pensioner population not only enables practitioners to capture their mortality improvements more accurately, but also allows for more accurate liability cash flows and enhanced pricing capabilities.
It should be noted that the difference in liabilities and the true scale of improvement may be larger than shown in the study because it was conducted at a county level. Counties can be large and diverse in composition.
Different societal groups experience different mortality improvement rates, and this includes DB pensioners. Tailoring mortality improvement assumptions to specific pensioner populations is therefore crucial for accurate projections.
DB pensioners possess characteristics that differ from the general population, requiring practitioners to refine mortality improvement assumptions accordingly. Calculating pensioner liabilities using a single mortality scale, such as MP-2021 which is calibrated to the U.S. population data, increases the risk of misestimating mortality improvements.
Understanding the socioeconomic differences between a pensioner population and the general population is vital for precise projections and for preventing negative financial impacts due to underestimating pension liabilities.
69É«ÇéƬ is equipped with both the technical expertise and financial security to manage risks for as long as policyholders survive. Contact us to find out more about 69É«ÇéƬ’s longevity de-risking solutions.
1. Forbes; 4 Reasons Life Expectancy Has Increased in the Past 200 Years ()
National Institutes of Health; Life expectancy in the U.S. increased between 2000-2019, but widespread gaps among racial and ethnic groups exist ()
Lauren Medina, et al.; Living Longer: Historical and Projected Life Expectancy in the United States , 1960 to 2060 ()
2. Magali Barbieri, Ph.D.; Mortality by Socioeconomic Category in the United States ()
3. Tyler Bond and Frank Porell, Ph.D.; Examining the Nest Egg ()
4. Society of Actuaries; Pri-2012 Private Retirement Plans Mortality Tables Report ()
5. Calculated using the Pri-2012 base mortality table. Annuity factors calculated as of 2023. 5% discount rate.
