Paid Sick Leave in Washington State: Evidence on Employee Outcomes, 2016–2018


Objectives. To estimate if Washington State’s paid sick leave law increased access to paid sick leave, reduced employees’ working while sick, and relieved care burdens.

Methods. I drew on new data from 12?772 service workers collected before and after the law took effect in January 2018 in Washington State and over the same time period in comparison states that did not have paid sick leave requirements. I used difference-in-difference models to estimate the effects of the law.

Results. The law expanded workers’ access to paid sick leave by 28 percentage points (P?<?.001). The law reduced the share of workers who reported working while sick by 8 percentage points (P?<?.05). Finally, there was little evidence that the law served to reduce work–life conflict for Washington workers.

Conclusions. Mandated paid sick leave increased access to paid sick leave benefits and led to reductions in employees’ working while sick. However, covered workers did not experience reductions in work–life conflict in the period immediately following passage.

Compared with other developed countries, in the United States, access to paid time off from work is low and stratified by socioeconomic status.1 The dearth and the inequality of coverage arises from the lack of a federal provision or mandate for paid vacation, parental leave, or sick days.2 Yet the provision of these benefits can have important positive effects on worker well-being and population health. In the domain of paid sick leave, employees who lack the benefit are more likely to forgo needed medical care,3 and parents without paid sick leave are less likely to stay home to care for themselves or a family member.4 More broadly, expansion of paid sick leave appears to have positive population-level effects on health.5

In recognition of this evidence, over the past decade a number of cities and states have attempted to address this lack of federal policy by passing mandates that require private-sector employers to provide paid sick leave, with 11 states and 30 cities having passed such laws.6 However, the requirement to provide benefits does not necessarily equate with actual provision or use. A large body of research demonstrates significant noncompliance by private-sector employers with other provisions of labor law, including the classification of workers,7 wage and hour violations,8,9 and occupational health and safety rules.10

Evaluations of paid sick time statutes do show that following passage, employers report increased provision of paid sick time. In San Francisco, California, the share of firms reporting provision of paid sick time rose from 73% to 91%,11 and in Seattle, Washington, the share rose from 80% to 91%.12 Other studies only survey employers following passage and report similarly high levels of self-reported compliance—87% in New York City13 and 93% in Jersey City, New Jersey,14 after implementation.

However, relying on employer reports of provision of benefits is not without limitations. Employers may overstate their provision of paid sick time if they are concerned that reporting noncompliance could lead to sanction.10 Furthermore, and perhaps more importantly, if workers do not know about the availability of paid leave they cannot use it, and employer-side surveys cannot capture this dynamic.10

Yet there is much less evidence on the employee side, and what evidence exists does not align with employer reports. In Seattle, although the employer data showed a 91% compliance rate, 12 of 19 workers who did not have paid sick coverage before the ordinance reported still not having it afterward during in-depth interviews.15 A small-scale poll of 100 workers in Jersey City did not directly assess changes in employees’ reports of access to paid sick leave, but it did find that 50% had not heard of the law. Among workers, half had earned at least 1 paid sick day since the law went into effect, a quarter reported not earning any, and another quarter were unsure of whether they had earned any paid sick days, in contrast to the 93% of employers who reported providing earned sick days.14 Although this research raises the concern that there may be a substantial disconnect between legal mandates, employers’ reports, and employees’ experiences, the data are quite limited by the small samples available in Seattle and Jersey City.

Furthermore, beyond access to paid sick time, there is limited research that gauges the effects of expansion of paid sick leave on worker behavior. Observational research finds evidence of strong associations between having paid sick leave and staying home while sick,3 as well as parents taking time off to care for themselves or others.4 However, having paid sick leave is not random and is strongly correlated with other occupational and worker attributes,16 raising concerns about whether such associations are causal. The only existing research that gauged worker-level responses to expansions of paid leave in a quasi-experimental framework focused on labor market effects, and it found either no evidence of reductions in employment17 or modest reductions.18

The literature is largely silent on the question of whether expanding paid sick leave actually reduces rates of “presenteeism,” or employees’ reporting to work when sick.19 This question is particularly pressing in sectors that involve customer-facing interactions, perhaps especially food service,6 because of the risk of contagion from sick workers to the broader population.5

Paid sick leave can be used when employees themselves are ill, but also to care for family members. One might then expect that access to paid sick leave would make it easier to meet care obligations. The unequal distribution of informal care obligations by gender and by parental status20 would suggest that paid sick leave might reduce work–life conflict for women and for mothers.


