Mechanisms & Channels
Time-use evidence, wage expectations, selection effects, and NLSY cohort analysis exploring channels through which the gender gap operates.
ATUS Time-Use Gaps
Daily MinutesDaily Time-Use Gap (Female − Male, minutes)
Positive = women do more; negative = women do less
Source: American Time Use Survey, pooled recent waves.
| Activity | Gap (min) |
|---|---|
| Paid work | −67.96 |
| Housework | +31.91 |
| Childcare | +11.37 |
| Commute | −8.07 |
Women spend 68 fewer minutes on paid work per day but 43 more minutes on unpaid household and childcare work. The net paid + unpaid gap is roughly −25 minutes/day.
SCE Wage Expectations
Latest Wave (2025)Expected vs. Reservation Wages by Gender (2025)
Grouped bars — Survey of Consumer Expectations
| Measure | Female ($/hr) | Male ($/hr) | Gap ($) |
|---|---|---|---|
| Expected Offer Wage | 55.91 | 73.37 | 17.46 |
| Reservation Wage | 69.78 | 91.12 | 21.34 |
Men expect $17 more per hour in job offers and set reservation wages $21 higher. This expectations gap persists across all available survey waves.
Subgroup context (latest public SCE wave)
| Subgroup spread | Expected-offer gap ($) | Reservation-wage gap ($) |
|---|---|---|
| Gender (M − F) | 17.46 | 21.34 |
| Education (college+ − less than college) | 40.23 | 36.94 |
| Income (>$60K − ≤$60K) | 42.38 | 52.57 |
Men are above women in every public wave for both SCE measures, but the education and income gradients are larger than the gender gradient. This makes SCE useful for bargaining / outside-option interpretation without implying gender is the only gradient in the data.
Employment Selection
ACS Mean 2015–2023| Measure | Gap % |
|---|---|
| Combined expected annual earnings | 38.48 |
| Observed total annual earnings | 36.55 |
| Observed worker hourly wages | 18.83 |
| IPW-reweighted worker hourly | 19.73 |
The annual earnings gap (37–38%) is roughly double the worker hourly gap (19%), reflecting both lower hours and lower participation among women. IPW reweighting for employment selection adds approximately 1 pp to the hourly gap.
NLSY Cohort Deep Dive
NLSY79 · NLSY97Raw vs. Adjusted Gap by NLSY Cohort
Controls differ by cohort — see block-level table below
| Cohort | Raw Gap % | Adjusted Gap % | Reduction (pp) |
|---|---|---|---|
| NLSY79 (2000 earnings) | 44.43 | 34.10 | −10.33 |
| NLSY97 (2019 earnings) | 34.19 | 31.89 | −2.30 |
The raw gap narrowed from 44% (NLSY79, year 2000) to 34% (NLSY97, year 2019). Measured controls reduce the gap by 10 pp in NLSY79 but only 2 pp in NLSY97, and the blocks that matter most differ across cohorts.
What reduces the gap most?
| Cohort | Biggest measured reductions | Final gap |
|---|---|---|
| NLSY79 | Occupation sorting 7.02 pp; skills/noncognitive traits 5.62 pp; family background 0.04 pp | 34.10% |
| NLSY97 | Adult resources 6.60 pp; skills/school achievement 1.65 pp; occupation sorting 0.00 pp; family background −0.06 pp | 31.89% |
The cohort story changes. In NLSY79, occupation and skills matter most. In NLSY97, the biggest measured block is adult resources, which is post-market and should be treated as mechanism-sensitive accounting, not a clean pre-market explanation.