Waldron 2007 — Trends in Mortality Differentials and Life Expectancy for Male Social Security–Covered Workers, by Socioeconomic Status

Summary

Using a 1% longitudinal sample of Social Security administrative records (Continuous Work History Sample (CWHS) + Numident death file), Waldron tracks mortality and cohort life expectancy (LE) for 294,451 male Social Security–covered workers born 1912–1941, observing deaths at ages 60–89 from 1972–2001. Earnings from ages 45–55 (relative to the national wage index) serve as a proxy for adult socioeconomic status (SES). The study finds that mortality differentials by earnings group have widened across successive birth cohorts, and that the rate of mortality improvement has been substantially faster for men in the top half of the earnings distribution than for those in the bottom half.

Key Claims

The odds of dying for men in the bottom half of the earnings distribution relative to the top half at ages 60–64 widened from 27% greater (birth years 1912–1915) to 84% greater (birth years 1936–1938) — a 57-percentage-point widening over 25 birth years.
Cohort LE gap at age 60: 1.2 years (1912 cohort) → 5.8 years (1941 cohort). Over 29 cohorts, the top half gains 6.5 years of LE at age 60; the bottom half gains only 1.9 years.
Period LE (1999–2001): at age 65, the top-half exceeds the bottom-half by 1.9 years; the top quarter exceeds the bottom quarter by 2.3 years. The gap narrows and reverses at older ages due to sample frailty.
At age 65, U.S. male Social Security–covered workers in the bottom quarter of the earnings distribution have LE comparable to the average Irishman. Men in the top quarter have LE comparable to the average Japanese or Swiss man.
The sample excludes zero earners at ages 45–55 (15.6% of men dropped). Because non-participation is strongly correlated with poor health (men not in the labor force at age 45 have ~9 fewer years of LE than participants — Rogot et al. 1992), the study underestimates the true population LE gap at the bottom.
Mortality differentials narrow with age within each birth cohort — this reflects survivor selection (sample frailty), not genuine health convergence. Frailer low-earners die before reaching older ages; the survivors are selectively robust.
Waldron does not control for sample frailty and explicitly designates the findings as preliminary on that account.

Concepts Introduced or Extended

Entities Mentioned

Quotes

"The segment of the male Social Security–covered worker population experiencing slower mortality improvement is large — that is, the entire bottom half of the population, rather than just a limited group of disadvantaged at the lowest end of the earnings distribution."

"Eliminating the gap in probabilities of death by socioeconomic status by lowering probabilities of death for lower-earning men would increase average male life expectancy in the United States."

My Take

The paper's most important contribution is methodological: it is the first study to use Social Security Administration (SSA) administrative cohort data (rather than survey data or cross-sectional records) to document the widening SES-mortality gradient across successive birth cohorts. The LE gap widening from 1.2 to 5.8 years across the 1912–1941 cohorts is a large finding, but the numbers for later cohorts (especially 1941, observed only at age 60) are almost entirely projections under a simplified extrapolation assumption. Waldron is unusually transparent about this and discourages reading the projections as forecasts.

The international comparison (US bottom quarter ≈ Ireland; US top quarter ≈ Japan) is the sharpest framing of the U.S. SES-mortality problem: the U.S. performs poorly in international LE rankings not because high earners are unhealthy but because low earners are. This finding precedes and is confirmed by Chetty et al. (2016), which extends the analysis through 2014 using period rather than cohort methods. Together they form a continuous evidentiary record of widening income-LE inequality in the U.S. from the 1970s through 2014.