A to Z Index  |  FAQs  |  About BLS  |  Contact Us    

Article

January 2019

Racial and ethnic disparities in access to and use of paid family and medical leave: evidence from four nationally representative datasets

Statistics on overall access to and use of various types of paid family and medical leave for the U.S. workforce are widely available. However, much less is known about disparities in paid-leave access and use by race and ethnicity. This article examines this question, using data from four nationally representative surveys—the American Time Use Survey Leave Module, the Annual Social and Economic Supplement to the Current Population Survey, the National Study of the Changing Workforce, and the Survey of Income and Program Participation. The article’s most consistent finding is that Hispanic workers have lower rates of paid-leave access and use than their White non-Hispanic counterparts.

The United States, unlike peer countries, relies primarily on employer policies to supply paid leave for family and medical reasons. The federal Family and Medical Leave Act (FMLA) provides rights to only unpaid leave and only to some employees. Only a few states (California, Rhode Island, New Jersey, and New York) currently have public programs in effect that supply paid family or medical leave; Washington and the District of Columbia have passed legislation that will take effect in the next few years.

In the absence of a public framework for paid family and medical leave, employers are free to choose which types of paid-leave policies to offer, and to which employees. Data from the National Compensation Survey (NCS) indicate that low-paid workers are less likely to have access to paid leave than those with higher pay, but that survey does not provide statistics by race or ethnicity and does not indicate actual use of leave.1 This article provides new evidence about how access to and use of employer-provided paid family and medical leave vary by race and ethnicity.

To answer this question, we obtain data from four nationally representative datasets: the American Time Use Survey (ATUS) Leave Module, the Current Population Survey Annual Social and Economic Supplement (CPS ASEC), the National Study of the Changing Workforce (NSCW), and the Survey of Income and Program Participation (SIPP).

We analyze access to and use of various types of paid leave for four mutually exclusive groups: White non-Hispanics, Black non-Hispanics, Hispanics, and other non-Hispanics (thus, Hispanics are employees of any race who report being Hispanic). Taking advantage of the information in each of the datasets on characteristics that may be correlated with race and ethnicity, we use multivariate regression models to examine the extent to which paid-leave disparities persist after controlling for other demographic characteristics (such as age, gender, marital status, number and age of children, education, and immigrant status). Further models also control for employment characteristics (such as part-time versus full-time status, industry, occupation, union membership, public versus private sector, and state or region). The article focuses primarily on results that are statistically significant, but given the often large standard errors of many estimates, other differences cannot be ruled out.

Overall, we uncover evidence of paid-leave disparities by race and ethnicity, although this heterogeneity is often at least partly explained by differences in other demographic characteristics. Across the various types of leave considered, the most consistent finding is that Hispanic workers have lower rates of paid-leave access than their White non-Hispanic counterparts. These differentials—in access to any paid family or medical leave, as well as to specific types of leave, such as paid parental leave and paid leave to care for a sick family member, for one’s own illness, and for eldercare—are sizable in the raw data. While reduced in models controlling for differences in demographic and employment characteristics, they generally remain statistically significant. Our decomposition analysis indicates that the disparities between Hispanics and White non-Hispanics reflect differences in both endowments and returns to those endowments. The analysis also finds some evidence of differences in paid-leave access between Black non-Hispanics and White non-Hispanics, but these differences are smaller and less often statistically significant.

Background

According to the National Compensation Survey (NCS), just 15 percent of U.S. workers in 2017 had access to employer-sponsored paid family leave (defined as leave to care for a newborn child, an adopted child, a sick child, or a sick adult relative). Low-paid and part-time workers, and those employed in small firms, were less likely than other workers to have access to such leave, but the NCS did not indicate how access to or use of leave varies by race or ethnicity.

Understanding racial and ethnic disparities in access to and use of paid leave is important in light of evidence that mothers from racial and ethnic minority groups, as well as less educated and unmarried mothers, are less likely to take unpaid leave such as that provided by the FMLA.2 National surveys indicate that parents frequently do not take unpaid FMLA leave—even when they need to take time off to care for family members—because they cannot afford it.3 This finding is consistent with evidence that the FMLA has provided health benefits to infants of highly educated women and married women, but not to children of less advantaged mothers.4

The evidence further suggests that paid-leave policies can reduce these disparities. For example, Jenna Stearns’ study of paid temporary disability insurance programs in five U.S. states found that these programs led to improved infant health, with the largest effects accruing to disadvantaged African American and unmarried mothers.5 Research on California’s law on paid family leave also suggests an equalizing impact. Maya Rossin-Slater, Christopher Ruhm, and Jane Waldfogel showed that, for mothers overall, leavetaking more than doubled after the law came into effect, increasing by an average of about 3 weeks.6 Leavetaking increased the most for mothers who were non-White, who were not college graduates, and who were unmarried. These mothers took less leave than their counterparts before the law was enacted and subsequently increased their leavetaking the most (by more than 3 weeks, on average). Estimates for subgroups suggest that Black non-Hispanic mothers saw the largest absolute gain in leavetaking. Their maternity leave increased by 10.6 percentage points (relative to their baseline rate of 2 percent), for a predicted increase of about 6 weeks. The estimated increase for Hispanic mothers was also large—6.2 percentage points (relative to their baseline of 4 percent). These results are somewhat uncertain because of the often large standard errors and the very large effects for Black non-Hispanics. But even if the results are overstated, the evidence strongly suggests that paid leave leads to large increases in leavetaking for non-White mothers.

These results are all the more striking given that many women may not have been fully aware of the California paid-leave law. Eileen Appelbaum and Ruth Milkman found that about half of the California workers surveyed in 2009 and 2010 did not know about the program, and that this lack of knowledge was more common among younger workers, non-Whites, and those with less education, lower household income, or no access to employer-provided paid-leave benefits.7

Do racial and ethnic differences in access to and use of employer-provided paid leave exist among U.S. employees? A handful of studies have examined this question, using disparate data sources.8 For example, Sarah Glynn analyzed data from the American Time Use Survey (ATUS) Leave Module, which asks employees about access to and use of paid leave. She found that Hispanic parents are less likely than White non-Hispanic or Black non-Hispanic parents to have access to paid leave.9 The present article analyzes data from the 2011 ATUS Leave Module as well as information from three other datasets. While not as comprehensive as the ATUS Leave Module, these datasets help our understanding of specific types of leave access and use.

Data

The data for this article come from four nationally representative datasets. Together, these datasets provide the most comprehensive portrait of access to and use of paid family and medical leave in the United States. As summarized in table 1 and discussed below, the datasets are drawn from different periods and provide information on different types of leave access and use. Consequently, the results are not always comparable across datasets.

