Among Low-Income Women In San Francisco, Low Awareness Of Paid Parental Leave Benefits Inhibits Take-Up

Paid family leave policies have the potential to reduce health disparities, yet access to paid leave remains limited and unevenly distributed in the United States. Using California administrative claims data, we examined the impact of the San Francisco Paid Parental Leave Ordinance, the first in the US to provide parental leave with full pay. We found that the law increased parental leave uptake in San Francisco by 13 percent among fathers, but there was little change in leave among mothers. Data from a survey of mothers suggest that the limited impact may be partly a result of low understanding of benefits. Lower-income mothers reported even less knowledge of their maternity leave benefits than other mothers, and fewer than 2 percent of lower-income mothers had accurate information about the policy. The San Francisco policy also excludes small employers, which further limits its reach among low-income workers. A simpler universal policy may be more effective in expanding parental leave among vulnerable workers. 9 Evidence supporting these beneficial health outcomes is

P aid family leave policies have the potential to reduce persistent health disparities by enabling families to take time to care for themselves and their families without risking their jobs or paychecks. Robust evidence supports the connection between access to paid leave for new parents and a range of health outcomes, including increased breast-feeding; 1,2 fewer low-birthweight and small-for-gestationalage births; 3,4 decreased infant hospitalizations; 5 decreased infant mortality rates; 6 reduced likelihood of obesity, attention deficit hyperactivity disorder, hearing problems, and ear infections; 7 and improved maternal mental health, 8 particularly in lower-income households. 9 Evidence supporting these beneficial health outcomes is summarized elsewhere. 10 Despite this evidence, access to paid leave remains limited and unevenly distributed in the US. 11 Although 34 percent of US workers in the highest wage decile have access to paid family leave through their employers, just 6 percent of workers in the lowest wage decile do. 11 State-level paid family leave policies have increased maternity leave by three to five weeks, with the largest gains seen among less advantaged mothers 12,13 and with leave-taking increased by two to three days among fathers. 13 Nevertheless, leave-taking remains limited among lower-income workers, 14 possibly as a result of low awareness of relevant policies. A survey of California workers who had recently experienced a qualifying event (for example, becoming a parent or having a close family member become seriously ill) found that just 49 percent of respondents were aware of the state's Paid Family Leave program five years after the policy went into effect, and awareness was even lower among low-wage workers (38 percent), immigrants (34 percent), Latinos (34 percent), and workers with less than a high school diploma (21 percent). 15 More recent qualitative studies in California and elsewhere provide additional evidence of low awareness; even among those who have heard of the state-level policies, there was often a misunderstanding of key policy elements, including eligibility. 16,17 Using a quasi-experimental difference-indifferences analysis, our study examined how a pathbreaking fully paid parental leave policy, the San Francisco Paid Parental Leave Ordinance (PPLO), affected the probability and duration of parental leave.We used California administrative data on paid leave claiming, comparing changes over time in San Francisco with those in comparison counties. We also analyzed a survey of mothers who gave birth in the San Francisco Bay Area in 2016 or 2017 to examine the remaining barriers to leave-taking, including awareness and knowledge of parental leave benefits. Finally, we used the survey data to estimate the PPLO's reach-that is, the proportion of mothers who were eligible to benefit from the policy, given that it excluded small and informalsector employers. These complementary analyses indicate the actual relative to potential reach of the policy, particularly vis-à-vis the goal of expanding leave among vulnerable parents.

Policy Context
The US is the only country in the Organization for Economic Cooperation and Development that lacks any mandated paid leave for new mothers, and one of two without any paid leave for fathers. 18 Instead, workers in most states rely on their employers to provide leave voluntarily, resulting in socioeconomic disparities in access to both paid and unpaid leave. 11 State and local governments have increasingly stepped into this void: Eight states and the District of Columbia have passed paid family and medical leave laws, and dozens of state and municipal governments offer paid leave to their own employees. 19 California's Paid Family Leave program, enacted in 2002, provides partial wage replacement for up to six weeks of leave to care for one's own serious illness, care for family, or bond with a new child. In addition, California's State Disability Insurance program provides six to eight weeks of partially paid postpartum leave for physical recovery from childbirth. 20 Together, these programs provide more than three months of partially paid leave to birth parents, although qualitative evidence suggests that some women (and their employers) confuse these programs, potentially preventing them from taking advantage of all available leave. 21 Furthermore, low-income parents are less likely to be able to afford parental leave when it is offered at only partial pay.
