As I have noted in a recent blog post, life expectancies at older ages have increased notably in the US over the past several decades, providing justification for raising formal retirement ages in pension systems such as Social Security as part of reform proposals to close funding gaps. Yet, some have pushed back on this proposal, citing several studies that purport to show that although average life expectancies have indeed increased, this good fortune has not been equally shared. People in lower socio-economic strata, they argue, have fallen behind the well-to-do. While increasing life expectancy inequality could be addressed by making the benefit formulas more progressive, a more fundamental question remains: is the counter-claim true?
There are indeed many studies that claim that the life expectancies of the poor have not increased as fast as those of rich. (No one denies that the poor have always had lower life expectancies than those with higher incomes, and that fact is already reflected in the progressive structure of Social Security’s survivor, disability, and retirement benefits.) Most of these studies, however, suffer from data and methodological issues. Some are based on survey data with small sample sizes, which lowers their statistical power. Another problem with survey data is that the measures of socio-economic status used are quite incomplete. For example, educational level attained—while generally recorded accurately—is only somewhat related to lifetime income; consider, for instance, college dropout Bill Gates compared to a community college professor with a PhD. Survey data on current income, another commonly used measure, is imprecise. And even when drawn from administration data sources such as tax returns, it is not well correlated with lifetime income owing to its high variability over time. In fact, current income is a particularly poor measure for studying life expectancy inequality, because low current income may be caused by temporary or recent poor health, thereby introducing a bias of unknown size into the analysis.
One study often cited by those who argue for increasing inequality in life expectancy does not suffer from most of these problems. Published in 2007 by a researcher at the Social Security Administration (SSA), it uses agency data on the mortality experience of male Social-Security-covered workers aged 60 and older from 1972 through 2001, along with their average relative earnings—if positive—from ages 45 to 55 comparing the bottom and top halves of the distribution of average relative earnings. Based on a large 1 percent sample of all workers and using the ten-year average of positive earnings, the results should be more reliable. The study found that male workers born in 1941 who lived to age 60 with earnings in the top half would be expected to live 5.8 more years than those in the bottom half. In contrast, for male workers born in 1912, that difference was only 1.2 years. Yet even this study has limitations. Over time, the sample composition changed with the inclusion—by law—of new categories of workers from the non-profit and federal government sectors, as well as low-wage self-employed, farm, and domestic workers, possibly introducing a bias. It also includes disabled beneficiaries, whose mortality patterns are not relevant to setting retirement age policy. Finally, the study relies on assumptions about future mortality, meaning its findings are based in part on assumed projections rather than actual experience.
A much more recent SSA study, covering all all retired-worker beneficiaries from 1995 to 2022, examined annual relative mortality ratios across various age groups and earnings quintiles based on 35-year career-average earnings (Average Indexed Monthly Earnings, or AIME). This study appears free of the earlier issues: it includes zero earnings years, is not based on a sample, offers a complete long-run earnings history, and excludes disabled workers. Aside from the age 62 to 64 and female groups—influenced by the legislated rise in the retirement age, which reduced the incentive to retire early, and by the trend toward higher female labor force participation and earnings—the population characteristics appear relatively stable. The study’s findings on relative mortality rates for the five earnings groups for men at ages 65 to 69 from 2000 to 2022 are presented below.
As shown, there is little change over the past two decades in the relative mortality experience of Social Security retirement beneficiaries across lifetime earnings groups, aside from the lowest 20th percentile in the early years of the analysis. This finding of stability, based on the best data and methodology, challenges the consensus that life expectancy inequality is increasing. Using this study, policymakers can be confident that a uniform increase in the Social Security retirement age will be fair, requiring at most minor adjustments in benefit formula’s progressivity.
The post Is Life Expectancy Inequality Increasing? appeared first on American Enterprise Institute – AEI.