Table 1. Summary of the four datasets
Data elementATUS Leave ModuleCPS ASECNSCWSIPP Fertility History Module

Year of survey

2011201620082008

Who was surveyed

U.S. wage and salary workersU.S. noninstitutionalized  populationU.S. civilian labor forceU.S. women age 15 to 64

Coverage of family and medical leave access and use

Access to paid and unpaid leave for own illness or medical care, illness or medical care of another family member, childcare, eldercare, and birth or adoption of a childUse of leave for own illness or injury or medical problems, childcare problems, other family or personal obligations, and maternity or paternity obligationsAccess to and use of parental leave after childbirth, and access to paid sick leave and paid time off to care for a sick childUse of paid or unpaid maternity leave, paid or unpaid sick leave, or disability leave

Note: ATUS = American Time Use Survey, CPS ASEC = Current Population Survey Annual Social and Economic Supplement, NSCW = National Study of the Changing Workforce, SIPP = Survey of Income and Program Participation.

Source: U.S. Bureau of Labor Statistics, U.S. Census Bureau, and the Families and Work Institute.

The American Time Use Survey (ATUS) is a nationally representative cross-sectional household survey. It is sponsored by the U.S. Bureau of Labor Statistics and conducted by the U.S. Census Bureau. The ATUS primarily focuses on how individuals in the United States spend their time—on what activities and where. In 2011, the survey also included a Leave Module sponsored by the U.S. Women’s Bureau, aiming to improve our understanding of wage and salary workers’ access to paid and unpaid leave, reasons for taking leave, and ability to adjust work schedules. The 2011 ATUS Leave Module provides comprehensive data on access to leave benefits. It asks, first, whether the respondents receive any paid leave and, then, whether they are able to take paid leave (for own illness, to care for another sick family member, after birth or adoption of a child, for childcare other than for illness, for eldercare other than for illness, for vacation, and for personal reasons).10 In the 2011 ATUS Leave Module, 6,362 respondents answered the question about any paid-leave access.11

The Annual Social and Economic Supplement to the Current Population Survey (CPS ASEC) is a nationally representative survey administered each March. It provides information about employment status and leavetaking (for both men and women) during the week prior to the survey, referred to as the reference week. The CPS ASEC contains questions about the use of any leave, types of leave (as detailed below), and a rich array of demographic information. This article’s analytic sample is from the 2016 CPS ASEC and consists of 79,452 men and women who had a job and were either at work or on leave during the reference week.12 Respondents with a job but absent from work during the reference week were asked about the specific reason for their absence.13 To examine paid parental leave, this article uses the subsample (N = 2,685) of men and women with children under the age of 1. Men and women who reported they were absent from work because of maternity or paternity leave during the reference week and were paid during that absence are categorized as being on paid parental leave.

The National Study of the Changing Workforce (NSCW) is a nationally representative survey of the U.S. workforce. It has been conducted by the Families and Work Institute every 5 to 6 years since 1977. Until 2008, NSCW data were collected by telephone interviews from a sample drawn via random-digit dialing. In 2016, the data collection methodology changed to an online survey with a much smaller sample. For this reason, this analysis focuses primarily on data from the 2008 survey wave (N = 2,687).14 The NSCW provides data on access to and use of several types of leave. Some questions are asked only of a subsample, whereas others are asked of all respondents. First, the survey includes a set of questions about parental leave after childbirth for employees with children under age 6 (i.e., subsample of parents of young children, N = 405). Second, the survey asks all employees with children under age 18 (i.e., subsample of parents of all children, N = 1,052) whether they had access to any paid time off to care for a sick child without using vacation time. Third, the survey asks all employees, regardless of whether they had children (i.e., the full sample, N = 2,687), if they had access to any paid sick leave (paid time off for their own illness).15

Finally, this article uses data on parental leave from the 2008 panel of the Survey of Income and Program Participation (SIPP), which interviews respondents every 4 months. During the survey’s second interview (in spring 2009), women ages 15 to 64 were administered a topical module focused on their fertility history. This module gathered detailed information about leavetaking and employment history before and after the women’s first birth, including year of the first birth, employment during pregnancy, and types of leave taken (e.g., paid, unpaid, and disability).16 This article’s SIPP sample consists of 7,215 women who worked for pay at any time during the pregnancy of their first birth and who completed the topical module.17 These women reported on first births occurring between 1990 and 2009.18

As mentioned earlier, the types of leave covered (as well as the survey’s time period and reference period) vary depending on the dataset. In addition, responses reflect how survey questions are phrased and how leave policies are framed in particular firms. Respondents may not have access to a specific type of leave (e.g., childcare or eldercare leave) but may have access to paid time off that could be used for such purposes. In discussing our results, we note where such measurement issues are likely to affect our estimates.

Methods

Three sets of regression models are estimated. The first controls only for race and ethnicity, providing information on raw differences in paid-leave access and use across the four mutually exclusive groups. White non-Hispanics are the reference group in all models, so Black non-Hispanics, Hispanics, and other non-Hispanics are compared with them.

It is of interest to know to what extent the raw differences might be accounted for (in a statistical sense) by other demographic characteristics that vary by race and ethnicity and influence paid-leave access and use. For example, Hispanic workers might have less frequent access to paid leave than their White non-Hispanic counterparts, because Hispanic workers have lower average levels of education. For this reason, the second set of regression models controls for the following demographic characteristics: education, measured with three categories—high school or less (the reference group), some college, or college degree or more; gender, measured with two categories—male (reference group) or female; marital status, measured with two categories—neither married nor cohabiting (reference group) or married or cohabiting; number and age of children; immigrant status; and respondent age (in years) and age squared.19

It is also useful to understand whether racial and ethnic disparities persist after additionally controlling for employment characteristics that might differ by race or ethnicity and themselves be associated with access to or use of paid leave. For example, minority employees might be concentrated in industries that have relatively high or relatively low rates of paid-leave access or use. The third set of regression models therefore adds the following controls: full-time or part-time status (part-time employees are the reference group), industry, occupation, union status, employment sector, and region or state.20 This third set of models includes the most comprehensive array of covariates. However, it does not necessarily represent the preferred estimates, because the added characteristics may reflect involuntary sorting of Black non-Hispanics or Hispanics into, for example, industries or occupations with lower pay.

The regression models are estimated by ordinary least squares (OLS). Results from logit or probit models are similar, but OLS models are preferred for ease of interpretation.21 Survey participants who responded “don’t know” or “refused” to a question asking about a particular type of leave are not included in the analyses. Although this exclusion leads to slightly different sample sizes depending on the outcome variable, it has little or no effect on the results.

Regression results

Results from the three regression models just described—those comparing Black non-Hispanics and Hispanics with White non-Hispanics—are displayed in figures.22 Each figure indicates to what extent Black non-Hispanic or Hispanic employees are less likely than their White non-Hispanic counterparts to have access to or use the particular type of leave examined. Complete results for the regression models on which the figures are based are provided in the appendix.