In 2017 San Francisco went beyond the existing statewide programs to implement the first fully paid parental leave policy in the US. The PPLO mandates that covered employers provide supplemental wage replacement for the six weeks of parental leave that are provided by the Paid Family Leave program, bringing workers' wages up to 100 percent of their gross weekly wage, subject to a cap ($1,173 in 2017; exhibit 1). To receive PPLO benefits, employees must first apply for Paid Family Leave benefits through the state and then submit a separate form to their employers. Employers then pay employees their supplemental wages directly. Online appendix exhibit A1 provides details of the PPLO and related paid leave polices in California. 22 Two eligibility restrictions are notable. First, the PPLO covers only private-sector employers that have employees who work in San Francisco and at least twenty employees worldwide. This is more restrictive than the State Disability Insurance and Paid Family Leave programs, which also cover firms with fewer than twenty employees. Second, employees must meet minimum job tenure and San Francisco-based work hours requirements, in addition to being eligible for California Paid Family Leave benefits. Together these restrictions may disproportionately limit eligibility among lower-income workers, potentially blunting the impact of the PPLO on those who are least able to take partially paid leave.

Study Data And Methods
We drew on two complementary data sources: administrative data to estimate the effects of the Paid Parental Leave Ordinance on leave-taking and survey data to help understand the remaining barriers to leave-taking.
Administrative Data And Analysis The first data source is administrative records obtained by request from the California Employment Development Department, which processes claims for the state's Paid Family Leave program. Because parental leave beneficiaries must submit claims for Paid Family Leave through the state to apply for PPLO benefits, Employment Development Department claims should reflect changes in response to the PPLO. These data include the number of claimants and average parental leave claim duration (in weeks) separately for men and women for each month from January 2010 through June 2018 in San Francisco and in five comparison surrounding Bay Area counties not subject to the PPLO (Alameda, Contra Costa, Marin, San Mateo, and Santa Clara). To calculate the total weeks of leave taken, we multiplied the number of claimants by the average duration of claims within each month-year by county by claimant sex combination. We aggregated data to the quarter level to smooth monthly variation.
We analyzed the count of Employment Development Department parental leave claimants by sex and quarter from 2010 through the second quarter of 2018, comparing San Francisco residents (data are not available by employment county), other San Francisco Bay Area residents, and residents of the rest of the state. We first presented the data graphically, norming proportionately to that area's number of claimants in the fourth quarter of 2016 before the PPLO went into effect. We then quantified the magnitude of changes in counts of number of claimants and total weeks of leave taken in San Francisco after the implementation of the PPLO, using Poisson count data regressions (negative binomial regressions yield virtually the same results). We used count data rather than linear regression models because they most appropriately allow inference on these underlying counts and are naturally interpreted as proportionate differences. We used a quasi-experimental difference-indifferences conceptual approach, controlling for San Francisco residence, time dummies (quarters from 2010 to 2018), and a San Franciscospecific linear time trend. A post-PPLO San Participants were invited by mail to complete , for a further six weeks of leave to bond with a new child to follow the partially paid six weeks of leave covered by the California SDI program. The statewide SDI/PFL programs were amended in 2018 to increase the wage replacement rate from 55 percent to 60-70 percent, depending on wages, for the full twelve weeks, thus decreasing the employer share in San Francisco from 45 percent to 30-40 percent for the last six weeks of leave. "Higher wage" includes workers earning more than a third of statewide average weekly wages; "lower wage" includes workers earning below that threshold.
the survey online and were mailed a paper survey to complete if they did not initially respond. The first survey wave for women who gave birth in 2016 was conducted between December 2017 and December 2018. The second survey wave for women who gave birth in 2017 was conducted between November 2018 and May 2019. Our overall sample includes 1,304 mothers who either were employed or had an employed partner during their pregnancies and who could complete the survey in either English or Spanish (20.4 percent response rate; data not shown) (see appendix B1 for more detail on the representativeness of the sample). 22 We measured both mother-reported access to and duration of mothers' and partners' paid leave. We use these data to help interpret the Employment Development Department administrative estimates of proportionate leave increases, allowing them to be converted into changing population prevalence of leave-taking (appendix B2). 22 We do not report difference-indifferences estimates of leave-taking directly from the survey data here, as the survey sample sizes were not sufficient to yield estimates with confidence intervals small enough to be informative. See appendix B3 and appendix exhibits A2 and A3 for more details and results. 22 The primary outcomes of interest from our survey data included whether respondents understood the maternity leave benefits available to them, the extent to which employers were helpful in making sure they knew about benefits, and other sources of information about benefits. We determined knowledge of paid parental leave benefits among San Francisco employed mothers, comparing Medicaid-covered and non-Medicaid-covered respondents (as a proxy for income), and tracking the changes pre-versus post-PPLO (after extensive post-PPLO education efforts). We also assessed familiarity specifically with PPLO benefits post-PPLO, again comparing Medicaid-covered with non-Medicaid-covered respondents employed in San Francisco.
