Prevalence of Long Work Hours by Spouse’s Degree Field and the Labor Market Outcomes of Skilled Women

Abstract

Using 2009 to 2015 American Community Survey (ACS) data, this article estimates the effect of the prevalence of long hours and short hours of work in a husband’s field of work, as defined by his undergraduate degree field, on the labor market outcomes of skilled married women. When individuals work in fields that require longer hours of work, their spouses experience spillover effects. The labor market outcomes of female spouses are more negatively affected than are those of male spouses. Specifically, female spouses face lower total earnings, hourly wages, employment options, and hours of work for married women with children relative to married men with children or married women without children. Little evidence supports the idea that the rate of short hours of work in a spouse’s degree field differentially affects married women with children.

Keywords

labor market outcomes household economics marriage and labor supply work and family long hours

Prior literature has documented the lower labor force participation and lower earnings of college-educated women relative to college-educated men, and that these gender gaps tend to emerge and widen particularly after childbirth (Bertrand, Goldin, and Katz 2010). One potential mechanism is that women with children may select into lower-paying jobs in search of more flexible, family-friendly work conditions. Cortés and Pan (2016), for example, found that college-educated women, particularly those with children, avoid occupations that demand long hours.

This article considers whether the labor market outcomes of college-educated women are affected by whether their husband trained for a profession that demands long hours of work. Prior research documents that women remain responsible for the larger share of household responsibilities, even among dual-earner couples (Bianchi, Milkie, Sayer, and Robinson 2000; Stone 2007). Married women, especially those with children, may find work–family conflict exacerbated and their own labor market effort particularly costly when their husband works in a profession that demands long hours. Couples in which the husband works in such a profession may engage in greater household specialization, with the wife allocating relatively less effort to market work and shifting to jobs that are lower paying, lower effort, and more flexible.

This article evaluates the effects of the prevalence of long hours and short hours of work in a husband’s undergraduate degree field as the measure of interest, rather than the husband’s own hours of work. While a husband’s usual hours of work per week and a husband’s occupation both may respond to a wife’s characteristics, undergraduate degree field is time-constant and in most cases determined prior to spouse characteristics.

American Community Survey (ACS) data from 2009 to 2015 are used to calculate degree-field-specific measures of the prevalence of long hours and short hours of weekly work, and to estimate the effects of long hours and short hours in a spouse’s degree field in an analysis sample of college-educated women and men ages 25 to 44 who are married to college-educated spouses. More specifically, coefficient estimates for long hours and short hours in a spouse’s degree field for married women with children are compared to estimates for two comparison groups: married men with children and married women without children.

Literature Review

Prior research has studied the selection of women into lower-paying occupations and into lower-paying jobs within occupation, possibly in search of more flexible, family-friendly work conditions (Flabbi and Moro 2012; Pertold-Gebicka, Pertold, and Gupta 2016; Wiswall and Zafar 2018). The demand for long hours in skilled jobs appears to be an important feature driving gender gaps in labor market outcomes. Goldin (2014) calculated the gender wage gap and the return to long hours of work by occupation using 2009 to 2011 American Community Survey (ACS) data. She found that the gender gap is largest in occupations with the highest returns to long hours of work. Cha and Weeden (2014) evaluated the contribution of demand for long hours of work to the gender wage gap using Current Population Survey (CPS) data from 1979 and 2007. They found that increasing returns to long hours, combined with higher rates of long hours of work for male compared to female workers, acted to reduce gender wage convergence over this time period.

Cortés and Pan (2016) found that college-educated women with children avoid occupations that demand long hours and are more likely to drop out of the labor market if they trained in fields that demand long hours. Using 215 occupations and four decades of Decennial Census and ACS data, they found that the share of men in an occupation who report working 50 or more hours per week is negatively associated with the participation of college-educated married women with children in that occupation. In individual-level cross-sectional analysis using 2009 to 2014 ACS data, they found that college-educated married mothers who majored in undergraduate degree fields associated with longer hours of work are less likely to be currently employed.

Blau and Kahn (2017) estimated trends in the gender wage gap for 1980 to 2010 in the Panel Study of Income Dynamics (PSID). They found that by 2010 the gender wage gap was much wider at the top of the wage distribution than at the bottom or middle. Their discussion of potential explanations includes the role of long hours of work, based on the Goldin (2014) findings. Because occupations that have higher returns to long hours of work tend to also be higher-paying occupations, the gender gap in participation in occupations that demand long hours likely differentially affects the gender wage gap at the top of the wage distribution. Neither Blau and Kahn (2017) nor Goldin (2014) considered the additional role of long hours in a spouse’s occupation. If women’s labor market outcomes are negatively affected by demand for long hours in a spouse’s occupation, and if high-skilled women who might otherwise be at the top of the wage distribution disproportionately marry men who trained for demanding careers, then this could also differentially widen the gender gap at the top of the wage distribution.

Using the 1996 panel of the Survey of Income and Program Participation (SIPP), Cha (2010) found that having a husband who works 50 or more hours a week increases the probability that a married woman with children will quit her job, but having a wife who works 50 or more hours does not affect the probability that a husband will quit his job. Cha could not rule out the possibility that men choose longer work hours in anticipation of their wife’s lower labor market attachment. This article uses instead the rate of long hours of work in the degree field, rather than actual hours of work, to avoid the possibility that individuals endogenously choose work hours or occupation in response to a spouse’s characteristics.

Methods

Analysis for this article is conducted using the 2009 to 2015 waves of the American Community Survey (ACS).¹ The ACS is an annual cross-sectional survey of a 1% nationally representative sample of US residents conducted by the US Census Bureau. The empirical analysis makes use of two samples from 2009 to 2015 ACS data. First, the sample of college-educated male workers ages 25 to 55 is used to calculate degree-field-level characteristics. Regressions are then estimated using an analysis sample of college-educated women and men ages 25 to 44 who are married to college-educated spouses.

To focus the analysis on individuals with some labor market attachment, the analysis sample is further restricted to men and women who have worked in the past five years. Because the ACS data report occupation of most recent job in the past five years, individuals who do not report an occupation are excluded from the analysis sample.²

Degree Field Characteristics

Starting in 2009, the ACS data report the field of undergraduate degree for college-educated individuals using 181 detailed codes. Although the data report completion of an advanced degree, the field of the advanced degree is not reported. Following the literature, the proportion of prime-aged (ages 25–55), college-educated male workers in a degree field who report 50 or more usual hours of work per week is used as the measure of long hours by degree field. Using the same sample, the proportion of male workers in a degree field who report 34 or fewer usual hours of work per week is used as the measure of short hours by degree field. Whether estimating the effect of a husband’s degree field characteristics on a wife’s outcomes or the effect of a wife’s degree field characteristics on a husband’s outcomes, measures of long hours and short hours by degree field are calculated using prime-aged men.