Parental leave

The first results presented are for parental leave after childbirth. As discussed earlier, all four datasets provide some information about parental leave access or use, but the specific type of leave measured varies. Access to paid parental leave (e.g., for birth or adoption of a child) is measured in the 2011 ATUS Leave Module. Use of paid maternity leave for the first birth is captured in the SIPP Fertility History Module, and use of paid maternity or paternity leave during the past week is included in the CPS ASEC. Use of paid parental leave for the birth or adoption of the youngest child under age 6 is measured in the NSCW, but because that sample is quite small and we find no significant differences, the results for it are not displayed.

About 40 percent of respondents in the overall ATUS Leave Module sample report that they have access to paid parental leave at their job, but this percentage varies by race and ethnicity. (See figure 1.) The percentage is considerably higher than the share of workers with access to paid family leave as reported in the NCS, a difference reflecting the fact that employees may have access to paid parental leave through other forms of leave. Hispanic and, to a lesser extent, Black non-Hispanic workers are less likely than their White non-Hispanic counterparts to have access to paid leave (model 1). After controlling for demographics (model 2) and employment characteristics (model 3), the differences between Black non-Hispanics and White non-Hispanics are no longer significant, but a 9-percentage-point differential persists between Hispanics and White non-Hispanics.23

The SIPP asks female respondents about maternity leave use. In the full sample, just over 50 percent of women who were working at the time of their first birth said they took at least some paid maternity leave before or after the birth. As shown in figure 2, the raw data indicate that Hispanic workers are 5.5 percentage points less likely than their White non-Hispanic counterparts to report taking paid leave, but this difference changes sign and is no longer significant in models 2 and 3. The pattern is similar for Black non-Hispanic workers, but not significant in any of the models.

The CPS ASEC asks both men and women about leave use in the reference week. Among those with children under the age of 1, about 2.5 percent report being on paid maternity or paternity leave.24 In the raw data, Hispanic workers are 1.6 percentage points less likely than their White non-Hispanic counterparts to report taking paid paternity or maternity leave, but the differences shrink and are no longer significant in models 2 and 3. Paid-leave use is also less common for Black non-Hispanics than for White non-Hispanics, but the disparities are not statistically significant in any of the models. (See figure 3.) The much lower use rates of paid parental leave for the CPS ASEC are attributable to the survey’s relatively short reference period (past week), compared with that of the SIPP (time of first birth).

Leave to care for a sick child or family member

The next type of leave considered is paid leave to care for a sick child or family member. All children are ill from time to time, and about 15 percent of them have special healthcare needs that require their parents to be involved in their medical care.25 Many workers also have responsibility to care for other family members who are ill. Yet not all American workers have leave time specifically set aside to care for a sick child or family member.

In the 2011 ATUS Leave Module, 44 percent of all employees report having access to paid leave for illness or medical care of another family member, similar to the share reporting access to paid parental leave. In the raw data, in a pattern similar to that for parental leave, Hispanics and, to a lesser extent, Black non-Hispanics are significantly less likely to have access to such paid leave. These differences are no longer significant for Black non-Hispanic workers when controlling for individual and employment characteristics (models 2 and 3), but remain significant (with about a 7-percentage-point differential) for Hispanic workers.26 (See figure 4.)

Paid sick leave

The 2011 ATUS Leave Module data reveal that 55 percent of employees report having access to paid sick leave. This percentage is higher than that of employees reporting access to paid parental leave or paid leave to care for a sick family member. The difference reflects the fact that sick leave in the ATUS Leave Module is defined as leave to care for one’s own illness or medical care and, thus, may not include leave to care for a newborn or other family member. As for paid parental and family leave, Hispanic workers are less likely to have access to such leave than White non-Hispanic workers, with about a 6-percentage-point differential in models 2 and 3. No significant differences are observed for Black non-Hispanic workers in any of the models.27 (See figure 5.)

Other types of paid family and medical leave

The 2011 ATUS Leave Module also asks respondents about paid leave for other family responsibilities. About 29 percent of employees say they have access to paid leave for childcare, and around 27 percent say the same for eldercare. These percentages indicate that these benefits are less commonly available than the others considered earlier. Hispanic workers and, to a lesser extent, Black non-Hispanic workers again are less likely than their White non-Hispanic counterparts to have access to these types of leave. The differences in childcare leave are not significant in models 2 and 3 for Hispanics or Black non-Hispanics (see figure 6); however, those for eldercare leave remain significant, indicating a 3-to-4-percentage-point differential for Black non-Hispanic workers and a 7-to-8-percentage-point differential for Hispanic workers (see figure 7).

Any paid leave for family or medical reasons

The 2011 ATUS Leave Module asks about several different types of family and medical leave. While overall rates of leave access vary depending on which specific type of leave is considered, the patterns by race and ethnicity are broadly similar across different types of leave. This can be seen in figure 8, which shows the share of employees with access to any paid leave for family or medical reasons (e.g., any paid parental leave, leave to care for a sick family member, sick leave for own illness, paid leave for childcare, or paid leave for eldercare). Hispanics are less likely than White non-Hispanics to have access to any paid leave for family or medical reasons, a finding consistent with the pattern of results for specific types of leave in the previous figures. For example, in model 3, Hispanic workers are 6 percentage points less likely than their White non-Hispanic counterparts to have any paid family or medical leave. This difference is similar to the differentials between Hispanics and White non-Hispanics in access to specific types of paid family or medical leave in model 3 in the earlier ATUS figures (which showed differentials ranging from 5 to 9 percentage points).

Decomposition results

It is of interest to learn to what extent the significant differences in access to paid leave found between White non-Hispanics and Hispanics result from differences in the “endowments” (i.e., average characteristics) of these two groups, the heterogeneity in the “returns” to those characteristics (i.e., regression coefficients), and the interaction of the two. It is also important to identify which differences in endowments and returns have the greatest impact.

To answer these questions, we conducted an Oaxaca decomposition analysis of the differentials in paid-leave access between Hispanics and White non-Hispanics in the 2011 ATUS Leave Module. The threefold decomposition model divided the outcome disparities into components due to differences in endowments, returns to those endowments, and interactions between them. The endowment effect is the portion of the overall gap explained by differences in characteristics when holding returns to them constant. The returns effect is the portion of the overall gap explained by differences in returns when holding the characteristics constant. The interaction effect is the contribution of the differences in characteristics weighted by the differences in returns. These decompositions are based on statistical correlations and do not necessarily represent causal effects.

The decomposition equation is

where Y denotes access to leave, X the covariates, and β the returns. The subscripts W and H represent White non-Hispanics and Hispanics, respectively. In the equation, the first term () signifies the endowment effect, the second term () the return effect, and the third term () the interaction effect.