We also used data from the Bay Area Parental Leave Survey of Mothers to assess PPLO reach. We classified mothers as eligible for the PPLO if they gave birth in 2017 and were employed in San Francisco in a covered job (that is, employed at least eight hours a week for at least six months before childbirth, by a private-sector employer that had at least twenty employees).We also coded eligibility for the mother's live-in partner, who would be eligible for up to six weeks of PPLO parental leave within twelve months of the birth if they were the baby's parent and worked in San Francisco in a covered job in 2017. We then assessed the share of working parents eligible for PPLO benefits by income level.
For all survey data analyses, we incorporated probability weights to account for nonresponse and oversampling of San Francisco residents, Spanish speakers (proxied by mother's immigration from a Spanish-speaking country), and lowincome women (proxied by Medicaid status). All analyses were conducted in Stata, version 14.2. Study procedures were approved by the California Health and Human Services Agency's Committee for the Protection of Human Subjects.
Limitations This study had several limitations. The Employment Development Department administrative data to which we were given access allow precise estimation, given that they report on the universe of claimants, but as they are based on claimant residence, they do not contain pertinent information on claimant employer (firm size and county) to precisely identify PPLO eligibility. County of residence and county of employment are highly correlated in the Bay Area, but this misclassification could still bias downward estimates from administrative data. In our Bay Area Parental Leave Survey of Mothers sample, 76 percent of employed women who gave birth in San Francisco were also employed in San Francisco; 84 percent of employed women who gave birth outside San Francisco were employed outside San Francisco.
We sampled only respondents who gave birth in the San Francisco Bay Area; results might not be generalizable to parents who gave birth in regions with a different mix of demographic and labor-market characteristics.

Study Results
Effect On Parental Leave-Taking Employment Development Department administrative data on the number of male parental leave claimants showed steady upward trends in the pre-Paid Parental Leave Ordinance years, with similar trend increases seen in San Francisco, the surrounding counties, and the rest of the state (exhibit 2). In the post-PPLO period starting January 2017, however, the data indicate an increase in San Francisco male claimants compared with the prior trend; other regions did not show similar increases. Poisson regression model estimates indicate a 13.3 percent (standard error: 3.0) increase in San Francisco claimants post-PPLO beyond what would have been predicted from trend increases in other Bay Area counties (appendix exhibit A6).
For women, in contrast, the administrative data show no similar increase in parental leave claimants post-PPLO (appendix exhibit A5). 22 Of note, 89 percent of mothers in covered jobs were already taking at least twelve weeks of leave, as reported in Bay Area Parental Leave Survey of Mothers data, whereas only 43 percent of partners working in San Francisco pre-PPLO were reported to have taken a government-paid leave (appendix exhibits A2a and A2b). 22 Thus, there was considerably more opportunity for the PPLO to increase claimants among men. A detailed analysis of PPLO take-up among men is in appendix B2. 22 We next estimated effects on total weeks claimed. Among men, the PPLO was associated with a 15.6 percent (SE: 1.4) increase in total weeks claimed versus other Bay Area counties (appendix exhibit A7). 22 Among women, there was only a small effect of the PPLO on total weeks of leave, similar to the small increases seen in the number of female claimants.
Disparities In Knowledge Of Maternity Leave To understand why the PPLO did not increase leave-taking more dramatically, exhibit 3 uses the Bay Area Parental Leave Survey of Mothers data to compare knowledge of available maternity leave benefits between Medicaidcovered and non-Medicaid-covered women. Before the PPLO, 45 percent of Medicaid-covered respondents reported that they understood their maternity leave benefits "moderately well," "very well," or "extremely well" compared with 77 percent of non-Medicaid-covered respondents (p < 0:01). This gap narrowed after the PPLO was implemented to 54 percent versus 69 percent, respectively, although the differential change over time was not significant.
