Social Science Research 46 (2014) 72–84

Contents lists available at ScienceDirect

Social Science Research journal homepage: www.elsevier.com/locate/ssresearch

Housework, children, and women’s wages across racial–ethnic groups Heather Macpherson Parrott ⇑ Long Island University, Post, Department of Sociology and Anthropology, 720 Northern Boulevard, Brookville, NY 11548, United States

a r t i c l e

i n f o

Article history: Received 17 May 2013 Revised 7 February 2014 Accepted 9 February 2014 Available online 17 February 2014 Keywords: Motherhood penalty Household labor Wages Gender Race/Ethnicity Work effort

a b s t r a c t Motherhood affects women’s household labor and paid employment, but little previous research has explored the extent to which hours of housework may explain per child wage penalties or differences in such penalties across racial–ethnic groups. In this paper, I use longitudinal Panel Study of Income Dynamics (PSID) data to examine how variations in household labor affect the motherhood penalty for White, Black, and Hispanic women. In doing so, I first assess how children affect hours of household labor across these groups and then explore the extent to which this household labor mediates the relationship between children and wages for these women. I find that household labor explains a portion of the motherhood penalty for White women, who experience the most dramatic increases in household labor with additional children. Black and Hispanic women experience slight increases in housework with additional children, but neither children nor housework affects their already low wages. Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction Women continue to earn less than men despite efforts to equalize pay (OECD, 2012). The gap in earnings is most pronounced when parental status is taken into account—mothers encounter lower wages than women without children and are further penalized with each additional child (Budig et al., 2012; Correll et al., 2007; Glauber, 2007). This motherhood wage penalty has been linked to a number of larger gender issues that disproportionately affect mothers as compared to men and childless women, including occupational segregation (Shauman, 2006), employment discrimination (Benard and Correll, 2010; Correll et al., 2007), the cultural devaluation of women’s labor (Cohen and Huffman, 2003; England et al., 2002), and the availability of family-friendly public policies across countries (Budig et al., 2012). Each of the proposed explanations of the motherhood penalty touches on the actual, expected, or perceived impact of household labor on women’s wages, but little research has explicitly examined the relationship between household work and the motherhood penalty. Neither the motherhood penalty nor time spent in household labor is consistent across racial–ethnic groups. White women pay a higher price for motherhood than Black or Hispanic women (Budig and England, 2001; Glauber, 2007; Waldfogel, 1997), and there are some indications that White women may also complete more hours of household labor than minority women (Artis and Pavalko, 2003; Sayer and Fine, 2011; Silver and Goldscheider, 1994). In this paper, I link these two bodies of research to explore the relationships between children, housework, and women’s wages across ⇑ Fax: +1 516 299 3177. E-mail address: [email protected] http://dx.doi.org/10.1016/j.ssresearch.2014.02.004 0049-089X/Ó 2014 Elsevier Inc. All rights reserved.

H.M. Parrott / Social Science Research 46 (2014) 72–84

73

racial–ethnic groups, including the extent to which household labor may be implicated in racial–ethnic differences in the motherhood penalty. Specifically, I use longitudinal data from the Panel Study of Income Dynamics (PSID) to address the following research question: How do variations in household labor affect the motherhood penalty for White, Black, and Hispanic women? To answer this question, I first assess how children may differentially affect hours of household labor across these groups. I then explore the extent to which this household labor mediates the relationship between children and wages for these women.

2. Background The motherhood penalty has been a frequently studied phenomenon both within and beyond the US (ex. Budig et al., 2012; Gangl and Ziefle, 2009; Glauber, 2007; Gupta and Smith, 2002). Women in the US make an average of 7% less per child (Budig and England, 2001), the majority of which remains unexplained through empirical research. Scholars have explored the extent to which motherhood can affect women’s work hours (Bardasi and Gornick, 2008; Webber and Williams, 2008), work experience (Klerman and Leibowitz, 1999; Ludenberg and Rose, 2000), seniority (Budig and England, 2001), and opportunities for employment and advancement via employment discrimination (Blair-Loy, 2005; Correll et al., 2007; Kmec, 2011) – all of which affect wages. Women’s contributions to the home are implicated in all of the above explanations. For example, an increase in household labor after the birth of a child may lead women to decrease their work hours or sacrifice work experience by taking time out of the labor force altogether. Women’s hours of household labor are inversely correlated with market wages (Hersch and Stratton, 1997; John and Shelton, 1997; McLennan, 2000; Stratton, 2001) and daily household tasks such as cooking and cleaning have the strongest negative effects on wages (Hersch and Stratton, 2000; Kühhirt and Ludwig, 2012; Noonan, 2001). Even if couples split housework relatively equally before having children, the birth of a child increases the amount of time that mothers spend in housework more so than fathers—widening existing gaps in household labor (Baxter et al., 2008; Bianchi et al., 2000; John and Shelton, 1997; Sanchez and Thomson, 1997) and likely increasing pay disparities as well. The impact that household labor has on the motherhood penalty via differences in human capital development and occupational choices has been accounted for in previous studies; however, housework may affect women’s work lives in additional ways. Notably, household labor may affect women’s workplace productivity, or ‘‘work effort’’ (Keene and Reynolds, 2005; Maani and Cruickshank, 2010). According to Becker (1985), individuals have only a finite amount of time and energy to devote to the combination of paid and unpaid work, such that increases in household labor result in a decrease in the amount of effort that women are able to devote to paid labor. Changes in work effort have been repeatedly posited as a potential explanation for the motherhood penalty (Anderson et al., 2003; Budig and England, 2001), yet testing of this theory has been far less common. Bielby and Bielby (1988) discovered that motherhood, specifically motherhood of preschool age children, has a negative impact on women’s reported work effort, yet their study did not examine how differences in work effort affect wages. Anderson et al. (2003) find limited evidence that work effort affects motherhood wage penalties, though they only indirectly examine work effort by controlling for the age of the youngest child, a proxy for amount of household labor. Although Budig and Hodges (2010) address work effort in relation to the motherhood penalty, they do so by assessing time worked (years worked and hours per week), rather than productivity during work time, and do not directly measure household labor at all. In short, household labor may affect wages due to change in work effort, but this explanation has been under examined in previous research. The story is more complicated, however, in that both the motherhood penalty and hours of household labor vary by race and ethnicity (Budig and England, 2001; Glauber, 2007; Sayer and Fine, 2011; Wight et al., 2013). But there is a paradox. The motherhood penalty is one area where Black and Hispanic women are not economically disadvantaged in comparison to White women. White women face larger penalties than minority women (Anderson et al., 2003; Budig and England, 2001; Glauber, 2007), yet the causes of these racial differences have remained unclear. The paradox suggests that the reasons for women’s continued economic inequality vary across groups of women. Racial–ethnic differences in women’s contributions to the home, and how these contributions affect their work lives, may be part of this puzzle. White women have generally been found to take on more housework than minority women (Artis and Pavalko, 2003; John and Shelton, 1997; Silver and Goldscheider, 1994), with Black women completing the least amount of household labor of all racial–ethnic groups of women (Sayer and Fine, 2011; Wight et al., 2013). A number of factors may contribute to racial– ethnic differences in household labor, including differences in extended family assistance, differences in partner involvement, and pressures to perform household labor. Minority women traditionally receive more practical household assistance from extended family than White women (Cohen, 2002; Sarkisian et al., 2007; Sarkisian and Gerstel, 2004; Uttal, 1999). Although Hispanic women and Black women may use similar supports for negotiating household responsibilities, such as relying on extended family support networks (Cohen, 2002; Roschelle, 1999), there are indications that Black women receive more assistance and more clearly benefit in employment from the help they receive (Cohen, 2002; Coltrane, 2000; Cooke, 2007). Variation in women’s household labor may also be related to racial–ethnic differences in the division of household labor between partners. Generally, the time women devote to housework increases with marriage, but men do not similarly increase their household labor and may even decrease the amount of time they devote to housework (Bianchi et al., 2000; Davis et al., 2007; Hersch and Stratton, 1994). Some research has found that Black (Cooke, 2007; Cooksey and Fondell,

