Employment Lock

Definition

Employment lock (also called "job lock") is the tendency of workers to remain in or extend employment specifically to retain access to employer-sponsored health insurance (ESI). Because ESI in the United States is tied to active employment rather than to the individual, workers who anticipate high health care costs face a private cost of labor force exit that is not reflected in wages or Social Security (SS) financial incentives. This creates a distortion in retirement timing decisions that is especially acute for workers approaching age 62 — the earliest Social Security Old-Age (OA) entitlement age — and dissolves naturally at age 65 when Medicare provides universal coverage.

Key Ideas

ESI is tied to employment status, not to the person. Exiting employment generally means losing coverage unless the worker qualifies for Consolidated Omnibus Budget Reconciliation Act (COBRA) continuation coverage (expensive, time-limited), a spouse's plan, Medicaid (means-tested), or marketplace coverage (pre-Affordable Care Act [ACA]: typically unaffordable for older workers in poor health).
Workers who receive new diagnoses of chronic conditions (diabetes, cardiovascular disease) face a discrete increase in expected health care costs $\to$ increase in the value of continuing ESI $\to$ increase in the implicit cost of retiring or claiming SS benefits.
The effect is concentrated near the OA early entitlement age of 62: workers who could claim SS choose continued employment instead. Workers already past 65 (Medicare) face no such constraint.
The effect is stronger among more-educated workers, who have greater discretion to delay and who face higher returns to health investment (Grossman 1972).
The effect is strongest for workers who receive a new, previously unknown diagnosis — they are surprised by new information about their health-care cost trajectory.

How It Works

The Mechanism

At any moment a worker could claim Social Security (OA at 62+, Disability Insurance [DI] if eligible), the worker weighs:

Benefit of claiming: SS income; reduction in health-impairing work demands
Cost of claiming: Loss of ESI; actuarial reduction in OA benefits for early claiming; loss of work-related social capital

When a new health diagnosis arrives, expected medical expenditures rise, making ESI more valuable. For workers near 62 who are still employed, this tips the cost-benefit calculation toward extended employment and delayed SS claiming.

Causal Evidence: Singleton (2024)

Using National Health and Nutrition Examination Survey (NHANES) random morning/afternoon medical exam assignment as an instrument for new health information:

Outcome	Ages 50–58	Ages 59–61
Change in SS entitlement (period 2)	$+1.03$ pp (n.s.)	$-7.79$ pp (sig. at $5\%$ )
Change in SS covered earnings (period 2)	$-0.53$ pp (n.s.)	$+9.11$ pp (sig. at $5\%$ )
Implied ratio (delay / employment)	—	$\approx 1:1$

The cumulative entitlement delay for ages 59–61 is $0.380$ years. The delay dissolves to $1.8$ pp (standard error [SE] $= 3.1$ , not significant) by period 5 — when most participants have reached age 65 and Medicare eligibility. This Medicare-at-65 falsification is the strongest evidence that the mechanism is insurance-driven, not health-driven.

Who is most affected (ages 59–61, period 2):

Non-previously-diagnosed diabetics: $-10.85$ pp (vs. $+2.78$ pp for diagnosed: $13.63$ pp differential)
College+ education: $-17.10$ pp differential vs. high school (HS)-or-less
Labor force participants and DI-insured workers: larger point estimates (not significant)

Comparison With Prior Literature

Benitez-Silva and Dwyer (2005), using the Health and Retirement Study (HRS), find that a diabetes diagnosis transition increases expected retirement age by $0.764$ years (SE $= 0.444$ ). Singleton's $0.380$ -year effect on actual SS claiming is roughly half as large — consistent with a gap between expectations and behavior, and/or with employment lock partially (not fully) bridging the intention-action gap.

Gruber and Madrian (1995) show that COBRA continuation coverage availability — which reduces the cost of leaving an employer — is associated with earlier retirement, providing the first direct evidence that ESI creates employment lock. Garthwaite, Gross, and Notowidigdo (2014) find that Tennessee's Medicaid disenrollment induced labor force entry among formerly covered adults, confirming the channel from health insurance access to employment decisions.

Spousal Labor Supply Corollary (Boyle and Lahey 2016)

Employment lock has a household-level counterpart. When the primary earner exits employment after gaining non-employer health insurance (HI), the secondary earner may be drawn into the labor market to supply household health insurance coverage. Boyle and Lahey (2016) exploit the mid-1990s Veterans Affairs (VA) expansion: as older male veterans gained VA HI and exited employment (losing ESI), their wives increased employment probability by 1–2 pp and earnings by ~$23–27/week. The effect is concentrated among low–socioeconomic-status (SES) wives (high school education or less), confirming that financial constraints — not preferences for joint leisure — drive the response.

This "spousal lock" operates symmetrically to benefit-side lock: just as a DI beneficiary remains on the rolls to retain Medicare, a wife remains employed (or increases effort) to retain access to ESI after the husband's employment exit removes the household's coverage anchor. The ACA marketplace should weaken this mechanism post-2014 by providing an alternative coverage path, analogous to how Medicaid expansion weakens standard employment lock for low-income workers.