We observed similar differences in the proportion of respondents who reported that their em-ployers helped them understand which maternity leave benefits were available: 48 percent of Medicaid-covered respondents and 83 percent of non-Medicaid-covered respondents before the PPLO (p < 0:001). This gap narrowed slightly after implementation of the PPLO to 58 percent versus 83 percent, respectively (p < 0:01). Disparities also persisted in which resources respondents identified as their main source of information about maternity leave benefits: Before implementation of the PPLO, Medicaidcovered respondents were significantly less likely than their non-Medicaid-covered counterparts were to report receiving information from their employer (48-percentage-point difference; p < 0:001), friends and family (17-percentagepoint difference; p < 0:05), the government (7percentage-point difference; p < 0:01), or a nonprofit or legal aid group (4-percentage-point difference; p < 0:01), and were more likely to report that nobody helped them (41-percentage-point difference; p < 0:001) (appendix exhibit A8a). 22 In fact, pre-PPLO, the most commonly cited source among Medicaid-covered-respondents was "nobody" (51 percent). Post-PPLO, Medicaid-covered respondents continued to report more frequently that nobody helped them (30 percent versus 12 percent; p < 0:01). However, this difference decreased significantly over time (p < 0:05), and post-PPLO, it was no longer the modal response. Thirty-five percent reported friends and family and 29 percent reported health care providers as a main source of information (appendix exhibit A8b). Awareness of the PPLO also differed by Medicaid status: Although 61.6 percent of non-Medicaid-covered women employed in San Francisco had heard of the PPLO, just 9.7 percent of Medicaid-covered women had (p < 0:001) (appendix exhibit A9). 22 Moreover, only 1.4 percent of Medicaid-covered women had accurate knowledge of the PPLO (that is, knew that it includes fathers and ensures full pay for six weeks) compared with 31.5 percent of non-Medicaidcovered women.
Limited Reach Of The Ordinance Last, we explore the extent to which the PPLO could better reach low-income workers (as proxied by Medicaid status) by examining differences in the proportion of Bay Area Parental Leave Survey of Mothers respondents who were eligible or whose partners were eligible for PPLO benefits. Exhibit 4 illustrates the limited reach of the PPLO among lower-income workers. Levels of employment during pregnancy differed between these groups, as would be expected given Medicaid eligibility criteria, but 67.9 percent of Medicaid-covered women were employed during their pregnancies (89.1 percent of non-Medicaid-covered women were employed). Remarkably, among women who were employed, 65.4 percent of non-Medicaid-covered workers were employed in jobs that were covered by the PPLO, but only 33.3 percent of Medicaid-covered workers were in jobs that were covered by the PPLO, because of the concentration of lowincome workers in small firms or the informal sector. Collectively, this means that just 22.6 percent of Medicaid-covered women who had recently given birth were covered by the PPLO compared with 58.3 percent of non-Medicaidcovered women. Similarly, 31.3 percent of Medicaid-covered women had partners who were covered by the PPLO compared with 61.0 percent of non-Medicaid-covered women.

Exhibit 3
Percent of women employed in San Francisco, California, who had knowledge of available maternity leave benefits before versus after the San Francisco Paid Parental Leave Ordinance (PPLO), and source of that knowledge, by Medicaid status, 2016-17

Discussion
Results of this study suggest that San Francisco's Paid Parental Leave Ordinance, the first fully paid leave mandate in the US, moderately increased parental leave uptake among fathers and led to only small increases in the amount of leave taken by mothers. Men in San Francisco increased their uptake 13 percent more after the PPLO went into effect than did men in surrounding counties (appendix exhibit A6). 22 However, limited knowledge and reach of the PPLO likely dampened its potential effects.
Coverage restrictions may help explain the relatively limited uptake of the PPLO. Nationally and in California, small employers are the least likely to offer paid leave. 11,14 By excluding employers with fewer than twenty employees, the PPLO did nothing to address this gap.
Notably, the likelihood of working in a PPLOcovered job differed significantly by Medicaid status, with Medicaid-covered women far less likely than non-Medicaid-covered women to work in jobs or have partners employed in jobs covered by the PPLO. These groups differed in their likelihood both of working for a covered firm (private sector with at least twenty employ-ees) and of meeting minimum hours requirements (at least eight hours per week for at least six months before taking leave). This suggests that current policies are not well targeted for low-income working families.
Similar to prior studies of paid leave policies, we found limited awareness of the PPLO. [15][16][17] We extended this literature by showing that information about available maternity leave benefits and the PPLO was not equally accessible by all workers. By wide margins, lower-income workers reported less knowledge of their maternity leave benefits, less help from their employers in learning about their benefits, and lower awareness of the PPLO than their higher-income counterparts. The finding that fewer than 2 percent of all Medicaid-covered respondents had heard of the PPLO and had accurate information indicates that additional outreach efforts are necessary to reach all workers. The lack of knowledge persists despite substantial outreach efforts by city government, legal aid groups, and others. Our results suggest that one promising avenue for reaching potential beneficiaries is through the health care system. Health care providers were the only source of information reported as being used equally by Medicaid-and non-Medicaid-covered respondents, and in both groups, the frequency with which health care providers were cited increased over time.