The results (summarized in table 2, with full results provided in the appendix) indicate that differences in both endowments and returns play a role in explaining the Hispanic–White non-Hispanic gaps in access to the various types of paid leave examined here. Factors with a positive sign contribute to the gap, and factors with a negative sign reduce it. The statistical significance of the estimates is indicated with asterisks. In general, differences in endowments are most important in explaining the leave-access gaps between Hispanics and White non-Hispanics. They account for 11.2 percentage points of the 24.2-percentage-point difference in access to paid parental leave, 14.5 percentage points of the 23.0-percentage-point difference in access to paid leave to care for a sick child or family member, 17.3 percentage points of the 22.5-percentage-point difference in access to paid sick leave, and 7.7 percentage points of the 18.0-percentage-point difference in access to paid leave for eldercare. Differences in returns also matter, accounting for between 5.4 and 8.9 percentage points of the overall gaps across the four types of leave. Conversely, interactions between endowments and returns play a small or, in some cases, offsetting role.

Table 2. Summary of results from decomposition of Hispanic–White non-Hispanic differences in access to paid leave
VariablePaid parental leavePaid leave to care for a sick child or family memberPaid sick leavePaid leave for eldercare

White non-Hispanic

0.474***0.494***0.590***0.317***

(0.011)(0.011)(0.012)(0.009)

Hispanic

0.232***0.264***0.365***0.137***

(0.020)(0.022)(0.021)(0.016)

Difference

0.242***0.230***0.225***0.180***

(0.022)(0.026)(0.023)(0.018)

Endowments

0.112***0.145***0.173***0.077***

(0.029)(0.035)(0.032)(0.030)

Returns

0.089**0.0600.087**0.054*

(0.037)(0.041)(0.041)(0.030)

Interaction

0.0410.024-0.0350.049

(0.038)(0.044)(0.040)(0.040)

***p < 0.01, **p < 0.05, *p < 0.1

Note: Don't know/refused responses are not included. Standard errors are clustered by state and shown in parentheses.

Source: 2011 American Time Use Survey Leave Module.

With regard to specific endowments (see results in appendix table A-9), differences in immigration or citizenship status are particularly consequential for the gap in leave access between Hispanics and White non-Hispanics. These differences account for 6.5 percentage points of the 24.2-percentage-point gap in access to paid parental leave, 7.1 percentage points of the 23.0-percentage-point gap in access to paid leave to care for a sick child or family member, 9.2 percentage points of the 22.5-percentage-point gap in access to paid sick leave, and 3.4 percentage points of the 18.0-percentage-point gap in access to paid leave for eldercare. Differences in education are also important, accounting for 4.3, 3.6, 3.0, and 2.4 percentage points of these gaps, respectively.

With regard to specific returns, the roles of employment status (part time versus full time) and number and age of children in explaining leave-access gaps between Hispanics and White non-Hispanics are particularly striking. For example, differences in returns to employment status account for 9.4 percentage points of the 24.2-percentage-point gap in access to paid parental leave, 10.1 percentage points of the 23.0-percentage-point gap in access to paid leave to care for a sick child or family member, 11.8 percentage points of the 22.5-percentage-point gap in access to paid sick leave, and 6.9 percentage points of the 18.0-percentage-point gap in access to paid leave for eldercare. Differences in returns to number and age of children account for 5.0, 3.4, 4.2, and 1.9 (not significant) percentage points of these gaps, respectively.28 Some differences in returns to employee age are also large, suggesting that the age profile of access to leave differs between Hispanics and White non-Hispanics, although these differences are mostly not significant.

These results show that differences in leave access between Hispanics and White non-Hispanics are the result of differences in both observable characteristics and returns to those characteristics. Therefore, eliminating the leave-access gaps between the two groups would require policies that eliminate both sources of differences.

Conclusion

Paid family and medical leave is an important benefit for employees and a topic currently high on the policy agenda. Yet relatively little is known about racial and ethnic differences in access to and use of such types of leave. This article provides new evidence on this question, using four nationally representative surveys covering different types of leave, samples, and periods. Together, these surveys point to some common patterns of racial and ethnic differences in leave access and use.

The most consistent finding of the article is that Hispanic workers have lower rates of paid-leave access and use than their White non-Hispanic counterparts. The differentials are sizable in the raw data and, while reduced in models controlling for differences in demographic and employment characteristics, often remain statistically significant. Our decomposition analysis indicates that heterogeneity in both endowments and returns plays a role in accounting for these disparities in paid leave. We also find some evidence of differences between Black non-Hispanic and White non-Hispanic workers, but these differences are smaller and less often statistically significant.

An important topic for future research is the nature and extent of disparities in paid-leave access and use by factors other than race or ethnicity. While such factors are not the focus of this article, the multivariate results (available in the appendix) reveal substantial disparities by characteristics such as education and employment status.

Appendix

Table A-1. Regression results for access to paid parental leave
VariableModel 1Model 2Model 3

Black, non-Hispanic

-0.066**-0.027-0.022

(0.030)(0.029)(0.029)

Hispanic

-0.242***-0.089**-0.090**

(0.026)(0.034)(0.036)

Other, non-Hispanic

-0.0490.005-0.020

(0.055)(0.046)(0.049)

Some college

0.101***0.036

(0.021)(0.022)

College or more

0.235***0.064***

(0.018)(0.024)

Female

0.0010.009

(0.016)(0.017)

Married or cohabiting

0.034*0.022

(0.018)(0.017)

Foreign born, citizen

-0.128***-0.110***

(0.040)(0.037)

Foreign born, noncitizen

-0.203***-0.158***

(0.034)(0.034)

Age

0.042***0.016***

(0.004)(0.004)

(Age)2

-0.000***-0.000***

(0.000)(0.000)

Presence of children under age 6

0.080***0.056***

(0.020)(0.018)

Number of children under age 18

-0.0050.008

(0.009)(0.008)

Full-time status

0.245***

(0.018)

Additional controls

NoNoYes

State fixed effects

NoNoYes

Constant

0.474***-0.494***-0.158

(0.011)(0.067)(0.103)

Observations

5,9225,9225,922

R2

0.0280.1440.263

***p < 0.01, **p < 0.05, *p < 0.1

Note: Coefficients are from OLS models. Don't know/refused responses are not included. Standard errors are clustered by state and shown in parentheses. Reference groups are as follows (in same order as above): White non-Hispanic, high school or less, male, never or previously married, native born, citizen, no children under age 6 in household, and part-time status. Additional controls refer to union membership, industry, occupation, and employment sector. All results are weighted using the 2011 American Time Use Survey (ATUS) Leave Module probability weights.

Source: 2011 ATUS Leave Module.

