Meyer and Mok 2013 — Disability, Earnings, Income and Consumption

Summary

Using 42 years of Panel Study of Income Dynamics (PSID) data (1968–2009) on 6,741 male household heads, Meyer and Mok document the economic consequences of disability onset in a Poisson/linear fixed-effects event study. They classify disabled individuals into four groups by persistence and severity and find that Chronic-Severe disability — 17% of the disabled, ~9–15% prevalence at ages 50–56 — produces a 76% earnings decline, a 28% post-transfer income decline, and a 25% food-and-housing consumption decline over 10 years. Crucially, time-use evidence establishes that the consumption drop is real: extra time goes to TV, medical care, and sleep, with no home production offset, and diet quality deteriorates by 10–15% on key vitamins.

Key Claims

1. Four-group taxonomy: One-Time (24%), Temporary (31%), Chronic-Not Severe (28%), and Chronic-Severe (17%) of ever-disabled male PSID household heads, classified by whether more than half of post-onset years are work-limited (persistence) and whether the mean post-onset severity score exceeds 0.5 (severity ratio).
2. Chronic-Severe economic consequences at year 10: earnings −76%, post-transfer income −28%, food + housing consumption −25%, food consumption −18%; ~65% never return to work; ~20% poverty rate.
3. Social insurance gap: Public transfers rise to ~$14,000/year for Chronic-Severe by year 10; ~50% receive Social Security Disability Insurance (SSDI)/Supplemental Security Income (SSI); transfers offset only ~46% of the pre-transfer income gap, leaving a substantial residual consumption drop.
4. No home production substitution (Aguiar-Hurst channel): For Chronic-Severe, extra time goes to TV watching (+10.6 hrs/wk), medical care (+7.2 hrs/wk), and sleep (+6.8 hrs/wk) — not home food production or other home production activities. The time-use evidence rules out the hypothesis that measured consumption declines merely substitute for more home production.
5. Diet quality deteriorates: Chronic-Severe households have 10–15% lower vitamins A, C, and E intake; consumption decline is real, not measurement artifact.
6. Pre-onset anticipation pattern: Earnings, income, and consumption begin declining before disability onset, driven by deteriorating health (not unemployment shocks); the anticipation is consistent with a persistent health trajectory, not a sudden shock model.
7. Wives do not increase work hours (null added worker effect, AWE): In the PSID, wives of Chronic-Severe men show no statistically significant increase in work hours after disability onset — consistent with either no Added Worker Effect in this population or the AWE being crowded out by concurrent disability insurance (DI)/SSI income receipt.
8. Transfer under-reporting correction: Program-specific reporting rates from Meyer and Sullivan (2003, 2009) applied to scale up PSID benefits; SSDI/SSI are under-reported by ~20–40% in surveys; corrected estimates increase measured income replacement substantially.
9. Chetty (2006) optimal benefit calibration: With $\Delta c/c = -0.25$ (from consumption drop) and $D = 0.12$ (application rate), current DI is below the optimal benefit level for $\gamma \geq 3$ unless $\varepsilon_{D,b} \geq 0.66$ — well above the literature median (${\sim}0.49$). For most plausible calibrations, DI undercompensates severely disabled workers.

Concepts Introduced or Extended

• Morbidity-Mortality Distinction — no-home-production-substitution finding and diet quality decline confirm that disability is a genuine well-being shock, not a measurement artifact
• Nonhealth Risk and DI Insurance Value — provides the consumption-drop ($\Delta c/c = -0.25$) empirical input to the Chetty (2006) optimal benefit formula and evidence that transfers offset only ~46% of income loss
• Added Worker Effect — PSID-based null finding for wives of disabled men; consistent with AWE being crowded out by DI/SSI transfers
• DI Replacement Rate — ~46% transfer offset for Chronic-Severe; ~50% SSDI/SSI receipt at year 10

Entities Mentioned

• Bruce D. Meyer
• Wallace K.C. Mok
• Manasi Deshpande
• David Autor

Quotes

"The consumption of the chronically severely disabled falls by about 18 percent for food at home and 25 percent for food and housing."

"The additional time of the disabled is spent mostly on TV watching, medical care, and sleep."

My Take

The paper's most durable contribution is the combination of transfer-corrected income accounts with time-use evidence that rules out home production substitution. This methodological pairing closes the main loophole in consumption-based welfare analysis: the objection that survey consumption measures miss household production responses to disability. The Chetty optimality exercise is illuminating — and its conclusion that current DI is likely sub-optimal will age better than most calibration exercises because $\Delta c/c = -0.25$ is a direct estimate, not a modeling assumption. The main uncertainty is the coefficient of relative risk aversion $\gamma$, which is not identified from this data. The null AWE finding in the PSID may reflect the same DI-crowd-out mechanism documented by Chen (2012) rather than a true absence of the income effect on spouse's labor supply preferences.