74

H.M. Parrott / Social Science Research 46 (2014) 72–84

1996; Kamo and Cohen, 1998; Penha-Lopes, 2006) and Hispanic men (Coltrane et al., 2004; Cooksey and Fondell, 1996) contribute more to household labor than White men. In contrast, recent research suggests that the division of labor is most unequal among Hispanic couples and that White men do substantially more housework than their minority counterparts (Wight et al., 2013). Contributions of male partners can help alleviate women’s work-family strain, yet it is unclear which women receive from such assistance. There are additionally racial–ethnic differences in the total amount of household labor completed within the home, with less housework completed in Black homes than in White or Hispanic homes (Cooke, 2007; Sayer and Fine, 2011; Wight et al., 2013). The divide between Black and White homes may reflect racial–ethnic differences in pressures to perform household labor. For example, as Glenn (2000) points out, the social construction of motherhood for middle-class White women has historically included high standards of cleanliness and an emphasis on nurturing children. Such high domestic expectations have been especially unrealistic for minority women who have traditionally had to balance with paid employment. Currently, mothers of all racial–ethnic groups participate in the paid labor force. The increase of White mothers in the workforce corresponded with a significant decrease in the amount of total housework completed by women (Bianchi et al., 2000), a decrease that can be partially attributed to the outsourcing of domestic labor (Browne and Misra, 2003) and often the exploitation of minority domestic workers (Glenn, 1992; Glenn, 2000). Persistent racial–ethnic differences in household labor may reflect continuing disparities in the construction of motherhood, whereby White (and perhaps Hispanic) women hold themselves to higher standards of domestic labor. Although total hours of housework may affect women’s wages, assessing the impact of household labor on the motherhood penalty across racial–ethnic groups also calls for an exploration how additional children may affect such labor. Women’s hours of housework increase with additional children (Bianchi et al., 2000), yet we do not yet have an understanding of how race/ethnicity may moderate the relationship between children and household labor. Spouses appear to be the main source of household support for White women. However, White husbands typically increase participation in paid labor with additional children more than fathers of other races (Glauber, 2008), which could place increased pressure on White (married) women with regard to housework. In contrast, support from extended family may mitigate the negative effects of additional children on both housework and wages for minority women. Thus, I anticipate that children will affect women’s hours of household labor differently across racial–ethnic groups, with White women experiencing the most notable increases in housework with additional children. I further predict that hours of household labor will mediate the relationship between children and wages for all women, but will explain a larger portion of the penalty for White women as compared to Black and Hispanic women. If White women do have steeper increases in household labor with children, as hypothesized above, such increases in labor may help to explain why White women have larger per child wage penalties. Household labor may also explain more of the penalty for White women simply because Black and Hispanic women much lower residual penalties (Budig and England, 2001; Glauber, 2007), leaving more to be explained through additional research. Additionally, while research generally suggests that increases in household labor negatively affect wages for all women (Hersch and Stratton, 1997; Stratton, 2001), there are some indications that household labor is only correlated with wages for married White women (McLennan, 2000). Increases in household labor with children may therefore have a stronger impact on wages for White women, especially married White women.

3. Data, variables, and analytic strategy 3.1. Data I use PSID main family data from 1985 to 2011 to examine the relationships between household labor, children, and wages across groups of women. From 1968 to 1997, PSID data were collected yearly on the same sample of individuals and families. Several changes to data collection took place in 1997 due to funding issues and efforts to keep the sample nationally representative. These included a change to a semi-annual rather than an annual data collection, reducing the original core sample from 8500 to 6168 families, and adding a sample of post-1968 immigrant families and their adult children. The PSID contains an oversampling of Black, Hispanic, and low-income families—making the dataset ideal for examining racial–ethnic differences in household labor and the effects of such labor on wages. I chose 1985 as the starting year because the PSID did not start asking the race of the wife until 1985. Since women’s racial–ethnic group is a key consideration in this study, I chose to start the dataset at this point. Ultimately, the sample in this study contains 20 years of family data (yearly data from 1985 to 1997, plus bi-yearly data from 1999 to 2011). Person-years is the unit of analysis; thus, the sample sizes are reported in two ways: the total years of data among all of the participants (person-years) and the total number of women in the study. Using the main family data for these analyses required a transformation of the data from following families to following women over time. My interest in this study is with women, more so than with families, and it is important that I follow the same woman across data years in order to properly assess women’s wage growth over time. To increase the reliability and validity of the data and its structure, I conducted three main checks. First, I used family composition variables in the main family data to make sure I had accounted for all changes in the family over time. Second, I tracked the woman’s age over time, making sure that it advanced logically. I considered the woman to be new to the family if her age decreased

H.M. Parrott / Social Science Research 46 (2014) 72–84

75

between years or increased more than two years, and then advanced sequentially thereafter. Third, I matched the individual data files to the family data to check for consistency in date of birth for those women who were included in the individual data collection. In cases where differences seemed due to interviewer or reporting error—for example, if the year of birth was consistent over time except for one data collection when it was reported as one year higher or lower—I assumed that the woman was in fact the same. I had a total of 22,954 women and 138,428 person-years after making these changes and reshaping the data to include a line of data for each woman for each year. From this point, I used theoretical and methodological criteria for inclusion in the study. I first eliminated any years in which the woman was not working for pay and years in which the woman’s average hourly earnings were missing. I was unable to include women in the sample who did not work or report wages, since women’s hourly wages are one of the main dependent variables of interest. It is possible that these women left the work force because of they face more severe wage penalties; thus, mothers with the most severe employment or work-family balance issues unfortunately may not be included in the sample, potentially biasing my results. Nevertheless, I am able to capture a substantial sample of women who have had either consistent or sporadic labor force involvement. Additionally, I restricted the sample to women between the ages of 18 and 55 in 1985 to 2011. The sample only includes women who are White, Black, or Hispanic due to the relatively small sample sizes of other racial groups. With these restrictions and utilizing listwise deletion1, the final sample for this set of analyses consists of 14,755 women and 74,228 personyears with an average of 5.01 years of employment data for each woman.