Table A-2. Regression results for use of paid maternity leave for first birth
VariableModel 1Model 2Model 3

Black, non-Hispanic

-0.050.0210.013

(0.036)(0.033)(0.034)

Hispanic

-0.055**0.0360.014

(0.026)(0.027)(0.023)

Other, non-Hispanic

0.077**0.067**0.026

(0.031)(0.027)(0.020)

Some college

0.048***0.043***

(0.014)(0.013)

College or more

0.118***0.116***

(0.016)(0.015)

Married

0.010.012

(0.013)(0.013)

Age at first birth

0.078***0.058***

(0.009)(0.009)

(Age)2 at first birth

-0.001***-0.001***

(0.000)(0.000)

Full-time status before first birth

0.309***

(0.015)

Birth-year dummies

NoNoYes

State dummies

NoNoYes

Constant

0.519***-0.799***-0.811***

(0.014)(0.124)(0.118)

Observations

7,2157,2157,215

R2

0.0040.0820.142

***p < 0.01, **p < 0.05, *p < 0.1

Note: Coefficients shown are probabilities. Standard errors are clustered by state and shown in parentheses. Reference groups are as follows (in same order as above): White non-Hispanic, high school or less, unmarried, and part-time status. All results are weighted using the person-weight variable (wpfinwgt). Disability leave is included in paid leave.

Source: Wave-2 topical module of the Survey of Income and Program Participation 2008 panel.

Table A-3. Regression results for use of paid maternity or paternity leave in past week
VariableModel 1Model 2Model 3

Black, non-Hispanic

-0.009-0.001-0.001

(0.015)(0.013)(0.015)

Hispanic

-0.016***-0.002-0.005

(0.005)(0.008)(0.009)

Other, non-Hispanic

-0.012-0.009-0.009

(0.018)(0.015)(0.016)

Some college

-0.004-0.001

(0.008)(0.008)

College or more

0.0080.008

(0.007)(0.012)

Female

0.057***0.064***

(0.006)(0.007)

Married

0.022***0.023***

(0.007)(0.007)

Foreign born, citizen

-0.012-0.014*

(0.008)(0.007)

Foreign born, noncitizen

-0.000-0.003

(0.011)(0.013)

Age

0.004***0.003**

(0.001)(0.001)

(Age)2

-0.000***-0.000***

(0.000)(0.000)

Presence of children under age 6

0.009*0.011*

(0.005)(0.005)

Number of children under age 18

-0.004*-0.004*

(0.002)(0.002)

Full-time status

0.023*

(0.012)

Additional controls

NoNoYes

State dummies

NoNoYes

Constant

0.030***-0.099***-0.141***

(0.004)(0.022)(0.021)

Observations

2,6852,6852,685

R2

0.0020.0420.060

***p < 0.01, **p < 0.05, *p < 0.1

Note: Coefficients shown are probabilities. Standard errors are clustered by state and shown in parentheses. Reference groups are as follows (in same order as above): White non-Hispanic, high school or less, male, unmarried, native born, citizen, and part-time status. Additional controls refer to union membership, industry, occupation, and employment sector. All results are weighted using the person-weight variable (wtsupp). Self-employed persons are excluded.

Source: Annual Social and Economic Supplement to the Current Population Survey, administered in March 2016.

Table A-4. Regression results for access to paid leave to care for a sick family member
VariableModel 1Model 2Model 3

Black, non-Hispanic

-0.066**-0.026-0.030

(0.026)(0.024)(0.023)

Hispanic

-0.230***-0.069*-0.073*

(0.028)(0.040)(0.038)

Other, non-Hispanic

-0.0390.0200.002

(0.040)(0.039)(0.037)

Some college

0.099***0.027

(0.022)(0.021)

College or more

0.236***0.048**

(0.017)(0.021)

Female

-0.010-0.009

(0.014)(0.016)

Married or cohabiting

0.043***0.036**

(0.016)(0.014)

Foreign born, citizen

-0.115***-0.094**

(0.043)(0.038)

Foreign born, noncitizen

-0.214***-0.157***

(0.046)(0.041)

Age

0.042***0.013***

(0.004)(0.004)

(Age)2

-0.000***-0.000***

(0.000)(0.000)

Presence of children under age 6

0.080***0.059***

(0.020)(0.017)

Number of children under age 18

-0.0060.009

(0.009)(0.009)

Full-time status

0.262***

(0.023)

Additional controls

NoNoYes

State fixed effects

NoNoYes

Constant

0.494***-0.501***-0.169

(0.010)(0.074)(0.104)

Observations

6,1626,1626,162

R2

0.0250.1430.278

***p < 0.01, **p < 0.05, *p < 0.1

Note: Coefficients are from OLS models. Don't know/refused responses are not included. Standard errors are clustered by state and shown in parentheses. Reference groups are as follows (in same order as above): White non-Hispanic, high school or less, male, never or previously married, native born, citizen, no children under age 6 in household, and part-time status. Additional controls refer to union membership, industry, occupation, and employment sector. All results are weighted using the 2011 American Time Use Survey (ATUS) Leave Module probability weights.

Source: 2011 ATUS Leave Module.

Table A-5. Regression results for access to paid sick leave
VariableModel 1Model 2Model 3

Black, non-Hispanic

-0.041-0.002-0.001

(0.030)(0.027)(0.024)

Hispanic

-0.225***-0.064*-0.061*

(0.025)(0.036)(0.035)

Other, non-Hispanic

-0.0200.0290.009

(0.051)(0.044)(0.045)

Some college

0.095***0.024

(0.023)(0.021)

College or more

0.223***0.030

(0.017)(0.021)

Female

-0.036**-0.024

(0.014)(0.015)

Married or cohabiting

0.0300.025*

(0.018)(0.014)

Foreign born, citizen

-0.067-0.053

(0.043)(0.036)

Foreign born, noncitizen

-0.203***-0.156***

(0.043)(0.037)

Age

0.045***0.012**

(0.005)(0.005)

(Age)2

-0.000***-0.000**

(0.000)(0.000)

Presence of children under age 6

0.070***0.039**

(0.022)(0.018)

Number of children under age 18

-0.018*-0.002

(0.009)(0.009)

Full-time status

0.306***

(0.023)

Additional controls

NoNoYes

State fixed effects

NoNoYes

Constant

0.590***-0.479***-0.185

(0.012)(0.092)(0.117)

Observations

6,3376,3376,337

R2

0.0230.1500.307

***p < 0.01, **p < 0.05, *p < 0.1

Note: Coefficients are from OLS models. Don't know/refused responses are not included. Standard errors are clustered by state and shown in parentheses. Reference groups are as follows (in same order as above): White non-Hispanic, high school or less, male, never or previously married, native born, citizen, no children under age 6 in household, and part-time status. Additional controls refer to union membership, industry, occupation, and employment sector. All results are weighted using the 2011 American Time Use Survey (ATUS) Leave Module probability weights.

Source: 2011 ATUS Leave Module.