The increase in women's total leave-taking was small. This is perhaps not surprising given the relatively long average duration of leave among new parents in the Bay Area. Our Bay Area Parental Leave Survey of Mothers data show that about four-fifths of employed women in the Bay Area take twelve or more weeks of leave (appendix exhibit A4). 22 This is consistent with recent findings using California statewide administrative claims that showed a median leave duration among female claimants of twelve weeks. 14 Supplementary wage replacement for six of those weeks, as provided by the PPLO, may have some impact on financial security, but it appears to have had little effect on the (disproportionately low-income) remaining women taking shorter leaves. This is despite the fact that 93 percent of Bay Area Parental Leave Survey of Mothers respondents taking less than twelve weeks of leave replied that if leave were available at full pay, they would prefer to take a full twelve weeks of leave (data not shown).
Despite a modest increase in male leave claimants in response to the PPLO, leave-taking remains limited among fathers. As reported by the mothers in our Bay Area Parental Leave Survey of Mothers sample, one in ten covered partners employed in San Francisco after the PPLO went into effect did not take any leave, and just 59 percent took more than two weeks (appendix exhibit A2b). 22 Despite the PPLO increasing to full pay the six weeks of parental leave that the Paid Family Leave program had previously offered at partial pay, the PPLO did not appear to increase average leave duration. This partly reflects fear of employer reprisal, as reported by Bay Area Parental Leave Survey of Mothers respondents: Fear of losing one's job was reported by 20 percent of mothers as the reason they did not take more leave and as a main reason for 25 percent of partners who did not take any leave (data not shown). It also likely reflects limitations in knowledge and awareness we observed among mothers (the survey did not assess PPLO awareness among partners).
Access to paid leave has been shown to have important health and economic impacts. A large body of evidence has shown that work commonly factors into women's decisions about breastfeeding. [23][24][25][26] Ariel Pihl and Gaetano Basso found that California's Paid Family Leave program decreased childhood hospitalizations, likely as a result of decreased group child care among younger infants. 5 It is not clear whether health care needs such as well-baby and postpartum check-ups would change in response to a paid leave policy such as the PPLO, but this is an important topic for future research.
Our results contrast with those of a recent study of employers' responses to the PPLO, which showed a 15-percentage-point increase in the proportion offering paid leave in response. 27 This could reflect either employers reporting what they should offer, rather than what they do offer, or employees' limited knowledge of their enhanced parental leave benefits.

Conclusion
Paid leave policies are gaining momentum at the state and federal levels: Eight states and the District of Columbia have enacted paid family and medical leave policies, and the Family and Medical Insurance Leave (FAMILY) Act has bipartisan support in Congress. The novel coronavirus disease (COVID-19) global pandemic has increased the urgency of these policies. The pandemic underscores the lack of paid family leave for workers who are dealing with their own illnesses or caring for seriously ill or quarantined family members and places them at greater risk for illness and loss of income. 28 As these policies move forward, researchers must continue to study their impacts on health and other outcomes, taking into consideration different policy design features, such as wage replacement rate, coverage restrictions, and source of payment.
San Francisco's Paid Parental Leave Ordinance represents an ambitious policy design that requires employers to provide supplementary wage replacement when their workers take leave. It is the first policy to be enacted in the US that provides full wage replacement. Although the impact on employers appears minimal and support for the PPLO remains high, 27 the policy's coverage and eligibility restrictions (notably excluding public-sector employers and employers with fewer than twenty employees), as well as the complex application process, may limit the potential impact of the policy, particularly among lower-income workers. A simpler policy that directly expands the wage replacement rate of the current State Disability Insurance system could be more effective in reaching vulnerable parents and other caregivers already covered by State Disability Insurance and the Paid Family Leave program and would not require workers to file separate claims with their employers. An even more effective method of reaching low-income workers would be to incorporate the large share of low-income workers in the informal sector not yet covered by State Disability Insurance, although that would require even greater policy change and outreach efforts. ▪ Support for this research was provided by the Robert Wood Johnson Foundation. The views expressed here do not necessarily reflect the views of the foundation. The authors thank the four anonymous reviewers and the editorial team of Health Affairs for their very helpful comments. This is an open access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt, and build upon this work, for commercial use, provided the original work is properly cited. See https://creativecommons.org/licenses/ by/4.0/.