The long hours and short hours measures are calculated separately by advanced degree status within degree field. Using the 181 detailed degree field codes allows for 362 degree field×advanced degree status cells. Additional degree field×advanced degree status characteristics, also calculated using the same sample of college-educated prime-aged men, include the average hourly wage and wage variance.³ Because McKinnish (2008) documented that the earnings of married women are negatively affected by high migration rates in a husband’s occupation, the one-year cross-state migration rate is also calculated using the same sample. For simplicity, the text will refer to these degree field × advanced degree status characteristics as degree field characteristics. Degree field cells that contain fewer than 100 observations for calculating these characteristics are dropped from the analysis. This sample restriction eliminates 29 degree field × advanced degree status cells (leaving 333), but only eliminates 0.10% of the analysis sample.

Table 1 reports descriptive statistics using the sample of college-educated women ages 25 to 44 married to college-educated husbands. The first row of Table 1 reports distributional characteristics of long hours in degree field for the women in this sample. The median woman in the sample specialized in an undergraduate degree field in which 33.5% of prime-aged college-educated male workers report 50 or more usual hours of work per week. The second row of Table 1 reports distributional characteristics of long hours in degree field for the husbands in this sample. The median husband in the sample specialized in a degree field in which 34.5% of prime-aged, college-educated male workers report 50 or more usual hours per week.

Table 1.

Characteristics of Wife’s and Husband’s Degree Field and Occupation

	10th percentile	25th percentile	50th percentile	75th percentile	90th percentile
A. Degree field characteristics
Fraction male workers with ≥ 50 work hours in:
Wife’s degree field	0.239	0.275	0.335	0.377	0.446
Husband’s degree field	0.253	0.285	0.345	0.388	0.461
Fraction male workers with ≤ 34 work hours in:
Wife’s degree field	0.0369	0.0473	0.0697	0.0863	0.105
Husband’s degree field	0.0331	0.0399	0.0486	0.0728	0.0970
Average annual earnings for male workers in:
Wife’s degree field	53,089	63,834	76,350	95,327	120,680
Husband’s degree field	59,886	68,751	86,310	104,399	125,237
B. Occupation characteristics
Fraction male workers with ≥ 50 work hours in:
Wife’s occupation	0.125	0.185	0.274	0.366	0.491
Husband’s occupation	0.179	0.232	0.340	0.445	0.560
Fraction male workers with ≤ 34 work hours in:
Wife’s occupation	0.0163	0.0273	0.0528	0.0930	0.170
Husband’s occupation	0.0125	0.0184	0.0288	0.0609	0.142
Average annual earnings for male workers in:
Wife’s occupation	41,923	51,898	66,466	93,506	120,239
Husband’s occupation	45,794	61,935	85,759	107,594	144,197

Notes: Sample of college-educated women in 2009–2015 ACS data ages 25–44 married to college-educated men (N = 346,897). Fraction of male workers with ≥ 50 weekly work hours, fraction of male workers with ≤ 34 weekly work hours, and average annual earnings for male workers are calculated for each degree field and occupation using the sample of college-educated male workers ages 25–55.

The next two rows of Table 1 report distributional characteristics of the short hours measure. The median woman in the sample specialized in a degree field for which 6.97% of prime-aged, college-educated male workers report 34 or fewer usual hours of work per week. For the median husband, the analogous statistic is 4.86%. The final two rows of panel A in Table 1 report distributional characteristics of average male earnings in degree field for the same sample.

Panel B in Table 1 reports, for the same sample, distributional characteristics of long hours, short hours, and average male earnings calculated by detailed occupation rather than degree field. Comparing panels A and B, it is noteworthy that the gender gaps in long hours, short hours, and average male earnings are much larger for occupation than for degree field. Although women do typically specialize in degree fields in which long hours are less prevalent, short hours are more prevalent, and average male earnings are lower than degree fields typically chosen by men, the gender gap in these degree field characteristics is much smaller than the gender gap in the same occupation characteristics. These descriptive statistics suggest that women do not particularly select out of degree fields associated with long hours, but do select out of occupations associated with long hours.

A comparison of panels A and B of Table 1 highlights the benefits of using degree-field-level measures of long hours and short hours as opposed to measures based on occupation. Evidence supports sorting by women away from higher-paying occupations and away from occupations with longer hours. This sorting is likely affected by a husband’s characteristics. Therefore, in a regression of a husband’s labor market outcomes on long hours in a wife’s occupation, a negative relationship could indicate that women are less likely to sort away from demanding, high-paying occupations if they have a low-earning husband, rather than a direct causal effect of a wife’s occupation on a husband’s outcomes. Similarly, a negative effect of long hours in a husband’s occupation on a wife’s labor market outcomes could reflect husbands shifting into more demanding careers in response to wives who have low earnings or low attachment to the labor market. Although current occupation is likely endogenous to a spouse’s characteristics, field of degree is time-constant and, in most cases, determined prior to a spouse’s characteristics.

Positive Assortative Matching and Degree Field Characteristics

Table 2 provides evidence that degree fields associated with greater demand for long hours also tend to be higher skilled and higher wage. The top panel of Table 2 separates out individuals in the analysis sample with degree fields for which the long hours measure is less than or equal to 0.28 and individuals with degree fields for which the long hours measure is 0.38 or greater (approximately the 25th and 75th percentiles reported in Table 1). The table then reports, for each group, the means of average male wage in degree field and proportion of male workers in degree field with an advanced degree. Both average male wages in degree field and proportion with advanced degrees are notably higher for the degree fields with a higher prevalence of long hours. The bottom panel instead separates the sample based on the short hours measure. The reported means indicate that degree fields with a lower prevalence of short hours tend to be higher wage and higher skill. Overall, the descriptive statistics in Table 2 indicate that individuals from degree fields with a higher prevalence of long hours and a lower prevalence of short hours will tend to be positively selected relative to individuals from degree fields with a lower prevalence of long hours and a higher prevalence of short hours.

Table 2.

