Definition
Expected longevity is the true within-cohort gap in remaining life expectancy, computed by tracking actual individuals forward through time with their evolving health, wealth, and behavioral characteristics. It contrasts with period life expectancy (LE), which applies current-year age-specific death rates to a hypothetical cohort and implicitly treats each person's socioeconomic status (SES) as static throughout life.
Key Ideas
- Period LE overstates SES gradients: Because high-SES individuals tend to improve their health and wealth relative to low-SES peers over the life course, period LE attributes cross-sectional SES differences to the entire remaining lifespan — inflating the apparent causal effect.
- Pijoan-Mas and Ríos-Rull (2014): Using a hazard model with time-varying endogenous covariates (health, wealth, income, marital status, smoking) estimated on Health and Retirement Study (HRS) data (white Americans 50+, 1992–2008), they find period LE overstates the income gradient 7× and the wealth gradient 3× relative to expected longevity.
- Three-channel decomposition: For education (men), expected longevity gaps are partitioned into (a) initial health stock at age 50 (
26%), (b) differential health trajectory after 50 (74% — dominant), and (c) mortality conditional on health (0%). Education's effect operates entirely through health accumulation, not through a direct survival effect once health is held constant.
- Marital status and smoking differ: For these characteristics, conditional mortality explains one-third to one-half of the gap — consistent with behavioral/social mechanisms beyond health accumulation.
- Widening gradients 1992–2008: Expected longevity gap between college and no–high-school (HS) white men grew +1.7 years; wealth Q5–Q1 grew +2.1 years.
How It Works
Period LE at age x is computed from the current cross-section of age-specific death rates applied to a hypothetical individual who instantly "jumps" through all ages. Expected longevity tracks a real individual whose covariates evolve stochastically — using a hazard model estimated on longitudinal panel data, survival probabilities are integrated over the simulated covariate paths.
Why It Matters
Descriptive studies reporting large SES-mortality gradients using period LE substantially overstate the within-individual causal effect. A 7× inflation of the income gradient means that much of the dramatic income-mortality correlation reflects compositional dynamics (e.g., higher-income people are simultaneously healthier), not the marginal effect of income on survival. This distinction matters for policy design: interventions that raise income without improving health pathways will have much smaller effects than period-based estimates suggest.
Open Questions
- Whether the decomposition structure (health trajectory dominant, conditional mortality null for education) holds for non-white Americans or younger cohorts.
- How expected longevity gaps interact with Social Security redistributive claims — if the LE gap is smaller than period measures imply, progressive benefit formulas may be less offsetting than believed.
Related
Sources