The marriage market refers to the process by which potential partners sort into couples, governed by the economic and social characteristics each brings. In the Becker (1973) household specialization model, the gains to marriage arise from spousal earnings differences that enable household division of labor; when male relative earnings fall, these gains shrink and marriage rates decline. The Wilson (1987, 1996) hypothesis extends this: the absolute economic decline of men — through unemployment and idleness — renders them less desirable partners regardless of gender earnings differentials, eroding family formation even when women's own earnings are unchanged.
Becker (1973) models marriage as a gains-from-trade arrangement: if men and women have different comparative advantages (historically men in market work, women in home production), coupling generates surplus beyond what each could achieve alone. When male market earnings fall relative to female, this comparative advantage shrinks, reducing the surplus from household specialization: marriage becomes less attractive. The corollary: when female earnings fall (e.g., female-industry trade shock), marriage becomes more attractive, because the comparative advantage gap widens.
ADH 2019 confirms both sides: male-industry shocks reduce marriage by pp per unit shock; female-industry shocks raise marriage by pp. The symmetry validates the specialization mechanism rather than a simple income effect.
Wilson (1987, 1996) argues that the absolute disappearance of economically secure men matters: even holding gender earnings differentials constant, men who are idle, substance-involved, or incarcerated become less marriageable. The mechanisms are social, not just financial — community and institutional ties erode, role expectations collapse, and a new equilibrium of non-marriage becomes normative.
ADH 2019 provides direct evidence: trade shocks increase male idleness ( pp, ages –), reduce the male share of young adults in the Commuting Zone (CZ) ( pp — reflecting mortality, incarceration, or out-migration), and raise male mortality from drug and alcohol (D&A) poisoning, human immunodeficiency virus/acquired immunodeficiency syndrome (HIV/AIDS), and homicide. All three effects shrink the pool of marriageable men beyond the earnings channel alone.
Kubrin, Wadsworth, and DiPietro (2006), reviewed in Wray et al. (2011), applied Wilson's framework specifically to Black male suicide in the late 20th century. Deindustrialization removed manufacturing employment from urban cores, concentrated poverty, weakened social institutions, and reduced Black men's "marriageability" and social role — a form of structural anomie that elevated suicide rates. This is the theoretical predecessor to the causal identification ADH (2019) later provided. See Sociology of Suicide.
The most important empirical finding for children's welfare: fertility responds less elastically than marriage to male-industry shocks. A unit male-industry shock:
Because unmarried-couple and single-parent households have poverty rates of and respectively (vs. in married households, 1990), this asymmetry drives a pp increase in the share of children living in poverty per unit male-industry shock.
ADH 2019 uses commuting zones covering the U.S. mainland, 1990–2014. The results hold for the full population and specifically for non-Hispanic whites, establishing that the Wilson-type dynamic is not limited to urban African Americans but reflects a general feature of communities experiencing blue-collar employment decline.
Angrist (2001) provides one of the cleanest historical causal tests of the Becker marriage market model using a completely different identification strategy. He exploits the U.S. Immigration Acts of 1921 and 1924 — which imposed national-origin quotas on immigration for the first time, shifting immigrant sex ratios by favoring family reunification over male labor migration — as an instrument for the sex ratio of the foreign-born population. Using Integrated Public Use Microdata Series (IPUMS) data from the 1910, 1920, and 1940 censuses across year-ethnicity cells ( ethnic groups), he estimates the causal effect of sex ratios on second-generation marriage and labor market outcomes.
Key quantitative findings:
The high endogamy rates of early immigrant communities — over within-group marriage for Italians and Central/Eastern European Jews — validate the marriage-market channel: effects are stronger where co-ethnic marriage is more prevalent, confirming that sex-ratio shocks propagate through ethnic group marriage networks.
This paper is the historical IV counterpart to ADH 2019: both confirm Becker's specialization mechanism — more men relative to women raises female marriage rates and reduces female LFP — but using entirely different sources of variation (immigration legislation vs. trade shocks) and different time periods (1910–1940 vs. 1990–2014). The convergence across settings strengthens the causal interpretation.
Stevenson and Wolfers (2007) provide the long-run empirical backdrop that contextualizes the ADH 2019 results: the decline in male marriageability from manufacturing job loss is one identified mechanism within a 150-year structural shift in the economics of marriage.
Divorce peaked in 1981, not at a secular high. The divorce rate (per 1,000 married couples) peaked at 22.8 in 1979 and fell to 16.7 by 2005 — the lowest since the early 1970s. Long-run view: the 1950s–1960s were an anomalously low-divorce period; the 1970s overshot a gradually rising trend dating from 1860; post-1981 decline is reversion to the pre-WWII trend. The cohort-specific claim that "half of all marriages end in divorce" applies to the 1970s marriage cohort (48% dissolution within 25 years); 1980s and 1990s cohorts show lower and improving dissolution rates.