3.2. Variables I am interested in the effects of children on hours of household labor and household labor on the motherhood wage penalty; thus, hours of household labor is both a dependent and independent variable in these analyses. Hours of household labor, first a dependent variable, is the reported average number of hours that the woman spends ‘‘cooking, cleaning, and doing other work around the house’’ each week. When assessing the mediating effects of household labor on the relationship between additional children and women’s hourly wages, the dependent variable is the natural log of the woman’s hourly wages at her current job. Number of children and hours of housework are the main independent variable of interest. Number of children is measured two ways: continuously according to how many children under the age of 18 are in the household in the given year and as a set of dummy variables. By measuring number of children continuously, I am able to run mediation tests in this portion of the analysis to assess the impact of household labor on the motherhood penalty across models. I also assess children as a set of dummy variables because previous research has revealed a non-linear relationship between children and wages (Budig and England, 2001; Glauber, 2007; Waldfogel, 1997), as well as between children and hours of household labor (Killewald and Gough, 2010). I examine changes in these variables across racial–ethnic groups. The race/ethnicity variables are dummy coded according to the mother’s primary racial–ethnic identification. Women were coded as ‘‘Hispanic’’ if they either noted their race as Hispanic (as was an option in 1994–1999) and/or considered themselves to be of Hispanic origin (a separate question for ethnicity that was included in all data years). Despite the variation in questions about Hispanic origin, coding Hispanic in this way resulted in consistent racial–ethnic categorization across years in the PSID. I control for marital status, an important addition considering systematic differences in marital status across racial–ethnic groups. I include marital status as a series of dummy variables—never married, married, and divorced—with married as the reference category. The divorced category includes all women who are divorced, separated, and widowed. There are a number of work-related control variables used in the analyses. For human capital variables, I include seniority (number of years at current job), number of years the woman worked full-time during the observation period, number of years woman worked part-time in the observation period, years of education, and whether the woman is currently enrolled in school. I constructed the variable for years of education differently for given years, given limitations of the survey data. The data from 1994 to 2011 each contain a variable for the number of years of education. For 1985 to 1993, I constructed the variable using years completed of high school, years completed of college, and my approximation of years that it takes to complete professional degrees. I added two years of education for those completing master’s degrees, three for law degrees, five for doctorate degrees, and four for medical school. The means and standard deviations for the years in which I calculated the years of education are similar to years in which I use PSID constructed variables. The analyses also contain a set of variables for job characteristics that have been shown to affect wages. I include whether the woman worked full-time (35 + hours per week) and a set of dummy variables for occupational type. I control for occupation using the six main occupational categories from the 2000 US Census: professional; service; sales and office; farming, 1 As noted above, cases in which the woman’s hourly wage was missing were removed from the dataset. A total of 10 other variables used in these analyses contained missing values, including the 6 variables that rely upon occupational data. I explored using multiple imputation to address the problem of missing data and retain as many cases as possible, rather than using listwise deletion. More specifically, I used Royston’s ice command with Stata to create 5 complete datasets (Carlin et al., 2008). In each dataset, missing values are replaced with reasonable values based on other, observed values in the data. Analyses are then run on each of the datasets, and the results are combined into one common set of results with adjusted standard errors. When comparing the two methods of dealing with missing values – listwise deletion and multiple imputation – I found that the results were nearly identical and that listwise deletion ultimately only eliminated 2787 person-years of data (3.6% of the final sample). Thus, for the sake of simplicity, I use listwise regression throughout this paper.

76

H.M. Parrott / Social Science Research 46 (2014) 72–84

fishing, and forestry; construction and maintenance; and production and transportation. Professional occupations is the reference category. Finally, I controlled for who responded to the PSID survey, since this may have a significant impact on the reported hours of household labor. The respondent was either the head of household or wife in all but the 0.4% of the observations where another family member was surveyed. I coded the variable ‘‘female respondent’’ as 1 if the respondent was the wife or female head of household. See Table 1 for the means of all variables used in these analyses. 3.3. Analytic strategy I arranged the data into a pooled, cross-section time series. I ran the Hausman (1978) specification test on all models and determined that fixed-effects regression was more appropriate for these analyses than random-effects regression. There are two main limitations of fixed-effect methods. First, the standard fixed-effects models do not produce estimates for variables that are consistent over time. Second, in some cases the standard errors will be higher and p-values consequently larger than in random-effects models because fixed-effects models do not account for between individual differences (Allison, 2005). Fixed-effects can be used to explore categorical differences between people for characteristics that remain constant over time, such as race, by using one of two strategies: interacting variables that do not change over time with a variable that

Table 1 Descriptions of measurement and means of select variables, PSID data 1985–2011. Variables

Description

White Women

Black Women

Hispanic Women

Hourly Wages (2011 dollars) Natural Log of Hourly Wages Number of Children No Children 1 Child 2 Children 3 Children 4+ Children Hours of Housework

Women’s Hourly Wages (2011 dollars)

18.99

15.45*

15.28*

Natural log of women’s hourly wages for each year

2.30

2.13*

2.09*

Number of children under 18 years old in the household Dummy variable = 1 if no children under 18 years old are in the household Dummy variable = 1 if one child under 18 years old is in the household Dummy variable = 1 if two children under 18 years old are in the household Dummy variable = 1 if three children under 18 years old are in the household Dummy variable = 1 if four or more children under 18 years old are in the household Reported weekly hours spent ‘‘cooking, cleaning, and doing other work around the house’’

1.05 0.42 0.23 0.24 0.08 0.02 16.41

1.41* 0.28* 0.28* 0.26* 0.12* 0.06* 14.56*

1.48* 0.28* 0.25* 0.27* 0.14* 0.07* 19.65*

5.08 3.54

5.75* 3.73*

4.24* 2.39*

1.85

1.13*

0.89*

13.50 0.03

12.66* 0.04*

11.59* 0.04

36.54

38.01*

37.26*

0.40

0.21*

0.22*

0.17

0.30*

0.23*

0.34

0.32*

0.33*

0.00

0.00

0.03*

0.02

0.02

0.03

0.06

0.14*

0.16*

0.77 0.10 0.13 36.78 0.58

0.47* 0.29* 0.25* 36.15* 0.82*

0.72* 0.11* 0.17* 36.38* 0.64*

43,631 7867

24,598 4991

5999 1897

Human Capital Variables Years at Current Job Years at current job Years Worked FullYears worked full-time in paid labor force during observation period Time Years Worked PartYears worked part-time in paid labor force during observation period Time Years of Education Number of years of education Enrolled in School Dummy variable = 1 if woman is enrolled in an educational program Job Characteristics Weekly Work Hours

Hours worked per week

Occupational Category Professional Dummy variable = 1 if woman works in a professional or managerial occupation based on Census Code Service Dummy variable = 1 if woman works in a service occupation based on her Census Occupational Code Sales and Office Dummy variable = 1 if woman works in a service or office occupation based on her Census Occupational Code Farming, Fishing, Dummy variable = 1 if woman works in a farming, fishing, or forestry occupation and Forestry based on her Census Occupational Code Construction and Dummy variable = 1 if woman works in a construction, extraction, or maintenance Maintenance occupation based on Census Occupational Code Production and Dummy variable = 1 if woman works in a production, transportation, or material Transportation moving occupation based on Census Code Marital Status Married Never-Married Divorced Age Female Respondent N Person-Years N Women *

Dummy variable = 1 Dummy variable = 1 Dummy variable = 1 Age of woman Dummy variable = 1

if woman is married (reference category) if woman has never been married if woman is widowed, divorced, or separated if survey respondent was the sample female

Indicates a significant difference in means between given group of women (Black or Hispanic) and White women (p < 0.001, two-tailed tests).

