Disability and Marital Dissolution

Definition

Disability and marital dissolution refers to the causal pathway by which the onset of a work-limiting or work-preventing disability destabilizes a marriage and raises the probability of divorce. In the economic framework of marital instability (Becker, Landes, and Michael 1977), disability operates through two channels: it reduces the value of the marriage by impairing the disabled spouse's contribution to household income and production, and the magnitude of the dissolution effect depends critically on how unanticipated the disability was. Singleton (2009) provides the first systematic heterogeneous analysis of the disability-divorce relationship, finding that the divorce hazard effect is strongest among young, educated men with work-preventing disabilities ( $+13.3$ percentage points (pp) after marital quality controls) and near zero among older disabled workers.

Key Ideas

The two-factor model: Disability's effect on divorce is a product of the ex ante disability hazard (how likely was this disability before it happened?) and the disability's impact on marital value (how much does it damage earnings and household production?). Anticipated disabilities — common among older workers with elevated baseline hazard — are already priced in to the marriage, so their realization changes the divorce decision less.
Heterogeneity by age: Among men, the 3-year divorce hazard gap between disabled and nondisabled is $+5.2$ pp at ages 29–40, $+4.1$ pp at 41–52, and near-zero at 53–64 (high education, any disability). The near-zero result at older ages reflects declining nondisabled earnings at those ages, which shrinks the realized earnings gap.
Heterogeneity by education: Within age groups, the divorce effect is consistently larger for higher-educated men. High-education men have lower disability hazard (less anticipated) and larger pre-disability earnings trajectories (larger earnings loss), both of which amplify the divorce-triggering impact.
Heterogeneity by severity: Work-preventing disability produces a dramatically larger divorce effect than work-limiting disability. The fully controlled estimate for young, educated men with work-preventing disability is $+13.3$ pp (Singleton 2009, Table 5).
Female results (Singleton 2009): The disability-divorce gradient exists for women in the predicted direction but is statistically insignificant throughout, consistent with smaller disability-induced earnings losses among women.
Reconciles Charles and Stephens (2004): That paper found no aggregate disability-divorce effect (in contrast to job displacement). Singleton's subgroup analysis explains the null: disability incidence is concentrated among older, less-educated workers where the divorce effect is smallest, so the aggregate estimate is diluted to insignificance.
Women's disability-divorce gradient (Meyer and Mok 2014): Using Panel Study of Income Dynamics (PSID) marriage histories (1985–2011) and 44 years of panel data, Meyer and Mok find that a wife's disability does significantly raise the divorce hazard for women — $+0.6$ pp in the long run (years 6–10) on a 1.4% annual base rate. This is large relative to the base rate and contrasts with Charles and Stephens (2004). Crucially, the effect is nonlinear in severity: Temporary and Chronic-Not Severe groups drive the hazard elevation; Chronic-Severe wives do not experience elevated hazard. Possible explanations: guilt effect for leaving a seriously disabled spouse, high divorce compensation costs, or marriage-specific capital (children's welfare) interacting with severe disability. The study examines the wife's disability and husband's divorce response — a different direction than Singleton (2009), which focused on husband disability.
Marriage as informal insurance: Disability onset is precisely the event against which spousal support provides informal insurance — yet it is also when the marriage is most at risk of dissolution, creating a fragility in informal risk-sharing arrangements.

How It Works

Mechanism: Earnings Loss → Marital Value Decline → Divorce

The primary mechanism is the earnings channel: disability reduces the disabled spouse's earnings permanently (fixed-effects regressions in Singleton 2009 show no recovery over 7 years), which decreases the value of the marriage below the sum of outside options. The earnings-loss mechanism is supported by:

The education gradient: more educated men have larger pre-disability earnings trajectories and suffer larger absolute earnings losses — approximately $40,000 (2004 dollars) in annual earnings lost by year 7 for young, educated men.
The severity gradient: work-preventing disability (zero or near-zero income) produces much larger divorce effects than work-limiting disability.

The Anticipation Effect

The disability hazard among men aged 29–40 with high education is only 1.23% (any disability) and 0.28% (work-preventing) over a 3-year period. At ages 53–64 it is 4.23% and 1.61%. A more-anticipated disability — already reflected in the couple's expected marital value — produces a smaller shock to realized marital value relative to expectations and thus a smaller divorce-triggering deviation. This formalizes why disability's effect on divorce is not just a function of earnings loss but of how surprising that loss is.

Data and Design

Singleton (2009) uses pooled Survey of Income and Program Participation (SIPP) panels (1990–1996) with retrospective marriage and disability histories from wave 2 topical module, merged to Social Security Administration (SSA) Detailed Earnings Records (DER) for longitudinal earnings. The divorce hazard is measured over a 36-month window centered on disability onset. The design is observational — no instrument for disability onset is employed — so estimates capture association with marital quality controls (first-marriage indicator, age at marriage, marriage tenure) but not clean causal identification.

Why It Matters

Formal Insurance as Substitute for Informal Insurance

If the disability-divorce relationship reflects genuinely causal earnings loss, more generous formal disability insurance (DI) could stabilize marriages by reducing the earnings shock to marital value. However, DI income may simultaneously raise the outside option of the disabled partner, making the alternative to marriage more attractive. The net effect of DI generosity on divorce rates among the disabled is empirically unknown and flagged by Singleton (2009) as a priority for future research.

DI Beneficiaries' Household Structure

Because DI beneficiaries are disproportionately non-college, older, and male, the disability-divorce gradient has an ironic implication: the youngest, highest-earning beneficiaries — those most economically penalized by their disability — are also those whose marriages are most at risk. Dissolution of the marital household has downstream consequences for spousal Social Security benefit eligibility, household poverty risk, and access to informal caregiving. See DI Beneficiary Mortality and Added Worker Effect for the health and labor supply consequences for the DI-proximate household.

Connection to the Match Quality Framework

The Singleton (2009) framework is an application of the Becker-Landes-Michael marital instability model that also underlies Match Quality and Marital Dissolution (Weiss and Willis 1997). The key distinction is that Singleton studies heterogeneous effects of a single type of shock (disability), while Weiss and Willis study comparative effects across types of shocks (earnings trajectory revisions). Singleton's reconciliation of the Charles-Stephens null result resolves an apparent contradiction: the aggregate null arises from composition (disability concentrated among older workers), not from disability being inherently uninformative about spousal quality.

Open Questions

Is the disability-divorce effect causal? Reverse causality (marital conflict worsening health) and common-cause confounding (poor marital quality correlated with health risk) are not fully ruled out.
Does more generous formal DI reduce or increase divorce rates among the disabled? The earnings-replacement and outside-option mechanisms push in opposite directions.
How has the gradient changed post-2009, with opioid-crisis disability concentrated among younger workers at exactly the age-education cells where the divorce effect is largest?
Do the gendered patterns reflect different earnings gradients, different social norms about divorcing a disabled spouse, or different relevance of work-limitation measures for women's labor market attachment?
What explains the Chronic-Severe divorce null for wives? Is it guilt, high divorce compensation costs, or marriage-specific capital (Meyer and Mok 2014)? Identifying the mechanism has policy implications for whether DI stabilizes marriages by reducing the earnings shock or has no effect on the Chronic-Severe pathway.