Waldron 2004 — Heterogeneity in Health and Mortality Risk Among Early Retiree Men

Summary

Examines heterogeneity in health and mortality risk within the population of male Social Security retired-worker beneficiaries who claim before the full retirement age (65 for all cohorts studied). Using the Survey of Income and Program Participation (SIPP) matched to Social Security Administration (SSA) administrative data (health analysis) and the 1973 Current Population Survey (CPS) Exact Match (mortality analysis), the paper tests three competing models of early retiree composition: a health dichotomy (poor-health minority + healthy majority), a benefits-optimization model (self-selection on mortality risk produces a continuous gradient), and a health trichotomy (three distinct groups). The central finding is that the majority of early retirees are in poorer health and have higher mortality risk than age-65 retirees — reversing the conventional wisdom that only a minority of early retirees are health-compromised.

Key Claims

• Early retirees decompose into three groups: (A) extremely poor health/high mortality, (B) intermediate, (C) health/mortality equal to age-65 retirees. The trichotomy model fits the data better than the simpler dichotomy view.
• The majority of early retirees fall in groups A or B, not C — contradicting studies that find most age-62 claimants self-report good health (which conflates absolute health levels with relative health compared to age-65 retirees).
• The earlier men retire, the worse their health and mortality risk on average; even men in the top earnings quartile who claim at exactly age 62 have greater mortality risk than most later early retirees.
• The small group with health equal to age-65 retirees (Group C) is concentrated among men retiring at age 64 in the top earnings quartile.
• Early retirees with non-Social Security pension income in the top half of the distribution are equal in health to age-65 retirees; those below the median are not.
• Life expectancy differences by education widened over time even after controlling for changing educational composition by birth cohort.
• The benefits-optimization mechanism (individuals with below-average life expectancy self-select into early claiming because the actuarial reduction is implicitly favorable for them) is supported as a contributing driver alongside poor health.

Concepts Introduced or Extended

• Select and Ultimate Mortality Tables
• Income-Mortality Gradient
• Period Mortality

Entities Mentioned

• Hilary Waldron
• Social Security Administration
• Office of the Chief Actuary

Quotes

"The majority of early retirees are in poorer health and have higher mortality risk than that of age 65 retirees, and only a minority have health and mortality risk as good as that of age 65 retirees."

"If the data show that the majority of early retirees are in poorer health than age 65 retirees, then projections of their productivity may need to be adjusted downward to allow for the dampening effect of poor health on worker productivity."

My Take

The key methodological contribution is the shift from measuring absolute health levels (do early retirees report poor health?) to measuring relative health (are early retirees in worse health than age-65 retirees?). Prior studies found most age-62 claimants self-report good health — technically true but misleading, because the comparison group matters. This relative framing ties directly to how select-and-ultimate mortality tables work: the question is never whether Disability Insurance (DI) or early retirees are sick in absolute terms, but how their mortality compares to a reference population over time. The paper also provides early empirical grounding for the benefits-optimization / adverse-selection mechanism at age 62, which matters for DI research because it means the age-62 early retiree population is not a clean healthy counterfactual for DI beneficiaries.