77

H.M. Parrott / Social Science Research 46 (2014) 72–84

does vary over time within individuals, or using separate models for the different sub-groups. While this partially resolves the first limitation, it does not address the second. Despite these constraints, fixed-effects is the best method for these analyses because of its ability to control for unmeasured or even immeasurable stable differences between individuals (Allison, 2005; Halaby, 2004). The analyses in this paper are divided into two parts. I first use fixed-effects models to examine the degree to which race moderates the relationship between children and household labor (Table 2 and Table 3). I then assess the potential mediating effects of hours of household labor in the relationship between children and women’s hourly wages across racial–ethnic groups from fixed-effects models (Tables 4 and 5). Standard errors for Tables 2–5 are available upon request. I also conduct multi-level mediation tests using Stata’s ml_mediation command, adapted from Krull and MacKinnon (2001), to assess whether hours of housework independently affects the relationship between number of children and women’s hourly wages. The ml_mediation command calculates the indirect effect of household labor by multiplying the coefficients for paths a and b (see Fig. 1), and the proportion mediated is this indirect effect divided by the total effect (i.e. the sum of the direct and indirect effects). To run such tests through Stata, the sample was necessarily divided into subsets. I then ran mediation tests using full fixed-effects models for each racial–ethnic group (White, Black, and Hispanic), as well as each marital status within these racial–ethnic groups. I present all significant results with bootstrapped standard errors in Table 6.

4. Results and discussion 4.1. Race/ethnicity, children, and household labor The descriptive statistics and t-tests in Table 1 present evidence that there are racial–ethnic differences in both wages and hours of household labor. Hourly wages are significantly higher for employed White women ($18.99/h) than for employed Black ($15.45/h) or Hispanic women ($15.28/h). White women complete more household labor on average (16.41 h) than Black women (14.56 h), but fewer hours than Hispanic women (19.65 h) – a finding consistent with recent research by Wight et al. (2013). Racial–ethnic differences in the outsourcing of household labor could help elucidate why White women complete fewer hours of housework than Hispanic women, yet such differences do little to explain Black–White disparities in housework. (See Fig. 2). The fixed effects models in Tables 2 and 3 expand on the descriptive housework data to assess the effects of number of children on household labor across racial–ethnic groups. Table 2 demonstrates that there are in fact significant differences in

Table 2 Fixed effects regression models of women’s hours of household labor on number of children, PSID data 1985– 2011. Number of Children (vs. White) Children  Black Children  Hispanic

2.166*** 1.276*** 0.596**

Human Capital Variables Years at Current Job Years Worked Full-Time Years Worked Part-Time Years of Education Enrolled in School

0.106*** 0.671*** 0.411*** 0.161 0.619**

Job Characteristics Weekly Work Hours Occupation (vs. Professional) Service Sales and Office Fishing, Farming, Forestry Construction and Maintenance Production and Transportation

1.209*** 0.116 0.375 0.076 0.477*

Marital Status (vs. Married) Never-Married Divorced Female Respondent

1.215** 1.343*** 0.329*

Number of Person-Years Number of Women

0.087***

74,228 14,755

Notes: All models also include controls for age, age2, and dummy variables for survey years. The category of divorced women includes women who are divorced, separated, or widowed. Results are weighted. * p < 0.05, two-tailed tests. ** p < 0.01, two-tailed tests. *** p < 0.001, two-tailed tests.

78

H.M. Parrott / Social Science Research 46 (2014) 72–84

Table 3 Fixed effects regression models of women’s hours of household labor on number of children by racial–ethnic group, PSID data 1985–2011. Model 1 White Number of Children (continuous)

2.087***

Model 2 Black 0.859***

Hispanic

White

Black

Hispanic

1.473***

Number of Children (vs. No Children) 1 Child 2 Children 3 Children 4+ Children

2.794*** 4.692*** 6.372*** 7.871***

1.396*** 2.041*** 3.166*** 3.385***

1.891** 3.498*** 5.471*** 5.953***

Human Capital Variables Years at Current Job Years Worked Full-Time Years Worked Part-Time Years of Education Enrolled in School

0.119*** 0.763*** 0.430*** 0.000 0.857**

0.080*** 0.466*** 0.354** 0.029 0.286

0.146** 1.048** 0.956* 0.610 0.695

0.118*** 0.758*** 0.422*** 0.176 0.847**

0.080*** 0.466*** 0.349** 0.025 0.280

0.150** 1.043** 0.937* 0.614 0.635

Job Characteristics Weekly Work Hours

0.113***

0.036***

0.064***

0.112***

0.036***

0.063***

1.839*** 0.259 1.178 0.136 0.670

0.337 0.362 0.314 0.471 0.191

0.372 0.131 0.515 0.743 0.406

1.834*** 0.241 1.220 0.124 0.645*

0.330 0.375 0.238 0.484 0.186

0.307 0.101 0.470 0.683 0.456

1.608** 0.237 0.636***

1.221 2.377*** 0.266

1.693 1.102 0.412

1.482** 0.212 0.637***

1.217 2.359*** 0.255

1.217 2.359 0.255

Occupation (vs. Professional) Service Sales and Office Fishing, Farming, Forestry Construction and Maintenance Production and Transportation Marital Status (vs. Married) Never-Married Divorced Female Respondent Number of Person-Years Number of Women

43,631 7867

24,598 4991

5999 1897

43,631 7867

24,598 4991

5999 1897

Notes: All models also include controls for age, age2, and dummy variables for survey years. The category of divorced women includes women who are divorced, separated, or widowed. Results are weighted. * p < 0.05, two-tailed tests. ** p < 0.01, two-tailed tests. *** p < 0 .001, two-tailed tests.

the effects of household labor by racial–ethnic group. White women experience a significantly greater increase in household labor with children as compared to both Black and Hispanic women. The models in Table 3 reveal that though the effect of children on household labor may be larger for employed White women, all three racial–ethnic groups experience significant per child increases in household labor. For a given employed White women, having an additional child increases her hours of household labor by an average of approximately 2 h per week. Each additional child amounts to an increase in only 0.654 h of housework for an employed Black woman and 0.822 additional hours for an employed Hispanic woman. These findings are highlighted in Fig. 1, where I use these fixed-effects models to graph the predicted hours of housework by racial–ethnic group and number of children. The figure displays the relatively dramatic increase in hours of housework with each additional child for employed White women, while Black and Hispanic women experience comparitively small per child increases in household labor. Employed Hispanic women complete more household labor on average than employed Black and White women; however, as predicted, White women experience a more dramatic increase in household labor with additional children. I tested the interactions between race and marital status for these models (in analysis not shown), but there were no notable differences within racial–ethnic groups by marital status. Employed women within each racial–ethnic group experience comparable per child increases in household labor regardless of marital status. There are several possible explanations for these findings. The types of practical support that minority women receive, such as from extended family, may remain fairly stable or even increase with the addition of more children. Meanwhile, the support that White women receive, primarily from spouses, is likely to diminish with additional children as husbands invest more in paid labor (Glauber, 2008). Yet White women of all marital statuses experience similar steep per child increases in household labor, suggesting that other factors may be (additionally) at play. For example, White women as a whole may in fact hold themselves to higher (perhaps unattainable) standards of household labor due to the construction of White middle class motherhood. Such standards may necessitate more labor as the number of children increases, especially if men’s contributions to household labor decrease. 4.2. Race/ethnicity, household labor, and the motherhood penalty Tables 4–6 examine the effects of household labor on the motherhood penalty and differences in this relationship across racial–ethnic groups. Table 4 is a summary table, presenting only the coefficients for children (both as a continuous variable