Marriage rates reflect delayed timing and rising cohabitation. Marriage rates are at recorded-history lows, but median age at first marriage rose from ~23/22 (men/women, mid-1950s) to ~27/26 (2004), and 59% of early-2000s first marriages were preceded by cohabitation. The marriage rate decline overstates the retreat from partnership.
The Black–White gap is an entry gap, not a divorce gap. Among those who marry, divorce rates are similar by race. The gap is almost entirely explained by lower marriage entry among Black Americans, driven by labor market conditions and incarceration affecting male "marriageability" — the same Wilson mechanism documented by ADH 2019 for manufacturing workers.
The education divide runs in opposite directions for marriage and divorce. College-educated women's marriage rates have risen dramatically: 29% never-married (1960 census, women in their 60s) → 8% (2000 census). College graduates also have ~10 pp lower first-marriage dissolution than non-college graduates. Both facts are consistent with a consumption-complementarity model of modern marriage (see below).
Four driving forces:
From production to consumption complementarities. Becker's (1981) Treatise on the Family grounds marriage in production complementarities (household specialization, child-rearing). These are declining: 75% of married people lived with own children in 1880 vs. only 41% in 2005; rising longevity extends marriage far past childrearing; female LFP and household technology eliminate specialization gains. Stevenson and Wolfers argue that the theoretical foundation must shift toward consumption complementarities — joint leisure, shared public goods, insurance against health and income shocks. The empirical signature is rising positive assortative mating by education and income (consistent with consumption complementarity, where similar partners make better joint consumers; inconsistent with Becker's prediction of negative assortative mating based on specialization).
The paper provides the first clean causal test of Becker's household specialization model at scale, using exogenous geographic variation in trade exposure. It establishes that labor demand shocks — not just cultural shifts or women's economic independence — causally drive marriage rate changes.
Male idleness and premature mortality are intermediate outcomes on the path from manufacturing decline to marriage-market contraction. The same China shock that reduces the marriage-market value of men (via earnings and employment) also raises their mortality from D&A poisoning and violence. See Deaths of Despair for the causal mortality evidence.
The conditional Disability Insurance (DI) applicant mechanism (Autor and Duggan 2003; Lindner et al. 2017) identifies displaced low-skill manufacturing workers as the core population that flows into DI applications when employment disappears. ADH 2019 shows these same workers also exit marriage markets, become idle, and die prematurely — DI application is one of several competing behavioral responses to the same underlying labor demand shock. See Conditional DI Applicants.
The fraction of single-parent households is the single strongest correlate of low intergenerational income mobility across U.S. commuting zones (; Chetty et al. 2014c). This creates a causal pathway from manufacturing decline to intergenerational opportunity: ADH 2019 shows trade shocks cause non-marriage; Chetty et al. 2014c shows non-marriage predicts low child mobility. The community-level family structure effect (children of married parents still have lower mobility in high-single-parent CZs) implies the mechanism runs partly through peer networks and local institutions, not just individual household composition. See Intergenerational Income Mobility.
The ADH 2019 results explain a portion of the nonmarital fertility trend causally — specifically, the portion attributable to gender-specific trade shocks in specific commuting zones. Wu (2008) documents the broader 50-year backdrop: the share of women with at least one nonmarital birth by age 30 rose nearly monotonically from (1925–1929 cohort) to (1965–1969 cohort) for all U.S. women, and from to for white women. The manufacturing trade shock is one identified mechanism within a longer structural trend. See Nonmarital Fertility.
Scaling to observed China penetration 1990–2014:
Single-parent family structure is a component of the socioeconomic status (SES) disadvantage index in Autor, Figlio, Karbownik, Roth, and Wasserman (2019). The paper shows that boys in single-parent and low-SES families face disproportionately worse behavioral and educational outcomes than their sisters in the same households — a gradient confirmed to operate postnatally. This creates a downstream link from marriage market contraction (ADH 2019) to children's outcomes: ADH 2019 shows trade shocks reduce marriage rates and raise the share of children in single-parent households; Autor et al. (2019) shows that single-parent households produce the largest boy-girl gap in school suspension, absences, and high school (HS) graduation. The two papers together imply a causal chain: manufacturing decline male idleness non-marriage single-parent households disproportionate harm to boys intergenerational disadvantage. See Gender Gap in Educational Outcomes.
Within fragile families, the Black-white gap in post-birth marriage rates is largely explained by the local employability ratio (employed men / all women in a city). Black mothers marry at nearly the same rates as white mothers when this ratio is held constant (Harknett & McLanahan 2004). Because Black men face higher rates of unemployment and incarceration, the effective pool of marriageable men is smaller in predominantly Black urban areas. This connects the structural Wilson mechanism to the within-city variation in marriage rates among unmarried parents: it is not differential attitudes toward marriage (Black unmarried mothers hold equally pro-marriage attitudes) but differential access to marriageable men that drives the gap. See Multi-Partnered Fertility for the compounding role of multi-partnered fertility (MPF) in further reducing effective male investment per child.