Table A-6. Regression results for access to paid leave for childcare
VariableModel 1Model 2Model 3

Black, non-Hispanic

-0.0310.0060.006

(0.023)(0.020)(0.019)

Hispanic

-0.156***-0.050-0.056

(0.022)(0.034)(0.036)

Other, non-Hispanic

0.0380.0660.049

(0.060)(0.049)(0.049)

Some college

0.077***0.024

(0.020)(0.021)

College or more

0.191***0.052**

(0.017)(0.022)

Female

-0.028**-0.024

(0.014)(0.016)

Married or cohabiting

0.062***0.053***

(0.012)(0.012)

Foreign born, citizen

-0.075**-0.049

(0.033)(0.030)

Foreign born, noncitizen

-0.144***-0.098***

(0.039)(0.035)

Age

0.028***0.010***

(0.003)(0.003)

(Age)2

-0.000***-0.000***

(0.000)(0.000)

Presence of children under age 6

0.074***0.061***

(0.019)(0.018)

Number of children under age 18

-0.0000.006

(0.008)(0.008)

Full-time status

0.156***

(0.017)

Additional controls

NoNoYes

State fixed effects

NoNoYes

Constant

0.329***-0.332***-0.014

(0.009)(0.057)(0.083)

Observations

5,9575,9575,957

R2

0.0140.0980.176

***p < 0.01, **p < 0.05, *p < 0.1

Note: Coefficients are from OLS models. Don't know/refused responses are not included. Standard errors are clustered by state and shown in parentheses. Reference groups are as follows (in same order as above): White non-Hispanic, high school or less, male, never or previously married, native born, citizen, no children under age 6 in household, and part-time status. Additional controls refer to union membership, industry, occupation, and employment sector. All results are weighted using the 2011 American Time Use Survey (ATUS) Leave Module probability weights.

Source: 2011 ATUS Leave Module.

Table A-7. Regression results for access to paid leave for eldercare
VariableModel 1Model 2Model 3

Black, non-Hispanic

-0.065***-0.032*-0.039*

(0.021)(0.019)(0.021)

Hispanic

-0.180***-0.070***-0.076***

(0.019)(0.026)(0.025)

Other, non-Hispanic

-0.0300.009-0.005

(0.049)(0.044)(0.046)

Some college

0.053***0.003

(0.017)(0.018)

College or more

0.158***0.020

(0.018)(0.021)

Female

-0.024*-0.018

(0.013)(0.014)

Married or cohabiting

0.047***0.040***

(0.013)(0.012)

Foreign born, citizen

-0.068**-0.039

(0.029)(0.027)

Foreign born, noncitizen

-0.159***-0.116***

(0.026)(0.028)

Age

0.028***0.012***

(0.003)(0.003)

(Age)2

-0.000***-0.000***

(0.000)(0.000)

Presence of children under age 6

0.045**0.035*

(0.020)(0.019)

Number of children under age 18

-0.0050.001

(0.007)(0.007)

Full-time status

0.153***

(0.017)

Additional controls

NoNoYes

State fixed effects

NoNoYes

Constant

0.317***-0.324***-0.012

(0.009)(0.051)(0.080)

Observations

5,9545,9545,954

R2

0.0190.0870.166

***p < 0.01, **p < 0.05, *p < 0.1

Note: Coefficients are from OLS models. Don't know/refused responses are not included. Standard errors are clustered by state and shown in parentheses. Reference groups are as follows (in same order as above): White non-Hispanic, high school or less, male, never or previously married, native born, citizen, no children under age 6 in household, and part-time status. Additional controls refer to union membership, industry, occupation, and employment sector. All results are weighted using the 2011 American Time Use Survey (ATUS) Leave Module probability weights.

Source: 2011 ATUS Leave Module.

Table A-8. Regression results for access to any paid family or medical leave
VariableModel 1Model 2Model 3

Black, non-Hispanic

-0.040-0.003-0.002

(0.028)(0.027)(0.024)

Hispanic

-0.220***-0.064*-0.060*

(0.022)(0.033)(0.033)

Other, non-Hispanic

-0.0190.0250.003

(0.051)(0.043)(0.045)

Some college

0.098***0.029

(0.024)(0.022)

College or more

0.218***0.033

(0.018)(0.021)

Female

-0.036**-0.020

(0.014)(0.015)

Married or cohabiting

0.0270.020

(0.018)(0.014)

Foreign born, citizen

-0.047-0.034

(0.041)(0.035)

Foreign born, noncitizen

-0.207***-0.160***

(0.043)(0.037)

Age

0.046***0.012**

(0.005)(0.005)

(Age)2

-0.000***-0.000**

(0.000)(0.000)

Presence of children under age 6

0.069***0.038**

(0.021)(0.017)

Number of children under age 18

-0.016*0.000

(0.009)(0.008)

Full-time status

0.313***

(0.023)

Additional controls

NoNoYes

State fixed effects

NoNoYes

Constant

0.599***-0.477***-0.210

(0.011)(0.092)(0.126)

Observations

6,3516,3516,351

R2

0.0220.1480.307

***p < 0.01, **p < 0.05, *p < 0.1

Note: Coefficients are from OLS models. Don't know/refused responses are not included. Standard errors are clustered by state and shown in parentheses. Reference groups are as follows (in same order as above): White non-Hispanic, high school or less, male, never or previously married, native born, citizen, no children under age 6 in household, and part-time status. Additional controls refer to union membership, industry, occupation, and employment sector. All results are weighted using the 2011 American Time Use Survey (ATUS) Leave Module probability weights.

Source: 2011 ATUS Leave Module.

Table A-9. Decomposition of differences in paid-leave access between Hispanics and White non-Hispanics
VariablePaid parental leavePaid leave to care for a sick child or family memberPaid sick leavePaid leave for eldercare

A. Overall

White non-Hispanic

0.474***0.494***0.590***0.317***

(0.011)(0.011)(0.012)(0.009)

Hispanic

0.232***0.264***0.365***0.137***

(0.020)(0.022)(0.021)(0.016)

Difference

0.242***0.230***0.225***0.180***

(0.022)(0.026)(0.023)(0.018)

Endowments

0.112***0.145***0.173***0.077***

(0.029)(0.035)(0.032)(0.030)

Returns

0.089**0.0600.087**0.054*

(0.037)(0.041)(0.041)(0.030)

Interaction

0.0410.024-0.0350.049

(0.038)(0.044)(0.040)(0.040)

B. The role of specific factors

Education

-0.01-0.022-0.028-0.03

Endowments

0.043***0.036***0.030**0.024**

(0.008)(0.013)(0.012)(0.010)

Returns

-0.023**-0.027-0.030**-0.028**

(0.011)(0.020)(0.015)(0.013)

Interactions

-0.030***-0.031**-0.028**-0.026**

(0.010)(0.015)(0.012)(0.012)

Gender

-0.026-0.01-0.017-0.028

Endowments

0.006*0.0010.0020.002

(0.003)(0.003)(0.003)(0.002)

Returns

-0.027*-0.009-0.016-0.026**

(0.016)(0.015)(0.016)(0.012)