79

H.M. Parrott / Social Science Research 46 (2014) 72–84 Table 4 Summary of the effects of children on women’s hourly wages (ln) by racial–ethnic group from fixed-effects regression models, PSID data 1985–2011. Model 1 Number of Children (continuous) White Black Hispanic

0.036***

Model 2

Model 3

Model 4

0.050*** 0.008 0.009

Black 1 Child 2 Children 3 Children 4 + Children

Number of Person-Years Number of Women

0.002*** 74,228 14,755

Model 6

0.044*** 0.006 0.006

Number of Children (vs. No Children) White 1 Child 2 Children 3 Children 4 + Children

Hispanic 1 Child 2 Children 3 Children 4+ Children Hours of Housework

Model 5

0.032***

74,228 14,755

0.041*** 0.110*** 0.134*** 0.228***

0.033*** 0.096*** 0.116*** 0.206***

0.030 0.011 0.029 0.018

0.033 0.015 0.022 0.010

0.005 0.013 0.012 0.092

0.005 0.050 0.001 0.078 0.002***

0.002*** 74,228 14,755

74,228 14,755

74,228 14,755

74,228 14,755

Notes: All models also include controls for human capital variables, job characteristics, marital status, age, age2, and dummy variables for survey years. Results are weighted. * p < 0.05, two-tailed tests. ** p < 0.01, two-tailed tests. *** p < 0.001, two-tailed tests.

Table 5 Fixed effects models of women’s hourly wages (ln) on number of children and marital status by race/ethnicity, PSID data 1985–2011. White Women

Black Women

Hispanic Women

Number of Children Children  Never Married Children  Divorced

0.047*** 0.070*** 0.014

0.005 0.017 0.009

0.004 0.042 0.025

Hours of Housework Human Capital Variables Years at Current Job Years Worked Full-Time Years Worked Part-Time Years of Education Enrolled in School

0.003***

0.001

0.012

***

0.012 0.049*** 0.041*** 0.004 0.125***

***

0.009 0.018 0.021** 0.011 0.076***

0.007** 0.058*** 0.067*** 0.006 0.049

Job Characteristics Weekly Hours Worked

0.006***

0.010***

0.012***

0.213*** 0.062*** 0.304*** 0.008 0.034

0.101*** 0.037** 0.012 0.046 0.015

0.067 0.029 0.095 0.001 0.003

0.070* 0.024 0.037

0.037 0.028 0.049

0.054 0.224* 0.041

Occupation (vs. Professional) Service Sales and Office Fishing, Farming, Forestry Construction and Maintenance Production and Transportation Marital Status (vs. Married) Never Married Divorced Female Respondent Number of Person-Years Number of Women

43,631 7867

24,598 4991

5999 1897

Notes: All models also include controls for age, age2, and dummy variables for survey years. The category of divorced women includes women who are divorced, separated, or widowed. Results are weighted. * p < 0.05, two-tailed tests. ** p < 0.01, two-tailed tests. *** p < 0.001, two-tailed tests.

80

H.M. Parrott / Social Science Research 46 (2014) 72–84

Fig. 1. Mediation model.

Table 6 Summary of significant mediation tests assessing mediating effect of women’s hours of housework in the relationship between children and hourly wages across groups of women, PSID Data 1985–2011. Group

Indirect Effect Coefficient

Proportion of Total Mediated SE

All Women

0.004***

0.000

0.11

White Women (total) White Married Women White Divorced Women

0.008*** 0.007*** 0.003***

0.001 0.001 0.002

0.12 0.13 0.12

Note: Mediation tested using full fixed-effects models, SE calculated through bootstrapping. p < 0.05, two-tailed tests. ** p < 0.01, two-tailed tests. *** p < 0.001, two-tailed tests. *

Fig. 2. Predicted hours of housework by race–ethnicity and number of children from fixed regression models. Note: All models include controls for human capital variables, job characteristics, marital status, whether the respondent was female, age, age2, and dummy variables for survey years. Models were run separately by racial–ethnic group when calculating predicted hours of household labor. Results are weighted.

and set of dummy variables) from full fixed-effects models. Model 1 of Table 4 reveals a 3.6% motherhood penalty before hours of housework is included in the model. The inclusion of hours of housework (model 2) only decreases this percent to 3.2. Employed women as a whole experience a 3.2% decrease in hourly wages per child. However, these findings are not consistent across racial–ethnic groups. According to Table 4, employed Black and Hispanic women do not experience a significant motherhood penalty whether household labor is included in the model (model 4) or not (model 3). This trend remains even when number of children is assessed as a set of dummy variables interacted with racial–ethnic group. As summarized in models 5 and 6 of Table 4, Black and Hispanic women do not experience any motherhood wage penalty regardless of the number of children they have. In fact (in analysis not shown), these groups surprisingly don’t even experience penalties when human capital and job variables are removed from the models.

H.M. Parrott / Social Science Research 46 (2014) 72–84

81

White women are the only racial–ethnic group to experience a residual motherhood penalty. As shown in model 3 of Table 4, employed White women experience a 5.0% wage penalty with each child before hours of housework is included in the model. White women’s penalty decreases to 4.4% when hours of housework is added to the full model (model 4). Thus, while children increase household labor for all women, White women are the only group who experiences measurable negative effects of increases in household labor on their wages. In Table 5, I explore the effects of housework on wages within racial–ethnic groups to untangle the effects of race and marital status on motherhood wage penalties. Simply controlling for marital status is complicated in fixed-effects models, since the majority of women in this sample (92.4%) do not change marital status during the observation period. Therefore, I necessarily explore the effects of marital status within racial–ethnic categories using interaction effects (as in Table 5) or separate models (as in Table 6). For White women, I found significant differences in motherhood penalties across marital statuses. Only married and divorced White women face per child wage penalties, 4.7% and 3.3% respectively. Single White women do not experience a motherhood penalty in wages at all. This was true even when I included children as a set of dummy variables interacted with variables for marital status (in analysis not shown) – married and divorced women with children experience a wage penalty where single women did not. In contrast, motherhood penalties, or lack thereof, are not significantly different across marital statuses for Black and Hispanic women. No group of Black or Hispanic women experiences a wage penalty with children, even when number of children is assessed as a series of dummy variables (in analysis not shown). These findings runs counter to previous motherhood penalty research that has uncovered persistent motherhood penalties for at least some groups of Black women (Budig and England, 2001; Glauber, 2007). Notably, Glauber (2007) discovered motherhood wage penalties for married Black women with more than two children, a group which does not experience motherhood penalties here. This difference may have something to do with the data (PSID vs. NLSY) and resultant samples used. Although the average person-years per woman is smaller in this study, I examine a much larger number of women across racial–ethnic groups. For example, Glabuer’s (2007) sample of 1471 Black women can be compared with 4991 Black women here. Additionally, these data also include a much larger sampling of married Black women – 47% of Black women here as compared to 33% in Glauber’s study. Another noteworthy finding from Table 5 is that there are also racial–ethnic differences in the direct effect of housework on wages, a separate matter from per child wage penalties. White women experience a significant 0.3% decrease in wages per hour of weekly housework, while household labor has no significant direct impact on wages for Black and Hispanic women. This finding that housework only significantly negatively affects wages for White women corresponds with previous research by McLennan (2000). I conducted mediation tests to assess the extent to which household labor mediates the relationship between number of children and women’s hourly wages for each group of women.2 Hours of housework significantly mediates motherhood wage penalties for all women, and for a number of (White) subgroups – White women as a whole, married White women, and divorced White women. I present all significant results in Table 6. A total of 11% of the motherhood penalty for all women and 12% of the penalty for White women is mediated by hours of housework. Housework mediates a proportion of the penalty for married White women (13%) and White divorced women (12%), but not White single women. As noted above, I did not find significant differences in per child increases in household labor by marital status. Thus, while household labor does not increase more dramatically with additional children for White married women and White divorced women as compared to White nevermarried women, these groups are more affected by per child increases in household labor than never-married White women. Differences in the effects of household labor across White women likely has to do with married and divorced White women completing more household labor than never married women to begin with. There are indications that women increase their household labor when they get married (Bianchi et al., 2000) but do not necessarily decrease their labor significantly upon divorce or separation (Baxter et al., 2008), which may affect the baseline for White women who have children after marriage regardless of whether their marriage remains intact. Upon further investigation, married White women complete an average of 17.80 h of housework, while White divorced women complete an average 13.73 h and the disproportionately childless sample of never married White women complete an average of 8.86 h of housework per week. Even when just comparing White women who are mothers during the observation period, there are significant differences in housework across groups – married women (18.54 h) and divorced women (14.74) complete more weekly hours of housework than nevermarried women (12.75). Increases in household labor with children on top of already large contributions to household labor may cause strains on women’s work lives, since total hours of household labor have deleterious effects on women’s wages (Kühhirt and Ludwig, 2012; McLennan, 2000).