Characteristics of Degree Fields with High and Low Rates of Long Hours and Short Hours

	(≥ 50 hours in degree field) ≤ 0.28	(≥ 50 hours in degree field) ≥ 0.38
Average male wage in degree field	33.43 (7.00)	49.65 (10.32)
Proportion male workers with advance degree in degree field	0.077 (0.267)	0.795 (0.404)
	(≤ 34 hours in degree field) ≤ 0.04	(≤ 34 hours in degree field) ≥ 0.08
Average male wage in degree field	45.57(8.77)	32.48(5.41)
Proportion male workers with advanced degree	0.593(0.491)	0.192(0.394)

Notes: Sample of college-educated men and women in 2009–2015 ACS data ages 25–44 married to college-educated spouses. Table reports average degree field characteristics for designated subcategories. Standard deviations in parentheses.

Positive assortative matching is well documented in the marriage market (Mare 1991; Kalmijn 1998; Blossfeld and Timm 2003; Schwartz and Mare 2005; Blossfeld 2009). To the extent that there is positive matching on degree field characteristics implies that individuals married to spouses from degree fields with higher rates of long hours will tend to be positively selected and that individuals married to spouses from degree fields with higher rates of short hours will tend to be negatively selected. Table 3 provides evidence on matching on degree field characteristics by regressing a husband’s degree field characteristics on a wife’s. The first column reports results from simple bivariate regressions, whereas the second adds controls for advanced degree status, age, race/ethnicity, immigration/citizenship, age of marriage, state × urban fixed effects, and year fixed effects.

Table 3.

Relationship between Wife’s and Husband’s Degree Field Characteristics

	No controls	With controls
A. Full sample (n = 346,897)
Log(Average male wage in degree field)	0.268(0.012)***	0.178(0.011)***
≥ 50 Hours in degree field	0.173(0.019)***	0.137(0.018)***
≤ 34 Hours in degree field	0.225(0.017)***	0.217(0.016)***
B. Exclude same-degree field couples (n = 311,055)
Log(Average male wage in degree field)	0.202(0.010)***	0.095(0.009)***
≥ 50 Hours in degree field	0.100(0.017)***	0.044(0.014)**
≤ 34 Hours in degree field	0.132(0.014)***	0.116(0.012)***

Notes: Sample of college-educated men and women in 2009–2015 ACS data ages 25–44 married to college-educated spouses. Column (1) reports coefficient from simple bivariate regression of husband’s degree field characteristic on wife’s. Column (2) adds controls for both husband and wife: advanced degree indicator, age and age squared (interacted with advanced degree), race/ethnicity indicators, immigration and citizenship status as well as wife’s age of marriage (quadratic), state fixed effects, state×urban indicator fixed effects, and survey year fixed effects. Standard errors, clustered on spouse’s degree field, are reported in parentheses.

The results for the full sample in panel A of Table 3 confirm that wives from higher paying degree fields tend to have husbands from higher paying degree fields, wives from degree fields with higher rates of long hours tend to have husbands from degree fields with higher rates of long hours, and wives from degree fields with higher rates of short hours tend to have husbands from degree fields with higher rates of short hours. Panel B of Table 3 confirms that these associations persist even after eliminating from the sample the couples in which both spouses specialized in the same degree field.

Taken together, the analyses in Tables 2 and 3 suggest that regressions of labor market outcomes on a spouse’s degree field characteristics will be biased due to positive assortative matching. Men from degree fields with higher rates of long hours will tend to be married to higher-skilled, higher-wage women, which will positively bias estimates of the effect of long hours in a husband’s degree field on a wife’s labor market outcomes. Similarly, men from degree fields with higher rates of short hours will tend to be married to lower-skilled, lower-wage women, which will negatively bias estimates of the effect of short hours in a husband’s degree field on a wife’s labor market outcomes.

Of course, assortative matching will bias both estimates of the effect of a husband’s degree field characteristics on a wife’s outcomes and estimates of the effects of a wife’s degree field characteristics on a husband’s outcomes. It is therefore instructive to generate and compare estimates for both samples. If men married to women from degree fields with higher rates of long hours tend to have better labor market outcomes, this finding is much more reasonably interpreted as a positive bias due to assortative matching rather than an actual causal positive effect of a wife’s long hours on a husband’s labor market performance. Furthermore, this implies that estimates of the effect of long hours in a husband’s degree field on a wife’s labor market outcomes are also subject to positive bias due to assortative matching. If, by contrast, coefficient estimates indicate that women married to men from degree fields with higher rates of long hours tend to have worse labor market outcomes, this is consistent with a negative direct effect of a husband’s long hours that is sufficiently strong to outweigh the positive bias due to assortative matching.

We therefore compare estimates for wives (which reflect the combination of the direct effect of a husband’s long hours on wives and the assortative matching effect for wives) to estimates for husbands (which reflect the combination of the direct effect of a wife’s long hours on husbands and the assortative matching effect for husbands). More specifically, in this article, estimates for married women with children, who are most likely to be directly affected by long hours and short hours in a husband’s degree field, are compared to estimates for married men with children as well as to estimates for married women without children.

If the estimates for married women with children are typically more negative than the estimates for the two comparison groups, this finding is consistent with married women with children experiencing a disproportionate negative effect of a husband’s long hours. To be clear, however, this comparison of estimates does not necessarily difference out the assortative matching effect and leave an unbiased estimate of the difference in direct effects. This would only be true if the assortative matching effects were identical across subgroups. It is therefore more appropriate to think of this empirical approach as checking for evidence consistent with a differential direct effect of a spouse’s long hours on a wife’s labor market outcomes, rather than producing an unbiased estimate of the direct effect.

Regression Specification

The regression specification for analyzing a wife’s labor market outcomes is:

\begin{matrix} W i f e_O u t c o m e_{i j t} = β_{0} + β_{1} H u s b_D e g F i e l d H r s G t 50_{j} + β_{2} H u s b_D e g F i e l d H r s L t 34_{j} \\ + H u s b_D e g F i e l d C o n t r o l s_{j} β_{3} + β_{4} H u s b_E a r n_{j} + X_{i j} β_{5} \\ + \sum_{k} δ_{k} * W i f e_D e g F i e l d_{k i} + \sum_{k} ϕ_{k} * W i f e_D e g F i e l d_{k i} * W i f e_A d v_{i} \\ + γ_{t} + ε_{i j t} \end{matrix}

(1)

where for wife i married to husband j in ACS survey year t, Wife_Outcome is an individual labor market outcome: earnings, hourly wage, employment status, or usual hours of work per week. Table 4 reports descriptive statistics for these outcome variables separately for married women with children, married men with children, and married women without children.⁴

Table 4.

