Armour and Knapp 2021 — The Consequences of Claiming Social Security Benefits at Age 62

Summary

Armour and Knapp use nearest-neighbor matching on Health and Retirement Study (HRS) data to estimate the causal consequences of claiming Social Security at age 62 versus claiming later. Age-62 claimers exit the labor force at much higher rates, liquidate defined-contribution (DC) pension balances, and accelerate annuity income, but accumulate substantially less liquid wealth by their late 70s — driven mainly by later claimers continuing to build wealth, not by early claimers decumulating faster. No significant mortality or financial-hardship differences emerge through age 78, leaving open the key longevity-insurance question that only extends past the study window.

Key Claims

• Age-62 Social Security claimers are $40$ percentage points more likely to stop working by age 62 than matched later claimers; the employment gap persists into the early 70s.
• Early claimers are $16$ percentage points (pp) more likely to liquidate DC pension wealth near age 62, accelerating pension and annuity receipt alongside Social Security — a retirement bundling pattern.
• Annual household income is $\approx \$14{,}000$ lower for early claimers from age 62 through the mid-70s; the gap widens to $\approx \$23{,}000$ by ages 74–76 as later claimers' higher delayed-claiming benefits compound.
• Liquid non-housing wealth is $\approx 27\%$ lower for early claimers by age 70; the shortfall arises from later claimers accumulating more, not from early claimers spending down at higher rates.
• No significant mortality difference through age 78; no significant difference in financial hardship (food insecurity, medical debt) through the follow-up period.
• Heterogeneity: effects are larger for those with higher pre-62 earnings and those in physically demanding jobs (plausible occupational-health selection).

Method

Nearest-neighbor propensity-score matching (Abadie-Imbens estimator) on age-60 characteristics: health, wealth, earnings history, demographic controls, and expected claiming age. Estimand is the average treatment effect on the treated (ATET): what happens to age-62 claimers relative to what would have happened had they claimed later. Sample: HRS 1992–2016, RAND HRS Longitudinal File 2016; $n = 1{,}524$ after matching ($572$ age-62 claimants, $952$ later-claimant matches). Balance diagnostics confirm covariate balance post-matching.

Key identifying assumption: unconfoundedness — no unobserved variables at age 60 cause both the claiming decision and the outcomes. The authors defend this via the age-60 conditioning set and balance checks, but cannot rule out residual selection (e.g., health deterioration between 60 and 62 that is unobservable at 60).

Concepts Introduced or Extended

• Social Security Claiming Age
• Retirement Age Stickiness
• Employment Lock

Entities Mentioned

• Philip Armour
• David Knapp
• Health and Retirement Study
• Alicia H. Munnell

Quotes

"We find that individuals who claim Social Security benefits at 62 are 40 percentage points more likely to have stopped working by the time of their Social Security application."

"Rather than decumulating wealth more rapidly, early claimers simply fail to accumulate as much additional wealth as their later-claiming counterparts."

"The long-run consequences of early claiming — particularly around longevity insurance in one's 80s and beyond — remain outside the scope of this study."

My Take

The matching design is credible within its limits: conditioning on a rich age-60 covariate set and using the Abadie-Imbens estimator is defensible given the HRS data structure. The key limitation is that the HRS panel does not extend into the 80s for most study participants, precisely the period when the break-even calculus of delayed claiming most strongly favors later claimers (via longevity insurance). The null mortality result through age 78 is consistent with prior literature but cannot settle the longevity-insurance question. The retirement-bundling finding — claiming, work exit, and DC liquidation clustering at 62 — is a novel and policy-relevant observation, suggesting that raising the early entitlement age (EEA) would simultaneously delay retirement and preserve DC balances. Cross-read with Retirement Age Stickiness (Deshpande, Fadlon, and Gray 2021), which shows that full retirement age (FRA) changes shift claiming but not labor force exit: Armour and Knapp find the reverse at the EEA end — claiming at 62 does drive early labor exit.