Interactions

-0.005-0.002-0.003-0.004*

(0.003)(0.003)(0.003)(0.002)

Marital status

0.0040.0340.0330.011

Endowments

0.000-0.002-0.0030.002

(0.002)(0.003)(0.003)(0.002)

Returns

0.0030.0310.0310.008

(0.017)(0.020)(0.021)(0.016)

Interactions

0.0010.0050.0050.001

(0.003)(0.004)(0.003)(0.003)

Immigration/citizenship status

0.0640.0720.0960.032

Endowments

0.065***0.071***0.092***0.034**

(0.024)(0.025)(0.024)(0.016)

Returns

-0.013-0.0080.048-0.033

(0.038)(0.036)(0.039)(0.026)

Interactions

0.0120.009-0.0440.031

(0.035)(0.034)(0.036)(0.024)

Age

0.3790.2310.3110.058

Endowments

-0.0030.0070.016**-0.004

(0.007)(0.007)(0.007)(0.004)

Returns

0.376*0.2220.2950.053

(0.197)(0.188)(0.193)(0.117)

Interactions

0.0060.0020.0000.009**

(0.007)(0.007)(0.007)(0.004)

Number and age of children

0.0360.0240.0360.012

Endowments

0.0050.0030.0100.001

(0.006)(0.006)(0.006)(0.003)

Returns

0.050**0.034**0.042**0.019

(0.020)(0.017)(0.018)(0.013)

Interactions

-0.019**-0.013*-0.016**-0.008

(0.008)(0.007)(0.007)(0.005)

Full-time versus part-time status

0.0950.1050.1220.07

Endowments

0.0010.0020.0020.000

(0.004)(0.004)(0.005)(0.002)

Returns

0.094***0.101***0.118***0.069***

(0.033)(0.033)(0.033)(0.020)

Interactions

0.0000.0020.0020.001

(0.003)(0.003)(0.004)(0.002)

Industry

-0.064-0.045-0.037-0.023

Endowments

0.014**0.011**0.011**0.004

(0.007)(0.005)(0.005)(0.005)

Returns

-0.067*-0.051-0.043-0.026

(0.038)(0.032)(0.034)(0.037)

Interactions

-0.011-0.005-0.005-0.001

(0.007)(0.005)(0.005)(0.005)

Occupation

-0.0030.0560.0560.019

Endowments

0.024**0.042***0.042***0.032***

(0.011)(0.013)(0.012)(0.008)

Returns

-0.0500.0120.012-0.023

(0.034)(0.045)(0.041)(0.033)

Interactions

0.023**0.0020.0020.010

(0.011)(0.013)(0.012)(0.010)

Union

-0.022-0.051-0.0660.009

Endowments

0.009**0.008**0.015**0.003

(0.005)(0.004)(0.006)(0.003)

Returns

-0.004-0.0000.0130.002

(0.009)(0.011)(0.011)(0.009)

Interactions

0.0040.0040.0000.004

(0.003)(0.003)(0.005)(0.004)

Sector

0.0180.0160.0080.001

Endowments

-0.0080.0040.0040.003

(0.005)(0.007)(0.007)(0.008)

Returns

0.014*0.0060.001-0.002

(0.008)(0.010)(0.008)(0.013)

Interactions

0.012**0.0060.003-0.000

(0.006)(0.006)(0.006)(0.008)

Region

0.0180.0110.0040.006

Endowments

-0.044**-0.039**-0.046**-0.025**

(0.018)(0.018)(0.018)(0.012)

Returns

0.1190.1040.0530.020

(0.093)(0.088)(0.100)(0.093)

Interactions

0.048**0.044**0.049***0.033**

(0.021)(0.020)(0.019)(0.014)

Observations

4,8104,9905,1414,837

***p < 0.01, **p < 0.05, *p < 0.1

Note: Don't know/refused responses are not included. Standard errors are clustered by state and shown in parentheses.

Source: 2011 American Time Use Survey Leave Module.

ACKNOWLEDGMENTS: The authors are grateful to Pivotal Ventures for funding support and to Ipshita Pal for assistance with the NSCW data.

Suggested citation:

Ann P. Bartel, Soohyun Kim, Jaehyun Nam, Maya Rossin-Slater, Christopher Ruhm, and Jane Waldfogel, "Racial and ethnic disparities in access to and use of paid family and medical leave: evidence from four nationally representative datasets," Monthly Labor Review, U.S. Bureau of Labor Statistics, January 2019, https://doi.org/10.21916/mlr.2019.2.

Notes


1 National Compensation Survey: employee benefits in the United States, March 2017, Bulletin 2787 (U.S. Bureau of Labor Statistics, September 2017), https://www.bls.gov/ncs/ebs/benefits/2017/ebbl0061.pdf.

2 Wen-Jui Han and Jane Waldfogel, “Parental leave: the impact of recent legislation on parents’ leave-taking,” Demography, vol. 40, no. 1, February 2003, pp. 191–200; Wen-Jui Han, Christopher J. Ruhm, and Jane Waldfogel, “Parental leave policies and parents’ employment and leave-taking,” Journal of Policy Analysis and Management, vol. 28, no. 1, winter 2009, pp. 29–54; and Jane Waldfogel, “Family leave coverage in the 1990s,” Monthly Labor Review, October 1999, pp. 13–21, https://www.bls.gov/opub/mlr/1999/10/art2full.pdf.

3 A workable balance: report to Congress on family and medical leave policies (Washington, DC: U.S. Commission on Family and Medical Leave, 1996); David Cantor, Jane Waldfogel, Jeffrey Kerwin, Mareena McKinley Wright, Kerry Levin, John Rauch, Tracey Hagerty, and Martha Stapleton Kudela, Balancing the needs of families and employers: family and medical leave surveys (Rockville, MD: Westat, 2001); Jacob Klerman, Kelly Daley, and Alyssa Pozniak, Family and medical leave in 2012: technical report, prepared for the U.S. Department of Labor (Cambridge, MA: Abt Associates, 2014), http://www.dol.gov/asp/evaluation/fmla/FMLA-2012-Technical-Report.pdf; and Jane Waldfogel, “Family and medical leave: evidence from the 2000 surveys,” Monthly Labor Review, September 2001, pp. 17–23, https://www.bls.gov/opub/mlr/2001/09/art2full.pdf.

4 Maya Rossin, “The effects of maternity leave on children’s birth and infant health outcomes in the United States,” Journal of Health Economics, vol. 30, no. 2, March 2011, pp. 221–239.

5 Jenna Stearns, “The effects of paid maternity leave: evidence from Temporary Disability Insurance,” Journal of Health Economics, vol. 43, 2015, pp. 85–102.