Descriptive Statistics by Subsample

	Married women with children	Married men with children	Married women w/o children
Earnings	46,042(52,763)	87,905(78,382)	49,116(42,862)
Weekly hours	31.58(16.86)	44.92(11.06)	39.83(13.03)
Employed	0.813	0.967	0.911
N	237,345	208,781	109,552
Workers only
Earnings	54,984(53,226)	92,136(77,779)	52,589(42,242)
Wage	32.34(53.36)	40.68(42.30)	26.85(171.7)
Weekly hours	36.47(12.56)	45.58(9.565)	41.65(10.20)
N	198,746	199,195	102,317
≥ 50 Hours in degree field	0.327(0.0776)	0.345(0.0747)	0.330(0.0774)
≤ 34 Hours in degree field	0.0689(0.0285)	0.0588(0.0274)	0.0735(0.0318)
N	237,345	208,781	109,552

The explanatory variables of interest are the proportion of prime-aged college-educated male workers in a husband’s degree field reporting 50 or more hours of work per week and the proportion of prime-aged college-educated male workers in a husband’s degree field reporting 34 or fewer hours of work per week. The degree-field-level controls for a husband’s degree field are: average male wage, male wage variance, average wage for male workers working 50 or more hours a week, average male wage for workers working 34 or fewer hours a week, and out-of-state migration rate in past year. Average male wages in a husband’s degree field control for the fact that degree fields with higher rates of long hours also tend to be higher paying. The wage variance measure controls for differences across degree fields in dispersion around the mean.⁵ Standard errors are clustered on a spouse’s degree field.

It is important to include measures of long hours and short hours in a husband’s degree field together in the same regression. Excluding one of these variables exacerbates the bias due to assortative matching on the remaining variable. The long hours and short hours measures are negatively correlated, and assortative matching induces a negative correlation between the short hours measure and spouse labor market outcomes (individuals from short hours–prone degree fields tend to be negatively selected and their spouses, therefore, also tend to be negatively selected). If the short hours measure is not included as a control, this exacerbates the positive bias due to assortative matching on the long hours measure (and vice versa).

X is a vector of controls that includes, for both husband and wife, an indicator for advanced degree, age and age squared (also interacted with advanced degree), race/ethnicity indicators (white, black, Hispanic), immigration, and citizenship status. X also includes couple-level controls: number of children, number of children under age 5, wife’s age at time of marriage (quadratic), and state × urban indicator fixed effects. Fixed effects for a wife’s detailed degree field (interacted with advanced degree indicator) help control for heterogeneity in a wife’s human capital and potential earnings. Year fixed effects are also included in the model.

The specification also includes a control for a husband’s own earnings. This control is included because husbands from degree fields with higher rates of long hours tend to be higher earners, and higher husband earnings tend to reduce a wife’s labor market effort. A husband’s earnings are therefore included as a control to help isolate the effect of the time demands in a husband’s degree field separate from a husband’s earnings. Because of the concern that a husband’s earnings are endogenous to a spouse’s labor market outcomes, Table A.1 in the Online Appendix reports sensitivity of the baseline results in Tables 5 and 6 to excluding a husband’s earnings from the regression. The results are highly robust to the exclusion of this variable.

Table 5.

Earnings and Wage Effects of Long Hours and Short Hours in Spouse’s Degree Field

	(1)Married women with children	(2)Married men with children	(3)Married women w/o children	(4)Diff 1(1) vs (2)	(5)Diff 2(1) vs (3)
A. Log(Earnings)
≥ 50 Hours in spouse’s degree field	−1.221*(0.594)	0.207(0.246)	0.0662(0.317)	−1.428*(0.635)	−1.287*(0.520)
≤ 34 Hours in spouse’s degree field	−2.301(1.545)	−1.279*(0.560)	−1.916***(0.569)	−1.022(1.761)	−0.385(1.253)
N	237,345	208,781	109,552	446,126	346,897
B. Log(Earnings), workers only
≥ 50 Hours in spouse’s degree field	−0.352*(0.176)	0.173(0.117)	0.193(0.119)	−0.525**(0.180)	−0.545***(0.140)
≤ 34 Hours in spouse’s degree field	−0.624(0.474)	−0.685**(0.209)	−1.020***(0.268)	0.0610(0.491)	0.396(0.352)
N	198,746	199,195	102,317	397,941	301,063
C. Log(Wage), workers only
≥ 50 Hours in spouse’s degree field	−0.125(0.0795)	0.0578(0.0869)	0.116(0.0755)	−0.183⁺(0.0928)	−0.241**(0.0754)
≤ 34 Hours in spouse’s degree field	−0.199(0.193)	−0.569***(0.168)	−0.487*(0.189)	0.370(0.240)	0.288(0.203)
N	198,746	199,195	102,317	397,941	301,063

Notes: Column (1) uses sample of college-educated women in 2009–2015 ACS data ages 25–44 married to college-educated men and residing with a child under age 18. Column (2) uses sample of college-educated men in 2009–2015 ACS data ages 25–44 married to college-educated women and residing with a child under 18. Column (3) uses sample of college-educated women in 2009–2015 ACS data ages 25–44 married to college-educated men and who do not reside with a child under 18. Table reports, from Equation (1), estimates of the coefficients on fraction of male workers in spouse’s degree reporting 50 or more usual hours of work per week and fraction of male workers in spouse’s degree field reporting 34 or fewer usual hours of work per week. Differences estimates in columns (4) and (5) are generated using fully interacted models. Controls include all controls listed in notes of Table 4 as well as degree field fixed effects (interacted with advanced degree indicator), spouse’s earnings, and additional characteristics of spouse’s degree field: average hourly wage, average hourly wage for workers with 50 or more hours, average hourly wage for workers with 34 or fewer hours, wage variance, and cross-state migration rate. Standard errors, clustered on spouse’s degree field, in parentheses.

p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

Table 6.

Labor Supply Effects of Long Hours and Short Hours in Spouse’s Degree Field

	(1)Married women with children	(2)Married men with children	(3)Married women w/o children	(4)Diff 1(1) vs (2)	(5)Diff 2(1) vs (3)
A. Employed
≥ 50 Hours in spouse’s degree field	−0.135**(0.0491)	0.0271⁺(0.0146)	−0.0220(0.0370)	−0.162**(0.0523)	−0.113*(0.0499)
≤ 34 Hours in spouse’s degree field	−0.133(0.122)	−0.0368(0.0332)	−0.107(0.0651)	−0.0958(0.128)	−0.0258(0.108)
N	237,345	208,781	109,552	446,126	346,897
B. Weekly hours
≥ 50 Hours in spouse’s degree field	−3.395(2.612)	5.528***(1.334)	4.361*(1.689)	−8.923**(2.694)	−7.756***(2.109)
≤ 34 Hours in spouse’s degree field	−9.446(7.845)	0.290(2.302)	−7.589*(3.523)	−9.736(7.499)	−1.857(5.803)
N	237,345	208,781	109,552	446,126	346,897
C. Weekly hours, workers only
≥ 50 Hours in spouse’s degree field	−0.724(1.806)	4.374**(1.310)	3.655**(1.241)	−5.099*(2.056)	−4.380**(1.470)
≤ 34 Hours in spouse’s degree field	−6.521(5.312)	0.00538(2.267)	−7.911**(2.682)	−6.527(5.057)	1.390(4.043)
N	198,746	199,195	102,317	397,941	301,063

p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

Equation (1) is estimated separately for married women with children and married women without children. Equation (1) is also estimated for married men with children, where a husband’s labor market outcomes are the dependent variables, long hours and short hours in a wife’s degree field are the key explanatory variables, and fixed effects for a husband’s degree field replace fixed effects for a wife’s degree field.