Washington State’s paid sick leave law was enacted directly by voter initiative in November 2016. The law requires that nearly all employers, regardless of size and industry, provide paid leave and that it be provided to nearly all workers, regardless of full- or part-time status. Workers accrue leave at the rate of at least 1 hour for every 40 hours worked and may take leave to care for themselves or a family member when sick or to seek preventative care. The law went into effect January 1, 2018.21

I drew on a repeated cross-section of workers in the retail and food service industries. Although Washington’s paid sick leave law covers nearly all employees in the state, workers in the targeted industries are of particular policy relevance because they are among the most unlikely to have had paid sick leave coverage prior to the law3 and may be among the most financially unable to take unpaid time off from work or to hire others to care for sick family members.22 For these reasons, the effects of the law are likely to be greatest for retail and food workers, a population of considerable policy interest. For these same reasons, however, the estimates in this study are unlikely to provide insight into the broader effect of the law across industries.

The Shift Project (of which I am co-principal investigator) recruited workers to a custom survey using paid advertisements on Facebook that targeted workers on the basis of their employer. The data collection methodology and tests of bias are described in detail in Schneider and Harknett.23 The Shift Project fielded recruitment advertisements in the fall of 2016, the fall of 2017, and the spring of 2018. The project fielded the advertisements nationally to workers at 87 of the largest retail or food service firms in the United States. It recruited respondents using incentives that ranged in value from an entry into a drawing for an iPad to a $15 gift card. At each wave, the survey oversampled Washington State workers. An important limitation of the data is that this convenience sampling approach probably yielded a sample that was not representative of the target population of retail and food service workers employed at large firms in Washington and comparison states.

I restricted the sample to 12?772 workers employed in Washington State (outside of Seattle and Tacoma) or in 1 of 41 states (listed in Table A, available as a supplement to the online version of this article at that had not passed a paid sick time law as of June 2018 and who worked at 1 of 24 large employers whose workers were surveyed at each wave and had complete data on all covariates. I also examined a subsample of female workers (reducing the sample to 9618 respondents) and of mothers (reducing the sample to 4796 respondents). These sample restrictions are described in detail in Appendix A (available as a supplement to the online version of this article at

Key Variables


To define the key independent variable, I geocoded responses based on Internet Protocol (IP) addresses for Washington State (coded 1) versus all other states (without a paid sick leave law, coded 0; a full list is included in Table A). I interacted this dichotomous variable with a measure of time relative to the implementation date of Washington’s paid sick leave law: January 2018. I disaggregated the pretreatment period and separately examined fall 2016 and fall 2017 versus the posttreatment period of spring 2018.

Paid sick leave.

I took respondents’ reports of whether they were provided with paid sick time as the first dependent variable. These data were collected with an item that instructed respondents to “Please look at the following list of benefits that employers sometimes make available to their employees. Which of the benefits on this list can you receive as part of your job at [name of employer]?” The first option was “paid sick days.”

Working while sick.

I also examined respondents’ reports of presenteeism, or working while sick. Respondents were asked, “In the past month, did you ever work at [name of employer] even though you were feeling sick?” Respondents could select “Yes,” “No, I was sick but I stayed home,” or “No, I haven’t been sick in the past month.” This item was asked in fall 2017 and spring 2018. In fall 2016, respondents were asked a variant of the question: “In the past month, how often have you been sick, but gone to work anyway?” where the response options were “once,” “twice,” “three or more times,” or “never.” I harmonized the 2 items and compared respondents who either reported that they were sick and worked anyway (fall 2017 and spring 2018) or who were sick and worked at least once (fall 2016; all 3 items coded as 1) against all others (coded as 0).

Work–life conflict.

The survey data contained 2 items used in the literature to capture work–life conflict and that apply to the case of paid sick leave. The first item asked respondents to rate their agreement or disagreement with the statement, “In my work schedule at [name of employer], I have enough flexibility to handle family needs.” Responses were “always true” (coded 1), “often true” (coded 2), “sometimes true” (coded 3), or “never true” (coded 4). The second item asked respondents to rate their agreement or disagreement with the statement, “My shift and work schedule at [name of employer] cause extra stress for me and my family.” Response categories were the same (reverse coded). I also created a scale that combined these 2 items (??=?0.69).

Control variables.

I was able to account for a set of potentially confounding worker characteristics. I adjusted estimates for gender, race/ethnicity, age, presence of children by age group, educational attainment, school enrollment status, marital status, usual work hours, tenure at current employer, hourly wage, and managerial status. These controls guarded against the risk of bias introduced if the composition of the workforce changed over time, especially if such compositional changes differed between Washington and comparison states.