6 Maya Rossin-Slater, Christopher J. Ruhm, and Jane Waldfogel, “The effects of California’s paid family leave program on mothers’ leave-taking and subsequent labor market outcomes,” Journal of Policy Analysis and Management, vol. 32, no. 2, spring 2013, pp. 224–245.

7 Eileen Appelbaum and Ruth Milkman, Leaves that pay: work and employer experiences with paid leave in California (Washington, DC: Center for Economic and Policy Research, 2011).

8 See Barbara Gault, Heidi Hartmann, Ariane Hegewisch, Jessica Milli, and Lindsey Reichlin, Paid parental leave in the United States: what the data tell us about access, usage, and economic and health benefits (Washington, DC: Institute for Women’s Policy Research, 2014).

9 See Sarah J. Glynn, Working parents’ lack of access to paid leave and workplace flexibility (Washington, DC: Center for American Progress, 2012), https://www.americanprogress.org/wp-content/uploads/2012/11/GlynnWorkingParents-1.pdf; and Sarah J. Glynn and Jane Farrell, Latinos least likely to have paid leave or workplace flexibility (Washington, DC: Center for American Progress, 2012), https://cdn.americanprogress.org/wp-content/uploads/2012/11/GlynnLatinosPaidLeave1.pdf.

10 To measure any paid leave, the 2011 ATUS Leave Module used this question: “Thinking about the job where you worked the most hours last week, do you receive paid leave on your job?” It then asked, for each type of leave, the following question: “I’m going to read you a list of reasons why you might have to miss work. For each reason, please tell me if you are able to take paid leave in your current/main job.” The survey also collects information about “paid time off” policies, but this article does not use that information because it does not specify which types of leave are covered.

11 In the analyses using this dataset and the others, the self-employed are excluded.

12 Those who were laid off, unemployed, or out of the labor force are not included in the sample. As mentioned earlier, the sample also excludes the self-employed.

13 The 2016 CPS ASEC used this question: “What was the main reason…was absent from work last week?” Possible responses include the following: vacation/personal days, own illness/injury/medical problems, childcare problems, other family/personal obligation, maternity/paternity leave, labor dispute, weather affected job, school/training, civic/military duty, and other. On the basis of this information, respondents are categorized as being on any paid leave if they were absent for reasons including vacation/personal days, own illness/injury/medical problems, childcare problems, other family/personal obligation, maternity/paternity leave, or other reasons, but not including labor dispute, weather affected job, school/training, or civic/military duty.

14 All analyses are weighted with the use of the appropriate sample weights. Supplemental models using data from the 2016 wave (N = 1,423) were also estimated. Data from the two waves could not be combined because of differences in survey methodology and the lack of comparable weights.

15 The 2008 NSCW used the questions “Did you take time off work after your child’s birth or adoption, or were you not allowed any time off?” for paid parental leave, “Are you allowed to take at least five days off per year to care for a sick child without losing pay, without using vacation days, AND without having to make up some other reason for your absence?” for paid leave to care for a sick child, and “Are you allowed at least five days per year of paid time off for personal illness, or not?” for paid sick leave. The NSCW also asks whether employees had any paid vacation time and, if yes, the number of vacation days available, the number of vacation days used (if any used), and access to any paid holidays.

16 The 2008 SIPP Fertility History Module used the following questions to measure the use of leave before and after the childbirth, respectively: “In order for…to stop working before…first child was born, was…on paid maternity leave/unpaid maternity leave/paid sick leave/unpaid sick leave/disability leave?” and “What about after her child was born, and up to the time the baby was 12 weeks old? What types of leave, if any, did she use then? Anything else?”

17 Women who were self-employed before the first birth are not included.

18 This is the most recent SIPP for which this analysis can be conducted. Following a redesign, the SIPP no longer collects this retrospective information on maternity leavetaking at first birth.

19 The specific controls included varied by data availability. For example, age of children was replaced with the presence of a child under a certain age in the ATUS and CPS ASEC models. The information on children and immigrant status was not included in the NSCW models. Gender was not included in the SIPP models, which used women’s information only.

20 The NSCW model controlled for region instead of state. The SIPP model controlled for full-time employment status before first birth, state, and birth year. A small number of cases were missing data for employment characteristics. To avoid dropping these cases, we created and included dummy variables for missing items.

21 See, for example, Joshua Angrist, “Estimation of limited dependent variable models with dummy endogenous regressors: simple strategies for empirical practice,” Journal of Business and Economic Statistics, vol. 29, no. 1, 2001, pp. 2–15; and Steven B. Caudill, “An advantage of the linear probability model over probit or logit,” Oxford Bulletin of Economics and Statistics, vol. 50, no. 4, 1988, pp. 425–427.

22 Other non-Hispanic employees are also included in the models and controlled for separately, but results for that group are not reported because of small sample sizes.

23 In the smaller NSCW sample (not shown), Hispanic workers are less likely than White non-Hispanic workers to report access across all three models, but the differences are not significant. Black non-Hispanic workers are slightly more likely than White non-Hispanic workers to report access (but not significantly so in models 2 and 3).

24 Rates of leave use are relatively low because data are collected for only a single reference week, which may occur considerably after the birth (when most leaves have been completed). Conversely, the other data sources refer to leaves at any point during the period surrounding childbirth and so report much higher rate of leavetaking.

25 Mark A. Schuster, Paul J. Chung, and Katherine D. Vestal, “Children with health issues,” The Future of Children, vol. 21, no. 2, September 2011, pp. 91–116.

26 The NSCW asked specifically about paid leave to care for a sick child, but the data did not reveal any statistically significant differences by race or ethnicity.

27 There are no significant differences in access to paid sick leave in the NSCW.

28 We conducted a similar decomposition analysis of differentials between White non-Hispanics and Black non-Hispanics. While less pronounced, such differentials exist for some types of paid leave. The results (available on request) indicate that differences in endowments usually play a larger role than differences in returns, and that among the endowments, differences in marital status, occupation, and education tend to be most important.

article image
About the Author

Ann P. Bartel
apb2@gsb.columbia.edu

Ann P. Bartel is the Merrill Lynch Professor of Workforce Transformation at the Columbia Business School, Columbia University.

Soohyun Kim
soohyun.kim@columbia.edu

Soohyun Kim is a doctoral student at the Columbia School of Social Work, Columbia University.

Jaehyun Nam
jn2375@gmail.com

Jaehyun Nam is an assistant professor at Pusan National University.

Maya Rossin-Slater
mrossin@stanford.edu

Maya Rossin-Slater is an assistant professor in the Department of Health Research and Policy at the Stanford University School of Medicine.

Christopher Ruhm
ruhm@virginia.edu

Christopher Ruhm is a professor of public policy and economics at the University of Virginia.

Jane Waldfogel
jw205@columbia.edu

Jane Waldfogel is the Compton Foundation Centennial Professor for the Prevention of Children’s and Youth Problems at the Columbia School of Social Work, Columbia University.

close or Esc Key