5. Conclusion Ultimately, this work contributes to a greater understanding of inequality among women, the connection between household labor and women’s wages, and how the motherhood penalty differs by racial–ethnic group. Although mitigating the amount of housework that women complete may not entirely alleviate the motherhood penalty for White women, this 2 Hayes (2009) recommends still running mediation tests even if the direct effect is not significant. With this in mind, I ran mediation tests on all subgroups regardless of whether the direct effect (motherhood penalty) was significant.

82

H.M. Parrott / Social Science Research 46 (2014) 72–84

domestic labor does play a role in the penalty. The more rapid increase in household labor that White women experience with additional children has an impact on their wages that is not explained by human capital variables (work experience, education, tenure) or occupational choices. Time spent doing housework does not affect wages or wage penalties for Black or Hispanic women. Black women spend fewer hours in housework and less dramatic increases in housework with children, perhaps reflecting higher levels of assistance from others in performing household tasks. This pattern of racial differences has been supported in other research (e.g., Cooksey and Fondell, 1996). These contributions may be a mechanism by which both Black and Hispanic women balance work and family in ways that limit some of the long-term negative effects of children on wages. The negative correlation between housework and wages could be construed as support for Becker’s (1985) work effort hypothesis. Increases in household labor may be met with decreases in work productivity, which in turn affects women’s wages. This interpretation should be approached with caution for two reasons. First, Stratton (2001) found that controlling for self reported work effort did little to explain the connection between household labor and women’s wages. Since the PSID data does not include a measure for work effort, I am unable to directly asses the connections between household labor, work effort, children, and wages in this study. Second, previous research has not uncovered systematic racial differences in work effort (Anderson et al., 2003). We would therefore expect to at least see that hours of household labor affect wages somewhat equivalently across racial–ethnic groups if the relationship between housework and motherhood penalties was in fact driven by changes in work effort. This is not the case. As an alternative explanation, increases in children and household labor may affect wages via employer discrimination. Specifically, employers may expect that mothers will be less devoted to their jobs as their family and family demands expand (Correll et al., 2007). Such expectations could result in stagnant wages even without any decrease in work effort. This form of discrimination may vary by race. For example, employers often stereotype Black women as single-mothers, resulting in the perception that they are more reliable at work because they need to support their families (Kennelly, 1999). Mothering and paid labor are not seen as at odds with one another for minority women, as the often are for White mothers (Duncan et al., 2003). There are also lingering perceptions, despite evidence to the contrary, that middle-class women are more likely than working-class women to leave the labor force after the birth of a child (Damaske, 2011). This may leave White middle-class women more susceptible than other groups to expectations that their dedication to the labor force will decrease as family demands increase. While women experience disadvantages in the labor force in comparison to men, the main sources of their disadvantage vary across racial–ethnic groups. Motherhood and household labor are key factors economically affecting White women. However, for minority women these factors are likely overshadowed by a host additional issues that have an impact on their wages, including racial–ethnic discrimination (Huffman and Cohen, 2004; Kennelly, 1999; Roscigno, 2007). The lack of wage penalties among Hispanic and Black women certainly do not signal greater economic well-being for these groups. Rather the overall wages are lower for minority women than White women (as seen in Table 1), leaving more variation to be explained for White women. Minority women tend to be concentrated in low-paying, unskilled jobs that have little wage variation to begin with (Anderson and Shapiro, 1996; Maume, 1999). Black women and immigrant Latina women also both experience difficulty in turning their human capital investments, such as education, into more tangible economic benefits, such as promotions or wage increases (Hall and Farkas, 2008; McGuire and Reskin, 1993). Therefore, the fact that these groups do not experience persistent motherhood penalties, in combination with their low average wages, may indicate that they have been confined to a wage floor. In exploring the relationships between children, household labor, and wages, across employed White, Black, and Hispanic women, I have addressed a number of holes in the literature. First, I examined racial–ethnic differences in the effects of additional children on women’s housework. Although previous research has explored racial–ethnic differences in total household labor and division of household labor, scholars had not explored differences in how children impact household labor across groups. I found that while Hispanic women complete more household labor per week on average, White women experience the steepest increase in household labor with additional children. I also directly assessed the effect of household labor on motherhood penalties for US women, only previously examined among West German women (Kühhirt and Ludwig, 2012), and I explored racial–ethnic differences in this relationship. I found that housework does mediate the relationship between children and women’s wages, but only for White women – specifically married and divorced White women. Although there are systematic racial–ethnic differences in the effects of children on household labor, I found that these effects do little to explain racial–ethnic differences in the motherhood penalty. Despite the contributions of the present research, four main issues related to the motherhood penalty remain unresolved and thus provide avenues for future research. First, why does household labor affect the motherhood penalty in wages? In this study I am only able to speculate as to why household labor affects women’s wages. PSID data do not contain a measure of work effort, which would be very helpful for determining whether increases in household labor actually decrease work effort. Additionally, the PSID’s definition of housework–‘‘cooking, cleaning, and doing other work around the house’’– is very vague. It is unclear whether childcare is even included in this definition, though whether an individual counts this as housework could change their hourly reports dramatically. The definition additionally fails to capture a variety of other tasks that may affect women’s lives and employment, including setting appointments for children, chauffeuring them between activities, and caring for them when they are ill. Larger household tasks such as grocery shopping would also not necessarily be captured in this definition. Multiple and more detailed measures of housework would help us understand what tasks are