Note that long hours and short hours in degree fields are certainly correlated with unobserved work characteristics. For example, jobs that demand longer hours may also tend to be more stressful or may require that the worker be accessible by phone or e-mail most hours of the day, which could also affect spouse labor market outcomes. The effects estimated in this article can therefore better be thought of as the effects of a bundle of work characteristics associated with demand for long hours or short hours of work (conditional on pay).

Results

Baseline Results

Table 5 reports estimates from Equation (1) using earnings and wages as outcomes. Standard errors are clustered on a spouse’s degree field. Panel A reports estimates for logged earnings, specifically log(earnings + 1), whereas panels B and C restrict the sample to current workers and report estimates for logged earnings and logged wages. The first three columns report estimates using the subsamples of married women with children, married men with children, and married women without children, respectively.

It is instructive to first look at columns (2) and (3), which report estimates for the comparison groups of married men with children and married women without children. Across all three panels, the estimates in these columns are consistent with positive assortative matching. All coefficient estimates for long hours in a spouse’s degree field are positive and all coefficient estimates for short hours in a spouse’s degree field are negative, though with varying degrees of significance. This outcome is consistent with assortative matching in which individuals from degree fields with high rates of long hours tend to be positively selected and therefore tend to be married to positively selected higher-earning spouses. Individuals from degree fields with low rates of long hours tend to be negatively selected and therefore tend to be married to negatively selected lower-earning spouses. The magnitudes are quite similar when comparing the long hours and short hours coefficient estimates across the two comparison groups in columns (2) and (3).

Column (1) reports estimates for the sample of married women with children, and the coefficient estimates for long hours in a spouse’s degree field differ substantially from those for the two comparison groups. For married women with children, coefficient estimates on long hours in a spouse’s degree field are negative in all three panels and statistically significant in panels A and B. Column (4) directly tests the difference in coefficient estimates between married women with children and married men with children using a fully interacted model. Column (5) similarly tests the difference in coefficient estimates between married women with children and married women without children. Across all three panels, differences in long hours coefficient estimates between married women with children and the comparison groups are negative and statistically significant.

Unlike coefficient estimates for the long hours measure, coefficient estimates for short hours in a spouse’s degree field are the same sign for married women with children as the two comparison groups, and none of the differences in estimates tested in columns (4) and (5) are statistically significant. It does not appear that short hours in a spouse’s degree field are differentially consequential for the earnings and wages of married women with children.

The negative effect of long hours in a spouse’s degree field on the earnings of married women with children is larger when estimated using the full sample in panel A than when using the sample of workers in panel B. This finding is consistent with a sizeable negative participation effect of long hours in a spouse’s degree field on married women with children, and suggests that mothers married to husbands from degree fields with higher rates of long hours are disproportionately likely to be non-workers. This possibility is tested more directly in Table 6.

Table 6 reports estimates from Equation (1) using labor supply outcomes: employment and weekly hours. The structure of Table 6 is identical to Table 5. The results for long hours in a spouse’s degree field mirror those in Table 5. The long hours coefficient estimates for the comparison groups are largely positive, whereas the long hours estimates for married women with children are all negative. The differences in long hours coefficient estimates between married women with children and the comparison groups, reported in columns (4) and (5), are all negative and statistically significant. The negative hours of work effects in columns (4) and (5) both decline by roughly 43% when non-workers are excluded from the sample, suggesting that slightly less than half of the total hours of work effect of long hours in a husband’s degree field is attributable to effects on the extensive margin.

Also similar to the Table 5 results, there are no statistically significant differences in short hours coefficient estimates between married women with children and the comparison groups. Unlike Table 5, however, the differences in short hours estimates between married women with children and the comparison groups are mostly negative, though statistically insignificant. The fact that the short hours estimates are even more negative for married women with children than the comparison groups suggests that the negative selection into marriage to spouses from short hours–prone degree fields is more consequential for the labor supply of married women with children than the two comparison groups. This is very plausible given that married women with children have lower employment and average hours than both comparison groups, and negatively selected mothers likely respond to childbearing with larger reductions in labor supply than positively selected mothers. Notice, however, that this argument implies that positively selected women (who tend to marry men from long hours–prone degree fields) are disproportionately less likely to reduce labor supply when they have children, making the negative coefficient estimates for long hours in a spouse’s degree field reported in Table 6 even more striking.

Tables A.2 and A.3 in the Online Appendix replicate the analyses in Tables 5 and 6, but vary the hours of work cutoffs used to define long hours and short hours of work. The patterns of results in Tables A.2 and A.3 are consistent with those in Tables 5 and 6, though magnitudes are sensitive to the definitions of long hours and short hours used to construct the degree-field-specific measures.

The differential negative effects of long hours in a spouse’s degree field on married women with children reported in Tables 5 and 6 are consistent with a negative direct effect of husbands’ long hours on married women with children. An alternative explanation for a negative relationship between long hours in a husband’s degree field and a wife’s labor market outcome is that women who anticipate having low participation or low earnings after having children are more likely to marry men who have trained for careers that demand long hours. This form of negative assortative matching is a plausible alternative explanation for the estimates in Tables 5 and 6, but note that this explanation is inconsistent with all of the available evidence on marital sorting based on observed characteristics. The analysis in Table 3 indicates positive marital sorting on long hours in degree field and average male wages in degree field. The coefficient estimates on long hours in a spouse’s degree field for married women without children reported in column (3) of Tables 5 and 6 indicate positive correlations between a wife’s hours and earnings and long hours in a husband’s degree field. The coefficient estimates on long hours in a spouse’s degree field for married men with children reported in column (2) of Tables 5 and 6 indicate positive correlations between a husband’s hours and earnings and long hours in a wife’s degree field. All of these results indicate that wives of men from degree fields with long hours tend to be positively selected on labor supply, earnings, and long hours in degree field, and that wives of high-earning men are more likely to have specialized in degree fields with long hours. If the results in columns (4) and (5) of Tables 5 and 6 are generated by negative sorting of low-earning women into marriages with men trained for careers that demand long hours, this explanation requires negative sorting on unobservables, conditional on positive sorting on observables, that is only revealed after childbearing.

