Brown, Choi, Coile, and Woodbury 2020 — Social Security and Financial Security at Older Age

Summary

Brown, Choi, Coile, and Woodbury (2020) synthesize five years of National Bureau of Economic Research (NBER) Retirement Research Center (RRC) research (2013–2018) on Social Security and financial security at older ages, published in the Social Security Bulletin. The article covers three major themes: (1) work, retirement, and Social Security claiming behavior — including causal effects of the earnings test and Full Retirement Age (FRA) reforms; (2) health and financial well-being — including the income-mortality gradient and education–Disability Insurance (DI) participation link; and (3) behavioral determinants of retirement saving — including auto-enrollment, present bias, and passive-saver evidence from Denmark.

Key Claims

Earnings test reduces employment by $\geq 3.7$ pp at ages 63–64: Gelber, Jones, and Sacks (2014; 2017; 2019) use sharp eligibility thresholds to identify the causal effect of the Social Security earnings test. At ages 63–64, the earnings test reduces the employment rate by at least $3.7$ percentage points (pp). Notably, earnings-test frictions persist after the test is lifted — workers continue to bunch earnings as if the test still applies at ages 70–71 — suggesting forward-looking rules of thumb distort behavior even beyond the binding age.
FRA reforms delay retirement and claiming: A 1-year increase in Switzerland's FRA delayed labor force exit $7.9$ months and claiming $6.6$ months; some delay reflected rule-of-thumb behavior (passively accepting the FRA as a default) rather than financial incentive response. A 1-year increase in Austria's early retirement age raised job exit by $0.4$ years and claiming by $0.5$ years. Norway's 2011 reform replacing implicit work taxes with actuarially neutral incentives raised registered employment $+26\%$ and earnings $+15\%$ among affected workers.
Psychology of claiming: Individual variation in loss aversion (measured by gambling choices) predicts $\approx 6$ -month earlier Social Security (SS) claiming. Longevity salience — prompting respondents to think of someone they knew who lived long — delays intended claiming by $9$ months. Informational messages designed to promote reflection delay planned claiming by $5$ – $10$ months across multiple experimental studies.
Health insurance access accelerates retirement: State guaranteed-issue and community-rating reforms that expand individual health insurance access raise the monthly retirement probability at age 63 by $2.2$ pp, nearly doubling the baseline rate. Workers in fair/poor health show even larger responses.
DI enrollment gradient: $6\times$ by education: Venti and Wise (2014) find DI enrollment is more than $6\times$ higher for workers with less than a high school (HS) diploma than for those with a college degree or more. College graduates are $25+$ pp less likely to claim Old-Age and Survivors Insurance (OASI) early. Education's correlation with DI runs primarily through cumulative health and wealth pathways — consistent with Poterba, Venti, and Wise (2017) decomposition.
Low wealth is highly persistent: $70\%$ of low-wealth individuals (under $\$50$ k total assets) remain low-wealth at death; $45\%$ of couples in the lowest earnings quintile have less than $\$100$ k at retirement; for many low-income households, Social Security is the only financial resource.
Medical shocks deplete wealth: Strokes reduce net worth by $\$4{,}682$ (low-wealth) to $\$59{,}290$ (high-wealth); new lung disease diagnoses reduce net worth by $\$9{,}986$ to $\$84{,}959$ . Other diagnoses have smaller effects.
Auto-enrollment raises retirement wealth by $\approx 10$ pp of first-year salary: For employees with 5+ years of tenure, automatic 401(k) enrollment raises plan balances by $10$ pp of starting salary; borrowing partially offsets the gain but net wealth still rises.
Present bias and exponential-growth bias reduce saving substantially: Eliminating both behavioral biases could increase retirement wealth by up to $70\%$ . A Thrift Savings Plan (TSP) default-fund switch (government securities → lifecycle fund) increased passive acceptance but lowered contribution rates — default selection has ambiguous effects depending on which default parameter is changed.
Danish evidence: $85\%$ passive savers, tax subsidies near-ineffective: Chetty et al. (2014) find $85\%$ of Danish savers are passive (do not respond to tax subsidies). $\$1$ of tax expenditure on subsidies raises total saving by only $1$ cent; automatic contributions are effective because they require no action. Active savers fully offset policy changes via cross-account transfers.
OASI trust fund depletion projected 2034: Under the 2019 Board of Trustees intermediate assumptions, OASI costs exceeded income for the first time in 2020 and trust fund reserves will be depleted in 2034 absent Congressional action.

Concepts Introduced or Extended

Entities Mentioned

Quotes

"Social Security, combined with Supplemental Security Income for very low-income households, provides a solid base of annuitized income and financial security through later life. For a sizable share of households, it is their only resource."

"Evident throughout RRC research is the strength of the relationship between education, income, wealth, health, functional ability, Disability Insurance enrollment, and mortality; and how individuals struggling in any one of these domains are more likely to be struggling in others as well."

"[T]he framing of information about saving options and implications influences behavior [in retirement saving] — as with choosing when to claim Social Security benefits."

My Take

This is a review article, not an original study, and should be read as a curated bibliography of NBER RRC work rather than primary evidence. Its value is in synthesizing and confirming a coherent picture: SS policy shapes behavior, socioeconomic status (SES) gradients in health and wealth are large and persistent, and behavioral architecture powerfully determines saving. The earnings test causal evidence (Gelber et al.) is the most policy-relevant original finding surfaced by this review — it directly tests whether the earnings test distorts labor supply and finds it does so substantially. The behavioral saving material (auto-enrollment, passive savers) is more directly applicable to 401(k) policy than to SS disability policy, but illustrates the importance of structural defaults for low-information households. The DI enrollment gradient ( $6\times$ education) provides an important complement to the detailed pathway decomposition in Poterba, Venti, and Wise (2017).