H.M. Parrott / Social Science Research 46 (2014) 72–84

83

most detrimental to women and, by extension, would give us a better understanding of the ways in which household labor affects women’s work. Second, why are their racial–ethnic differences in the effects of children on household labor? Another limitation of this research is that the PSID data do not contain information on the division of household labor among partners, children, extended family, and even paid help—to my knowledge a deficiency of all existing longitudinal datasets. Detailed data collection on the division of household labor could help uncover the extent to which sharing household labor helps women to mitigate increases in household labor with additional children and the effects of housework on wages, as well as which types of assistance are most beneficial. This information could be helpful for determining, for example, whether Black women’s less dramatic increase in household labor per child is due to certain types of support for household labor or simply the completion of less domestic labor. Such measures could even assess the extent to which outsourcing domestic labor assists certain groups of women in the workplace. Based on this research, if White women are outsourcing domestic labor, this outsourcing does not appear to sufficiently mitigate their household labor or the effects of such labor. Third, why do Black and Hispanic women not experience motherhood wage penalties? As discussed above, the absence of a motherhood penalty for these groups cannot be interpreted as an indication of economic success. Black and Hispanic women have significantly lower average wages, wages that appear to be unaffected by both household labor and children. Motherhood may not further depress wages simply because their wages are already too low. These findings suggest that theorizing about the motherhood penalty and gender inequality may have limited applicability to minority women experiencing double jeopardy in the workplace (Greenman and Xie, 2008). This draws attention to the need for more research and theorizing on the economic strains experienced by minority mothers. Fourth, why do White women, specifically married and divorced White women, have residual motherhood penalties? I found that household labor accounts for a portion of the motherhood penalties for these groups, yet they still face significant and unexplained wage penalties with motherhood. Not only does labor within the home have limited explanatory power, but the individual level human capital and occupational factors included in these models are also insufficient. Structural-level explanations, such as employment discrimination and family-(un)friendly work environments, would likely prove useful in explaining residual penalties. Unfortunately, such factors are rarely assessed using quantitative data sets such as the PSID. Future data collection and research can address these gaps by figuring out ways to assess such structural factors as workplace policies and employment discrimination, and match these measurements with data for individual workers. Such data would help researchers evaluate, for example, the actual affect of employment discrimination on motherhood penalties across racial–ethnic groups and marital statuses. The causes of women’s economic inequality vary across racial–ethnic groups, leading to a variety of possible solutions. The finding that time spent in household labor explains a portion of White women’s motherhood penalty underscores the importance of supporting public policies that will lead to more equitable divisions of household labor. Gender-neutral parental leave policies may be part the solution, but the success of such policies is dependent upon cultural support for women’s employment (Budig and Misra, 2008) and the recognition of work-family balance as a public issue rather than a private problem (Gornick and Meyers, 2003). Policy solutions that seek to make household labor more equitable may have a limited impact on Black and Hispanic women, who additionally need policies that tackle racial and ethnic inequality in the workplace. Solutions may include addressing such issues as affirmative action and immigration reform in ways that assure minority women have equal opportunity and compensation in employment. Acknowledgments I would like to thank Elizabeth Cherry, Linda Grant, Maria Paino, Christopher Parrott, Linda Renzulli, Jeremy Reynolds, and the anonymous reviewers at Social Science Research for their helpful comments on earlier drafts of this paper. The collection of data used in this study was partly supported by the National Institutes of Health under Grant Number R01 HD069609 and the National Science Foundation under Award Number 1157698. References Allison, P., 2005. Fixed effects regression methods for longitudinal data using sas. SAS Institute Inc., Cary, NC. Anderson, D., Shapiro, D., 1996. Racial differences in access to high-paying jobs and the wage gap between black and white women. Ind. Labor Relat. Rev. 49, 273–286. Anderson, D.J., Binder, M., Krause, K., 2003. The motherhood wage penalty revisited: experience, heterogeneity, work effort, and work-schedule flexibility. Ind. Labor Relat. Rev. 56, 273–294. Artis, J.E., Pavalko, E.K., 2003. Explaining the decline in women’s household labor: individual change and cohort differences. J. Marr. Fam. 65, 746–761. Bardasi, E., Gornick, J.C., 2008. Working for less? Women’s part-time wage penalties across countries. Fem. Econ. 14, 37–72. Baxter, J., Hewitt, B., Haynes, M., 2008. Life course transitions and housework: marriage, parenthood, and time in housework. J. Marr. Fam. 70, 259–272. Becker, G.S., 1985. Human capital, effort, and the sexual division of labor. J. Organ. Behav. 3, S33–S58. Benard, S., Correll, S., 2010. Normative discrimination and the motherhood penalty. Gender Soc., 24. Bianchi, S.M., Milkie, M.A., Sayer, L.C., Robinson, J.P., 2000. In anyone doing the housework? Trends in the gender division of labor. Soc. Forces 79, 191–228. Bielby, D., Bielby, W.T., 1988. She works hard for the money: Household responsibilities and the allocation of work effort. Am. Jour. of Soc. 93, 1031–1059. Blair-Loy, M., 2005. Competing Devotions: Career and Family Among Women Executives. Harvard University Press, Harvard. Browne, I., Misra, J., 2003. The intersection of race and gender in the labor market. Ann. Rev. Sociol. 29, 487–513. Budig, M.J., England, P., 2001. The wage penalty for motherhood. Am. Sociol. Rev. 66, 204–225. Budig, M.J., Hodges, M.J., 2010. Differences in disadvantage: variation in the motherhood penalty across white women’s earnings distribution. Am. Sociol. Rev. 75, 705–728.