Heterogeneity by Age of Youngest Child

Tables 7A and 7B replicate the analyses in Tables 5 and 6 for subsamples based on age of youngest child. To reduce the total volume of reported estimates and to facilitate comparisons across tables, Tables 7A and 7B report only differences in coefficient estimates between married women with children and the two comparison groups, such as those reported in columns (4) and (5) of Tables 5 and 6.

Table 7A.

Earnings Analysis by Age of Youngest Child

	Log(Earnings)		Log(Earnings), workers only		Log(Wage), workers only
	Diff 1	Diff 2	Diff 1	Diff 2	Diff 1	Diff 2
A. Youngest child < 5
≥ 50 Hours, spouse’s degree field	−2.092**(0.783)	−1.810**(0.674)	−0.488**(0.178)	−0.495***(0.144)	−0.155(0.0955)	−0.187*(0.0757)
≤ 34 Hours, spouse’s degree field	−2.259(2.165)	−1.461(1.651)	−0.321(0.502)	0.0246(0.367)	0.127(0.224)	0.133(0.178)
N	275,429	250,942	240,999	215,347	240,999	215,347
B. Youngest child ≥ 5
≥ 50 Hours, spouse’s degree field	−0.516	−0.621	−0.540*	−0.568***	−0.197	−0.298**
	(0.601)	(0.494)	(0.224)	(0.165)	(0.131)	(0.101)
≤ 34 Hours, spouse’s degree field	0.281	0.780	0.672	1.079**	0.752*	0.565*
	(1.350)	(0.930)	(0.578)	(0.398)	(0.346)	(0.285)
N	170,697	205,507	156,942	188,033	156,942	188,033

The results in Tables 7A and 7B indicate some important heterogeneity based on age of youngest child. The negative participation effects of long hours in a spouse’s degree field are almost exclusively experienced by married women whose youngest child is younger than 5 years old. By contrast, when the sample is restricted to workers, negative effects of long hours in a spouse’s degree field on earnings, wages, and weekly hours are relatively similar for mothers with and without a child under the age of 5. Although the analyses in Tables 5 and 6 found very little evidence of effects of short hours in a spouse’s degree field, there is some evidence of positive wage and earnings effects of short hours in a husband’s degree field for married working mothers whose youngest child is age 5 or older.

Table 7B.

Labor Supply Analysis by Age of Youngest Child

	Employed		Weekly hours		Weekly hours, workers only
	Diff 1	Diff 2	Diff 1	Diff 2	Diff 1	Diff 2
A. Youngest child < 5
≥ 50 Hours, spouse’s degree field	−0.243***(0.0657)	−0.185**(0.0608)	−11.23***(2.982)	−9.820***(2.539)	−5.101*(2.193)	−4.241*(1.723)
≤ 34 Hours, spouse’s degree field	−0.230(0.173)	−0.176(0.148)	−12.61(8.840)	−5.592(7.173)	−6.392(5.657)	0.107(4.625)
N	275,429	250,942	240,999	215,347	240,999	215,347
B. Youngest child ≥ 5
≥ 50 Hours, spouse’s degree field	−0.0494(0.0510)	−0.0153(0.0477)	−5.669⁺(3.034)	−4.869*(2.162)	−4.785⁺(2.481)	−4.086*(1.667)
≤ 34 Hours, spouse’s degree field	0.0572(0.112)	0.156⁺(0.0861)	−6.744(6.766)	3.346(5.330)	−6.482(5.269)	4.794(4.207)
N	170,697	205,507	156,942	188,033	156,942	188,033

Notes:Tables 7A and 7B replicate the analysis in columns (4) and (5) of Tables 5 and 6, splitting the sample based on age of youngest child. Standard errors, clustered on spouse’s degree field, in parentheses.

p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

Occupational Choice

Long hours in a husband’s degree field could affect a wife’s labor market outcomes through a wife’s choice of occupation and/or through effects on a wife’s outcomes conditional on occupation. Tables 8 and 9 explore these pathways. First Table 8 replicates the analyses in Tables 5 and 6, but adds controls for detailed own occupation fixed effects. Results show a modest reduction in magnitude for the estimates in Table 8 compared to those reported in columns (4) and (5) of Tables 5 and 6. The similarity of the estimates in Table 8 to those in Tables 5 and 6 suggests that most of the effects estimated in Tables 5 and 6 operate within rather than between occupations.

Table 8.

Controlling for Detailed Occupation Fixed Effects

	Log(Earnings)		Log(Earnings), workers only		Log(Wage), workers only
	Diff 1	Diff 2	Diff 1	Diff 2	Diff 1	Diff 2
≥ 50 Hours, spouse’s degree field	−1.340*(0.551)	−1.203*(0.476)	−0.453**(0.154)	−0.452***(0.134)	−0.158⁺(0.0803)	−0.186*(0.0721)
≤ 34 Hours, spouse’s degree field	−0.866(1.486)	−0.695(1.119)	0.174(0.400)	0.244(0.299)	0.311(0.197)	0.135(0.183)
N	446,126	346,897	397,941	301,063	397,941	301,063
	Employed		Weekly hours		Weekly hours, workers only
	Diff 1	Diff 2	Diff 1	Diff 2	Diff 1	Diff 2
≥ 50 Hours, spouse’s degree field	−0.145**(0.0491)	−0.105*(0.0482)	−7.709**(2.414)	−7.020***(1.943)	−4.286*(1.819)	−3.678**(1.332)
≤ 34 Hours, spouse’s degree field	−0.0757(0.116)	−0.0341(0.100)	−6.116(6.571)	−2.091(5.163)	−3.245(4.285)	0.992(3.613)
N	446,126	346,897	446,126	346,897	397,941	301,063

Notes: Top panel replicates columns (4) and (5) results of Table 5, with the addition of detailed own occupation fixed effects. Bottom panel replicates columns (4) and (5) of Table 6, with the addition of detailed occupation fixed effects. Standard errors, clustered on spouse’s degree field, in parentheses.

p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

Table 9 reports results using the same Equation (1) specification from Tables 5 and 6, but uses an indicator for high-earning occupation as the dependent variable. In panels A and C of Table 9, the outcome is an indicator that equals 1 for individuals whose occupation has average male earnings greater than or equal to the median ($66,466 for women and $85,759 for men, as reported in Table 1). Because occupation is reported for the most recent job held in the past five years, not all who report a high-earning occupation are currently working. Panels B and D of Table 9, therefore, use instead an indicator for currently working in a high-earning occupation. Panels C and D restrict the sample to individuals who specialized in high-earning degree fields (above the median of $76,350 for women and $86,310 for men, as reported in Table 1).

