The Incidence of Mandated Maternity Benefits
Causal Question / Estimand
Who bears the cost of a group-specific benefit mandate? The estimand is the incidence of state/federal laws requiring comprehensive maternity coverage — specifically, the fraction of the mandate’s cost shifted to the wages of the targeted group (women of childbearing age), plus the effect on their net labor input (hours, employment).
Identification Strategy
The paper introduces the Triple-Difference (DDD) estimator. Because a mandate raising the cost of insuring women of childbearing age varies along three margins, Gruber differences across all three: the affected demographic group (married women 20–40, and their husbands) versus an unaffected within-state control group (older workers, single men); treatment states that passed maternity mandates (23 states, 1975–1979) versus comparison states that did not; and before versus after passage. The DDD nets out any national shock to the affected group’s wages and any state-specific shock common to all groups. Individual-specific cost of the mandate (from age-specific maternity cost × probability and type of insurance coverage) sharpens identification. The 1978 federal Pregnancy Discrimination Act, which forced the remaining states to adopt, provides a “reverse experiment” that flips treatment and control for a confirmatory check.
Key Assumptions
- Triple-Difference — the DDD contrast is the identification device.
- Parallel-Trends — in the relaxed DDD form: the wage-trend differential between the affected and unaffected groups would have evolved identically in treatment and comparison states absent the mandate (a single parallel-trends assumption, per OldenMoen2022-TripleDifference).
- SUTVA — one worker’s outcome is unaffected by others’ mandate exposure.
Threats to Validity
Differential group-specific shocks that vary across treatment and comparison states (e.g. state-level demand shifts hitting women of childbearing age differently); selection of which states passed mandates; composition changes in employment. The within-state control group and the reverse-experiment check are designed to blunt these.
Setting / Data
Current Population Survey (CPS), 1974–1982; wages, hours, and employment of married women of childbearing age and their husbands; state maternity-mandate laws passed 1975–1979 and the 1978 federal Pregnancy Discrimination Act.
Key Claims
- Costs of the maternity mandates were shifted to the targeted group’s wages at roughly 100 percent (nearly full shifting to group-specific wages).
- Correspondingly, there was little effect on net labor input (hours/employment) for the targeted group.
- Mandated benefits can be an efficient financing tool precisely when the affected group values the benefit and bears its cost through wages.
Connections
- Introduces the Triple-Difference design later formalized in OldenMoen2022-TripleDifference.
- Extends the classic two-group DiD lineage of AshenfelterCard1985-LongitudinalEarnings and Meyer1995-NaturalAndQuasiExperiments with a third contrast.
- See also DiD.
Citation
Gruber, J. (1994). The Incidence of Mandated Maternity Benefits. American Economic Review, 84(3), 622–641.