Added Worker Effect

Definition

The Added Worker Effect (AWE) is the predicted increase in a spouse's labor supply when the primary earner experiences a permanent or persistent adverse income shock — disability, unemployment, or early death. The mechanism is a standard income effect in the neoclassical household model: if leisure is a normal good, a fall in household income raises the marginal utility of earnings for all household members, inducing the secondary earner to work more. The AWE is also predicted by forward-looking models: a spouse who expects to outlive a disabled partner has stronger incentive to build labor market attachment to provide for future needs.

Key Ideas

Income vs. substitution: The AWE is a pure income effect, not a substitution effect. The spouse's own wage has not changed; what changes is household income, which shifts the budget constraint inward and reduces the optimal leisure of both spouses.
Government programs crowd out the AWE: Social insurance transfers — Disability Insurance (DI), Unemployment Insurance (UI) — replace part of the lost income, reducing or eliminating the budget constraint shift that would otherwise trigger the AWE. Chen (2012) estimates DI crowds out wife's labor force participation (LFP) by 6%, growing to 8.4% after 5 years.
Comparison: UI crowd-out: Cullen and Gruber (2000) estimate wives would work 30% more hours if UI did not exist during husbands' unemployment spells — a larger crowd-out, likely because unemployment is transitory (wives need to supply labor immediately), while disability is permanent (wives may defer adjustment or rely on DI indefinitely).
Pre-award dip as AWE evidence: In Chen (2012), both wives of accepted and rejected DI applicants reduce labor supply ~1 year before the husband's award decision — before DI income arrives. This timing shows the AWE response is triggered by the disability shock itself, not by the DI award outcome.
Heterogeneity: AWE and its crowd-out are larger for: (a) younger wives (higher baseline labor force attachment → more room for response); (b) wives of "healthier" (Stage 5, musculoskeletal) husbands — possibly because sicker husbands require more spousal caregiving time at home, reducing the wife's ability to increase labor supply even if financially motivated.
Caregiving offset: A husband's poor health may reduce the wife's valuation of market time via increased caregiving demands or changes in preferences for joint leisure — forces that work against the AWE. The net effect is ambiguous; most empirical studies find a positive AWE for husbands with substantial functional limitations.
Distributional effect: Crowd-out of the AWE falls primarily on the extensive margin (LFP), not the intensive margin (hours among workers). Wives who drop out tend to be lower earners; the conditional mean earnings of wives who remain employed may rise even as participation falls.
Causal AWE identification via Norwegian judge lottery (Autor, Kostøl, and Mogstad 2015): Using quasi-random assignment of Norwegian DI appellants to judges (75 judges, 1994–2005), AKM provide the first instrumental-variable (IV)-identified causal estimate of the DI-specific AWE. Denied applicants' spouses earn ~ $6,000 more in year 1 and$ 10,000–$12,000 more in years 2–4. This is the "direct AWE" response (denial withdraws DI income → spouse must supply more labor), as opposed to Chen's "crowd-out" framing (allowance supplies income → spouse works less). The two designs are complements: one instruments for the presence of DI income, the other for its absence.
Marital-status self-insurance: The AKM (2015) results show that the AWE is powerful enough to fully offset the consumption loss from DI denial for married couples — household consumption is statistically unchanged after denial for married applicants. For single applicants, with no spouse to supply the AWE, denial causes a large consumption drop. This heterogeneity implies DI's consumption-insurance value is ~3.4× higher for singles than for married applicants.
PSID null AWE after disability onset (Meyer and Mok 2013): In 42 years of Panel Study of Income Dynamics (PSID) data, wives of Chronic-Severe disabled men show no statistically significant increase in work hours after disability onset. This is consistent with DI/Supplemental Security Income (SSI) transfers crowding out the budget-constraint shift that would otherwise trigger the AWE — the same mechanism Chen (2012) identifies for DI recipients — but the PSID sample does not allow distinguishing true AWE absence from transfer crowd-out. The null is informative as a complement to the AKM (2015) Norwegian evidence: the AWE appears relatively small or crowded out in the U.S. PSID context, even without the Norwegian IV design.
Null AWE for husbands of disabled wives (Meyer and Mok 2014): Symmetric to the MM (2013) null, husbands of disabled wives in the PSID also show no significant AWE — neither in hours, earnings, nor extensive-margin work, across all disability groups. Distribution analysis reveals large heterogeneity (some husbands work more, some less, with interquartile range (IQR) and 90–10 spread widening over the disability spell), but the median change is near zero or slightly negative. This suggests that heterogeneous responses cancel in the aggregate rather than that no individual responds. The null is consistent with DI/Supplemental Nutrition Assistance Program (SNAP) transfer crowd-out, with caregiving demands offsetting income-effect incentives, and with women's lower pre-disability earnings making the household income shock smaller than in the male-disability case.
Reversed AWE trigger via positive insurance shock (Boyle and Lahey 2016): The AWE can be triggered not only by a negative income shock to the primary earner but by a positive insurance shock that causes the primary earner to exit employment. When older male veterans gained Veterans Affairs (VA) health insurance in the mid-1990s expansion, husbands left employment (losing employer-sponsored insurance [ESI]), causing their wives to increase labor supply — by +1–2 pp employment probability, ~+0.5 hrs/week, and +3–4% earnings. The effect is concentrated entirely among low-socioeconomic-status (SES) wives (high school (HS) or less education, low non-housing wealth, no pension), consistent with an insurance coverage-loss channel: wives enter or intensify employment to obtain their own ESI. High-SES wives, who have alternative coverage options or face no binding financial constraint, show no significant response. This is the "spousal lock" corollary to standard employment lock: the wife's employment is locked in to secure health insurance after the husband's exit strips the household of employer-sponsored coverage.

How It Works

In the standard unitary household model, the household maximizes a joint utility function over combined consumption and leisure. A negative income shock (disability, unemployment) shifts the budget constraint inward, reducing optimal consumption and leisure for both spouses. If leisure is normal, the wife's optimal labor supply increases. A government transfer that replaces the lost income partially restores the original budget constraint, crowding out part of the predicted AWE.

In the forward-looking lifecycle model, the mechanism is subtler: if a wife expects to outlive her disabled husband, the expected duration of her household income shortfall extends to her remaining lifetime (not just the husband's disability spell). This permanent-income channel implies a larger AWE than the contemporaneous income effect alone.

Why It Matters

DI's household insurance value: By crowding out the AWE, DI insures not only the disabled beneficiary but also spouses against being forced into unwanted additional labor market participation. This is an underappreciated dimension of DI's insurance value beyond what the standard individual-level welfare analysis captures. See Nonhealth Risk and DI Insurance Value.
Benefit calculation: DI's spousal benefit (up to 50% of the worker's Primary Insurance Amount (PIA), with a $12,000/year earnings ceiling in 2005) interacts with the AWE: the earnings ceiling creates a 50-cent implicit tax on spousal earnings above the threshold, potentially amplifying the crowd-out among wives who are near that ceiling.
Program complementarities: Understanding AWE crowd-out is necessary for correctly evaluating any social insurance program that targets one household member — the true household welfare calculation must account for the secondary labor supply response being displaced.

Open Questions

Does the AWE crowd-out vary with the nature of the husband's impairment (caregiving-intensive vs. not)?
Is the crowd-out symmetric — does wife's DI receipt crowd out husband's AWE?
How does the crowd-out change when both spouses are attached to the labor market (high wife LFP eras vs. low)?
Does the 6% crowd-out estimated by Chen (2012) generalize to the current DI program, given rising female LFP since the 1980s?