Table 9.

Participation in High-Earning Occupations, by Age of Youngest Child

	Youngest child < 5 years		Youngest child ≥ 5 years
	Diff 1	Diff 2	Diff 1	Diff 2
A. High-earning occupation
≥ 50 Hours in spouse’s degree field	−0.0286(0.0722)	0.0166(0.0567)	−0.150⁺(0.0815)	−0.189**(0.0633)
≤ 34 Hours in spouse’s degree field	0.196(0.133)	0.233*(0.0978)	0.278⁺(0.157)	0.0764(0.172)
N	275,429	250,942	170,697	205,507
B. Currently working in high-earning occupation
≥ 50 Hours in spouse’s degree field	−0.0974(0.0822)	−0.0519(0.0610)	−0.195*(0.0857)	−0.238***(0.0626)
≤ 34 Hours in spouse’s degree field	0.140(0.167)	0.172(0.122)	0.263(0.172)	0.0997(0.154)
N	275,429	250,942	170,697	205,507
C. High-earning occupation, high-earning degree field sample
≥ 50 Hours, spouse’s degree field	0.0503(0.0886)	0.0893(0.0838)	−0.183(0.121)	−0.284**(0.0970)
≤ 34 Hours, spouse’s degree field	0.328*(0.165)	0.358*(0.143)	0.372(0.233)	0.0389(0.218)
N	140,711	127,435	84,138	102,122
D. Currently working in high-earning occupation, high-earning degree field sample
≥ 50 Hours, spouse’s degree field	0.00244(0.104)	0.0163(0.0926)	−0.249⁺(0.127)	−0.349***(0.102)
≤ 34 Hours, spouse’s degree field	0.246(0.209)	0.230(0.173)	0.249(0.247)	−0.0113(0.222)
N	140,711	127,435	84,138	102,122

Notes: Outcome variable in panels A and C is an indicator for high-earning occupation. Indicator equals 1 if occupation (for most recent job in the past five years) has average male earnings that exceed the median reported in Table 1 ($66,466 for women and $85,759 for men). Outcome variable in panels B and D is an indicator for currently working in high-earning occupation. High-earning degree sample used in panels B and D is restricted to individuals for whom average male earnings in undergraduate degree field exceed the median reported in Table 1 ($76,350 for women and $86,310 for men). Control variables are the same as in Tables 5 and 6. Standard errors, clustered on spouse’s degree field, in parentheses.

p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

Table 9 reports estimates separately by age of youngest child. For married women with a child younger than age 5, estimates for long hours in a spouse’s degree field are small, statistically insignificant, and of mixed sign. For married women with a youngest child aged 5 and older, however, the estimates for long hours in a spouse’s degree field are all negative and statistically significant. These estimates suggest that for married women whose youngest child is 5 or older, long hours in a husband’s degree field are associated with a lower probability of working in a high-earning occupation.

Non-College Women Married to College-Educated Men

The analysis sample is restricted to married couples in which both the husband and the wife have college degrees, so that undergraduate degree field is reported for both the husband and the wife. Table 10 reports results from a similar analysis of women without college degrees who are married to husbands with a college degree. Because the spouse has a college degree, a spouse’s degree field, and therefore rates of long hours and short hours in a spouse’s degree field, are still observed. The specification in Equation (1) is still used, except that a wife’s own degree field fixed effects can no longer be included.

Table 10.

Non-College Women Married to College-Educated Husbands

	Log(Earnings)		Log(Earnings), workers only		Log(Wage), workers only
	Diff 1	Diff 2	Diff 1	Diff 2	Diff 1	Diff 2
≥ 50 Hours, spouse’s degree field	−1.079(0.942)	0.972(0.894)	−0.513⁺(0.276)	−0.546*(0.258)	−0.229⁺(0.136)	−0.322*(0.159)
≤ 34 Hours, spouse’s degree field	−0.199	2.025	−0.402	−0.865⁺	−0.0753	−0.406
	(2.167)	(2.111)	(0.750)	(0.515)	(0.363)	(0.324)
N	201,245	101,348	170,681	80,318	170,681	80,318
	Employed		Weekly hours		Weekly hours, workers only
	Diff 1	Diff 2	Diff 1	Diff 2	Diff 1	Diff 2
≥ 50 Hours, spouse’s degree field	−0.218**(0.0787)	−0.0862(0.0886)	−9.406*(4.171)	−0.0351(4.171)	−5.289⁺(2.719)	−2.944(3.037)
≤ 34 Hours, spouse’s degree field	−0.117(0.170)	−0.0489(0.195)	−6.215(9.902)	4.666(8.058)	−2.657(7.173)	0.169(5.962)
N	201,245	101,348	201,245	101,348	170,681	80,318

Notes: Diff 1 compares non-college women who have children and who are married to college-educated men and non-college men who have children and who are married to college-educated women. Diff 2 compares non-college women who have children and who are married to college-educated men and non-college women without children and who are married to college-educated men. Sample and specification are the same as those in Tables 5 and 6, except for exclusion of degree field fixed effects. Standard errors, clustered on spouse’s degree field, in parentheses.

p < 0.10; *p < 0.05; **p < 0.01; ***p < 0.001.

In Table 10, columns (1), (3), and (5) compare non-college mothers who have college-educated husbands to non-college fathers who have college-educated wives (Diff 1). Columns (2), (4), and (6) compare non-college mothers who have college-educated husbands to non-college women without children who have college-educated husbands (Diff 2). The coefficients on long hours in a spouse’s degree field are negative, with one exception, but statistical significance is reduced relative to Tables 5 and 6. The Table 10 estimates therefore are consistent with a differential negative effect of long hours in a husband’s degree field on non-college women with children who are married to college-educated husbands, but with less statistical significance than observed in Tables 5 and 6.

Conclusions

Estimates in this article are consistent with a differential negative effect of long hours in a spouse’s degree field on the earnings, wages, and labor supply of married women with children. Caution must be exercised, however, when interpreting the magnitude of the estimates. First, Table A.2 and A.3 in the Online Appendix indicate that magnitudes are sensitive to the exact definitions of long hours and short hours used in the analysis. Second, estimates for the comparison groups suggest that there is substantial bias due to assortative matching. Because it is not possible to claim that this bias is identical across married women with children and the two comparison groups, the differences in estimates do not necessarily difference out the assortative matching bias.