Why It Matters

For Retirement Timing

Employment lock reveals that workers near OA eligibility respond not just to SS financial incentives but to the interaction between those incentives and the health insurance system. Policies that separate insurance from employment status (e.g., ACA Medicaid expansion, marketplace subsidies) may affect the timing of SS claiming independently of any changes in SS rules.

For Interpreting DI Caseloads

Workers who extend employment due to employment lock do not show up in DI application data during that extension period, even if they are health-impaired. This creates a timing distortion: the eventual DI application may arrive after a delay of one or more years relative to the onset of the qualifying impairment. Employment lock thus contributes to the onset-to-filing gap documented in the DI literature.

For Measuring Health Effects on Labor Supply

Health shocks reduce labor supply on average. Employment lock attenuates this effect for workers with ESI near 62: a subset of health-impaired workers extends employment, working against the naive correlation between health and labor force exit. Failure to account for employment lock overstates the responsiveness of labor supply to health.

For the ACA

The ACA's Medicaid expansion and marketplace subsidies reduced the cost of leaving employer coverage. For workers under Medicaid income limits, the employment lock penalty effectively fell to zero. For workers above Medicaid but below marketplace subsidy phase-out, it fell substantially. The $0.380$ -year delay in Singleton (2024) was estimated from 1999–2004 NHANES surveys — pre-ACA — and may have diminished since.

DI Lock and SSI Lock: The Benefit-Side Analog

"DI lock" is the disability-program analog: beneficiaries may remain on DI or Supplemental Security Income (SSI) rolls specifically to retain Medicare or Medicaid coverage, even if they have sufficient work capacity to exit. Congress addressed this concern by legislating continued Medicare eligibility for 93 months after a DI beneficiary leaves the rolls for work (added in 1984, extended since), and permanent continued Medicaid eligibility for SSI exiters under Section 1619(b) (1980, made permanent 1986), as long as disability continues and income/assets qualify.

Coe and Rupp (2013) provide the first systematic empirical test. They exploit state-level variation in (1) non-group health insurance market regulation (guaranteed issue + community rating = "strict"), (2) Medicaid buy-in programs, and (3) Medicaid generosity, using Social Security Administration (SSA) administrative data on DI and SSI beneficiaries.

Key results:

Aggregate DI lock is small: Little overall relationship between state insurance access and beneficiary work or exit rates. The legislative extensions appear to have worked in the aggregate.
SSI lock persists heterogeneously: Beneficiaries with moderate Medicaid expenditures (not zero, not catastrophically high) are more likely to work in states with strict non-group regulation or generous Medicaid eligibility — both policies reduce the gap between the insurance value of staying on SSI and the insurance available off the rolls.
DI lock persists for the privately uninsured: Beneficiaries without any supplemental private insurance (Medicare as sole coverage) remain DI-locked; Medicaid buy-in programs specifically alleviate this.
High-expenditure beneficiaries are unresponsive: Severity too high for financial insurance incentives to matter.
Exit rates are very low overall ( $\approx 1\%$ /year DI; $\approx 4\%$ /year SSI), so health insurance lock is a marginal amplifier around an already-large work-disincentive structure.

The Substantial Gainful Activity (SGA) cliff interaction: DI beneficiaries who could qualify for SSI after losing DI benefits (because SSI provides an income floor and Medicaid access for earnings between the SGA and $2 \times$ Federal Benefit Rate (FBR) $+$ $85) face a smaller effective work disincentive. Coe and Rupp find suggestive evidence that this SSI-dampening of the SGA cliff is real, but estimates lack precision.

Application-Entry Lock (Coe et al. 2011)

The benefit-side DI/SSI lock above concerns workers already on the rolls. There is a parallel application-entry effect: workers who might otherwise apply to Social Security Disability Insurance (SSDI) in order to gain Medicare do not apply when they can access affordable non-group health insurance. Coe et al. (2011) find that states with strict non-group market regulation (guaranteed issue + community rating) have $0.054$ pp lower within-state SSDI application rates ( $p < 0.01$ ). This is the demand-side mirror of DI lock: employment lock deters applications, not just exits. The paper identifies this effect using state-level panel variation in regulation, pre-ACA (1993–2009). The ACA's insurance market reforms would be expected to extend this effect nationally. See Geographic Variation in Disability Insurance.

Open Questions

Has the ACA substantially reduced employment lock near age 62? Post-ACA data on the age-62 SS claiming notch could test this.
Does employment lock generate adverse selection in employer health insurance? If sick workers stay employed for ESI while healthy workers exit freely, employer pools become sicker, raising premiums and reinforcing the lock.
How does employment lock interact with the DI onset-to-filing gap? Workers who use ESI to delay labor force exit may also delay DI application, complicating timing analysis.
Does employment lock affect the composition of the gap population (workers aged 62–64 who are too impaired for DI/SSI but rely on early Old-Age and Survivors Insurance (OASI))? If so, workers who lack ESI may enter OA earlier than those with ESI, creating insurance-driven selection into the gap.