84

H.M. Parrott / Social Science Research 46 (2014) 72–84

Budig, M., Misra, J., 2008. How Care Work Employment Shapes Earnings in a Cross-national Perspective. United Nations Research Institute for Social Development, Geneva. Budig, M.J., Misra, J., Boeckmann, I., 2012. The motherhood penalty in cross-national perspective: the importance of work-family policies and cultural attitudes. Social Polit.: Int. Stud. Gender State Soc. 19, 163–193. Carlin, J.B., Galati, J.C., Royson, P., 2008. A new framework for managing and analyzing multiply imputed data in stata. Stata J. 8, 49–67. Cohen, P.N., 2002. Extended households at work: living arrangements and inequality in single mothers’ employment. Sociol. Forum 17, 445–463. Cohen, P.N., Huffman, M.L., 2003. Individuals, jobs, and labor markets: the devaluation of women’s work. Am. Sociol. Rev. 68, 443–463. Coltrane, S., 2000. Research on household labor: modeling and measuring the social embeddedness of routine family work. J. Marr. Fam. 62, 1208–1233. Coltrane, S., Parke, R.D., Adams, M., 2004. Complexity of father involvement in low-income mexican–american families. Fam. Relat. 53, 179–189. Cooke, L.P., 2007. Policy pathways to gender power: state-level effects on the us division of housework. J. Social Pol. 36, 239–260. Cooksey, E.C., Fondell, M.M., 1996. Spending time with his kids: effects of family structure on fathers’ and children’s lives. J. Marr. Fam. 58, 693–707. Correll, S.J., Benard, S., Paik, I., 2007. Getting a job: is there a motherhood penalty? Am. J. Sociol. 112, 1297–1338. Damaske, S., 2011. For the Family?: How Class and Gender Shape Women’s Work. Oxford University Press, New York. Davis, S.N., Greenstein, T.N., Marks, J.P.G., 2007. Effects of union type on division of household labor. J. Fam. Issues 28, 1246–1272. Duncan, S., Edwards, R., Reynolds, T., 2003. Motherhood, paid work and partnering: values and theories. Work Employ Soc. 17, 309–330. England, P., Budig, M., Folbre, N., 2002. Wages of virtue: the relative pay of care work. Soc. Probl. 49, 455–473. Gangl, M., Ziefle, A., 2009. Motherhood, labor force behavior, and womens careers: an empirical assessment of the wage penalty for motherhood in Britain, Germany, and the United States. Demography 46, 341–369. Glauber, R., 2007. Marriage and the motherhood wage penalty among African Americans, Hispanics, and Whites. J. Marr. Fam. 69, 951–961. Glauber, R., 2008. Race and gender in families and at work – the fatherhood wage premium. Gender Soc. 22, 8–30. Glenn, E.N., 1992. From servitude to service work: historical continuities in the racial division of paid reproductive labor. Signs 18, 1–43. Glenn, E.N., 2000. The social construction and institutionalization of gender and race. An integrative framework. In: Ferree, M.M., Lorber, J., Hess, B.B. (Eds.), Reinvisioning Gender. Rowman Altamira, Walnut Creek, CA. Gornick, J.C., Meyers, M., 2003. Families that Work: Policies for Reconciling Parenthood and Employment. Russell Sage Foundation. Greenman, E., Xie, Y., 2008. Double jeopardy? The interaction of gender and race on earnings in the united states. Soc. Forces 86, 1217–1244. Gupta, N.D., Smith, N., 2002. Children and career interruptions: the family gap in Denmark. Economica 69, 609–629. Halaby, C.N., 2004. Panel models in sociological research: theory into practice. Ann. Rev. Sociol. 30, 507–544. Hall, M., Farkas, George., 2008. Does human capital raise earnings for immigrants in the low-skill labor market? Demography 45, 619–639. Hausman, J.A., 1978. Specification tests in econometrics. Econometrica 46, 1251–1271. Hayes, A.F., 2009. Beyond Baron and Kenny: statistical mediation analysis in the new millennium. Commun. Monogr. 76, 408–420. Hersch, J., Stratton, L.S., 1994. Housework, wages and the division of household time for employed spouses. Econ. Issues Work Fam. 84, 120–125. Hersch, J., Stratton, L.S., 1997. Housework, fixed-effects, and wages of married workers. J. Hum. Resour. 32, 285–307. Hersch, J., Stratton, L.S., 2000. Household specialization and the male marriage premium. Ind. Labor Relat. Rev. 57, 78–94. Huffman, M.L., Cohen, P.N., 2004. Racial wage inequality: job segregation and devaluation across U.S. labor markets. Am. J. Sociol. 109, 902–936. John, D., Shelton, B.A., 1997. The production of gender among black and white women and men: the case of household labor. Sex Roles 36, 171–193. Kamo, Y., Cohen, E.L., 1998. Division of household work between partners: a comparison of black and white couples. J. Compar. Fam. Stud. 29, 131–146. Keene, J.R., Reynolds, J.R., 2005. The job costs of family demands: gender differences in negative family-to-work spillover. J. Fam. Issues 26, 275–299. Kennelly, I., 1999. ‘That single-mother element’: how white employers typify black women. Gender Soc. 13, 168–192. Killewald, A., Gough, M., 2010. Money isn’t everything: wives’ earnings and housework time. Social Sci. Res. 39, 987–1003. Klerman, J.A., Leibowitz, A., 1999. Job continuity among new mothers. Demography 36, 145–155. Kmec, J.A., 2011. Are motherhood penalties and fatherhood bonuses warranted? Comparing pro-work behaviors and conditions of mothers, fathers, and non-parents. Social Sci. Res. 40, 444–459. Krull, J.L., MacKinnon, D.P., 2001. Multilevel modeling of individual and group level mediated effects. Multivar. Behav. Res. 36, 249–277. Kühhirt, M., Ludwig, V., 2012. Domestic work and the wage penalty for motherhood in west germany. J. Marr. Fam. 74, 186–200. Ludenberg, S., Rose, E., 2000. Parenthood and the earnings of married men and women. Lab. Econ. 7, 689–710. Maani, S.A., Cruickshank, A.A., 2010. What is the effect of housework on the market wage, and can it explain the gender wage gap? J. Econ. Surv. 24, 402– 427. Maume, D.J., 1999. Glass ceilings and glass escalators: occupational segregation and race and sex differences in managerial promotions. Work Occup. 26, 483–509. McGuire, G.M., Reskin, B., 1993. Authority hierarchies at work: the impacts of race and sex. Gender Soc. 7, 487–506. McLennan, M.C., 2000. Does household labor impact market wages? Appl. Econ. 32, 1541–1557. Noonan, M.C., 2001. The impact of domestic work on men’s and women’s wages. J. Marr. Fam. 63, 1134–1145. OECD, 2012. Closing the Gender Gap: Act Now. OECD Publishing. Panel Study of Income Dynamics, 1985–2011. Family Public Use Dataset. Produced and Distributed by the Survey Research Center, Institute for Social Research, University of Michigan, Ann Arbor, MI, 2008–2011. Penha-Lopes, V., 2006. ‘‘To cook, sew, to be a man’’: the socialization for competence and black men’s involvement in housework. Sex Roles 54, 261–274. Roschelle, A.R., 1999. Gender, family structure, and social structure: racial ethnic families in the United States. In: Ferree, M.M., Lorber, J., Hess, B.B. (Eds.), Revisioning Gender. Sage Publications, Thousand Oaks, CA. Roscigno, V.J., 2007. The Face of Discrimination: How race and Gender Impact Work and Home Lives. Rowman and Littlefield Publishers, Lanham, MD. Sanchez, L., Thomson, E., 1997. Becoming mothers and fathers: parenthood, gender, and the division of labor. Gender Soc. 11, 747–772. Sarkisian, N., Gerstel, N., 2004. Kin support among blacks and whites: race and family organization. Am. Sociol. Rev. 69, 812–837. Sarkisian, N., Gerena, M., Gerstel, N., 2007. Extended family integration among Euro and Mexican Americans: ethnicity, gender, and class. J. Marr. Fam. 69, 40–54. Sayer, L.C., Fine, L., 2011. Racial–ethnic differences in U.S. married women’s and men’s housework. Soc. Indic. Res. 101, 259–265. Shauman, K.A., 2006. Occupational sex segregation and the earnings of occupations: what causes the link among college-educated workers? Soc. Sci. Res. 35, 577–619. Silver, H., Goldscheider, F., 1994. Flexible work and housework: work and family constraints on women’s domestic labor. Soc. Forces 72, 1103–1119. Stratton, L.S., 2001. Why does more housework lower women’s wages?: testing hypotheses involving job effort and hours flexibility. Social Sci. Quart. 82, 67–76. Uttal, L., 1999. Using kin for child care: embedment in the socioeconomic networks of extended families. J. Marr. Fam. 61, 845–857. Waldfogel, J., 1997. The effect of children on women’s wages. Am. Sociol. Rev. 62, 209–217. Webber, G., Williams, C., 2008. Mothers in ‘‘good’’ and ‘‘bad’’ part-time jobs: different problems, same results. Gender Soc. 22, 752–777. Wight, V.R., Bianchi, S.M., Hunt, B.R., 2013. Explaining racial/ethnic variation in partnered women’s and men’s housework: does one size fit all? J. Fam. Issues 34, 394–427.