A negative relationship between long hours in a husband’s career and the labor market outcomes of married women with children is also consistent with differential selection into marriage, in which women who anticipate lower participation or earnings after they have children are more likely to marry men trained for demanding careers. Although it is not possible to rule out this alternative explanation, negative selection on unobserved characteristics into marriage with men trained for demanding careers is not consistent with considerable evidence in this article of positive selection on observed characteristics into marriage with men trained for demanding careers.

The findings in this article provide important context when considering labor market outcomes of high-skilled women who have trained for demanding, high-earning careers. Much attention has been given to the fact that the long hours and inflexibility in many careers are not family friendly and may impede the labor market success of women with children. Less attention has been paid to the fact that, because of assortative matching, these high-skilled women are often also married to men who have trained for similarly demanding careers. Therefore, married women in these fields may find it difficult to reconcile the demand for long hours in their career with the demand for long hours in their husband’s career and the time demands of children. To the extent that couples choose to increase household specialization after having children, couples in which at least one member experiences demand for long hours of work may choose to specialize even more. The results in this article suggest that women married to a spouse from a demanding profession are more likely to reallocate effort away from the labor market and to increase specialization in the household than are men who are married to a spouse from a demanding profession.

Finally, although most analysis in this article is conducted using a sample of married couples in which both spouses have completed college, additional analyses of women without college degrees married to college-educated husbands produce similar estimates, though with less statistical significance. It is not possible to produce analogous estimates for women without college degrees married to husbands without college degrees.

Supplemental Material

ILRR_McKinnish_Supplemental-Online-Appendix – Supplemental material for Prevalence of Long Work Hours by Spouse’s Degree Field and the Labor Market Outcomes of Skilled Women

Supplemental material, ILRR_McKinnish_Supplemental-Online-Appendix for Prevalence of Long Work Hours by Spouse’s Degree Field and the Labor Market Outcomes of Skilled Women by Terra McKinnish in ILR Review

Footnotes

For information regarding the data and/or computer programs used for this study, please address correspondence to terra.mckinnish@colorado.edu.

ORCID iD

Terra McKinnish

An Online Appendix is available at .

1

ACS data were accessed through the Integrated Public Use Microdata Series (IPUMS) web server ().

2

Estimates of the effect of long hours in a husband’s degree field on the employment of married women with children become even more negative if individuals who have not worked in the past five years are included in the analysis sample.

3

To calculate wage characteristics, the hourly wage is first calculated for each worker by dividing annual earnings by annual hours. Annual hours are calculated by multiplying weeks worked last year × usual hours per week. Because post-2008 ACS reports weeks of work in intervals, weeks of work are taken as the midpoint of the reported interval. Specifically, week values of 7, 20, 33, 43.5, 48.5, and 51, are used, respectively, for the reported intervals 1–13, 14–26, 27–39, 40–47, 48–49, and 50–52.

4

Differences in means across the three columns are all statistically significant at the 1% level for all variables in .

5

Using residual wages, conditional on an age polynomial, race categories, Hispanic ethnicity, immigrant and citizenship status, when calculating degree-field-level controls has no appreciable effect on the coefficient estimates for long hours and short hours in a husband’s degree field.

References

Bertrand

Marianne

Goldin

Claudia

Katz

Lawrence F.

2010. Dynamics of the gender gap for young professionals in the financial and corporate sectors. American Economic Journal: Applied Economics 2(3): 228–55.

Bianchi

Suzanne M.

Milkie

Melissa A.

Sayer

Liana C.

Robinson

John P.

2000. Is anyone doing the housework? Trends in the gender division of household labor. Social Forces 79(1): 191–228.

Blau

Francine D.

Kahn

Lawrence M.

2017. The gender wage gap: Extent, trends, and explanations. Journal of Economic Literature 55(3): 789–865.

Blossfeld

Hans-Peter

. 2009. Educational assortative marriage in comparative perspective. Annual Review of Sociology 35: 513–30.

Blossfeld

Hans-Peter

Timm

Andreas

(Eds.). 2003. Who Marries Whom? Educational Systems as Marriage Markets in Modern Societies. Dordrecht, the Netherlands: Kluwer Academic.

Cha

Youngjoo

. 2010. Reinforcing separate spheres: The effect of spousal overwork on men’s and women’s employment in dual-earner households. American Sociological Review 75(2): 303–29.

Cha

Youngjoo

Weeden

Kim A.

2014. Overwork and the slow convergence in the gender gap in wages. American Sociological Review 79(3): 457–84.

Cortés

Patricia

Pan

Jessica

. 2016. Prevalence of long hours and skilled women’s occupational choices. IZA Discussion Paper No. 10225. Bonn, Germany: Institute of Labor Economics.

Flabbi

Luca

Moro

Andrea

. 2012. The effect of job flexibility on female labor market outcomes: Estimates from a search and bargaining model. Journal of Econometrics 168(1): 81–95.

10.

Goldin

Claudia

. 2014. A grand gender convergence: Its last chapter. American Economic Review 104(4): 1091–119.

11.

Kalmijn

Matthijs

. 1998. Intermarriage and homogamy: Causes, patterns, trends. Annual Review of Sociology 24: 395–421.

12.

Mare

Robert D.

1991. Five decades of educational assortative mating. American Sociological Review 56(1): 15–32.

13.

McKinnish

Terra

. 2008. Spousal mobility and earnings. Demography 45(4): 829–49.

14.

Pertold-Gebicka

Barbara

Pertold

Filip

Gupta

Nabanita Datta

. 2016. Employment adjustments around childbirth. IZA Discussion Paper No. 9685. Bonn, Germany: Institute of Labor Economics.

15.

Ruggles

Steven

Flood

Sarah

Goeken

Ronald

Grover

Josiah

Meyer

Erin

Pacas

Jose

Sobek

Matthew

. 2019. Integrated Public Use Microdata Series USA: Version 9.0 [data set]. Minneapolis, MN: IPUMS. https://doi.org/10.18128/D010.V9.0.

16.

Schwartz

Christine R.

Mare

Robert D.

2005. Trends in educational assortative marriage from 1940 to 2003. Demography 42(4): 621–46.

17.

Stone

Pamela

. 2007. Opting Out? Why Women Really Quit Careers and Head Home. Berkeley: University of California Press.

18.

Wiswall

Matthew

Zafar

Basit

. 2018. Preference for the workplace, investment in human capital, and gender. Quarterly Journal of Economics 133(1): 456–507.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.20 MB