I used a difference-in-differences model to estimate the effect of the Washington paid sick leave law. Importantly, this model relied on the strong assumption that absent intervention, the difference between treatment and control units observed in the time period prior to treatment would be unchanged in the period posttreatment. Although this counterfactual cannot be tested directly, the presence of steady differences in the pretreatment period can be seen as evidence in support of this assumption.24 I estimated a linear probability model for the first 2 outcomes and an ordinary least squares regression model for the 3 work–life conflict outcomes in which the key term was an interaction between working in Washington State and time. For the binary outcomes, I estimated robust standard errors. For each outcome, I estimated a model that allowed for distinct estimates of the dependent variables in each of the 2 pretreatment periods. This allowed at least partial assessment of the parallel trends assumption, although I was limited by having only have 2 pretreatment observations. For the analysis of the work–life conflict measures, I estimated 3 versions of each model: first for the full sample, then for women, and finally for women with children.

In addition to controlling for a set of worker-level demographic and work-related characteristics, I was able to include employer fixed effects, which allowed within-employer comparisons over time. Appendix B (available as a supplement to the online version of this article at presents a set of robustness tests.


I first present results from the analysis of the effects of Washington’s paid sick leave law on employee’s reports of having paid sick time. I then examine the effects of the law on employees’ reports of working while sick. Finally, I examine how the law affected work–life conflict.

Paid Sick Leave Coverage

Table 1 presents the key regression result for the model examining how worker access to paid sick time changed after the implementation of the Washington law. In model 1, I separated the preimplementation period into 2 blocks—fall 2016 and fall 2017—and contrasted the predicted levels of the outcome with the postimplementation period: spring 2018. The interaction terms indicate that the Washington State and comparison state trend between fall 2016 and fall 2017 is indeed parallel (??=?0.02; P?=?.782) but that there is significant divergence after implementation (??=?0.28; P?<?.001).


Figure 1 plots predicted values from this model. In the left-side panel, the gray vertical line denotes when the Washington law went into effect (and the black line when it was passed). The share of workers reporting paid sick leave benefits was nearly identical in Washington State (33%) and in comparison states (27%) prior to implementation. Although there were clear parallel trends in the preimplementation period for Washington State and comparison states, there was also a clear upward slope in both between fall 2016 and fall 2017. However, following implementation, the slope is significantly flatter for comparison states than for Washington State, where there is a distinct steepening to the slope. At follow-up, the share of workers in Washington reporting paid sick leave was 72% versus 38% in comparison states, an estimated effect size of a 28 percentage point increase in coverage.

Working While Sick

The Washington State paid sick leave law appears to have increased access to paid sick time. In model 2 of Table 1, I examined whether that translated into workers being less likely to come to work when sick. Here too, there was clear evidence in support of parallel trends: there was no difference in the change in working while sick between Washington and comparison workers between fall 2016 and fall 2017 (??=?0.02; P?=?.761). Following implementation, however, there was a significant difference, with Washington workers less likely to work while sick (??=??0.08; P?<?.05).

Figure 1b plots predicted values from this model using the same approach as before. There was no evidence of change in the share of workers in comparison states who reported working while sick; it stayed at a steady 64% across all 3 periods. In contrast, the share was steady in Washington State in the preimplementation periods (68% and 70%), but then dropped significantly after implementation to 59%. The plot shows the emergence of difference between Washington State and comparison states after implementation of the paid sick leave law.

Work–Life Conflict

Estimates from the difference-in-differences models of work–life conflict are presented in Table 2. The table presents results for 3 models (all respondents, female respondents, and mothers) for 3 outcome measures. The first 3 models (1–3) show the estimates for respondents’ reports of having enough flexibility in their schedules to deal with personal and family matters. This outcome variable most closely captures the dimension of work–life conflict that paid sick leave might help to mitigate. Across all 3 models, the key interaction is in the expected direction but is not statistically significant. Similarly, there were no significant effects of the paid sick leave law on the alternative measure of work–life conflict (if the respondent’s work schedule caused stress on self or family; models 4–6) or on the combined measure (models 7–9). Figure A (see Appendix A) shows the estimates from models 3, 6, and 9.



In the absence of federal regulation, paid sick leave laws have been enacted in 11 states and 30 cities across the country. Prior evaluation research has drawn on employer self-reports of compliance or used employee data to gauge labor market effects. The result is that we know relatively little about whether passage leads to workers having more access to paid leave and, secondly, whether it allows workers to use that leave to avoid working while sick and to reduce their burden of caring for others.

I have drawn on unique survey data collected from 12?722 retail and food service workers over a 2-year period spanning the implementation of Washington State’s paid sick leave law. My estimates from difference-in-differences models show that the Washington State law expanded workers’ access to paid sick leave by 28 percentage points. The Washington law also reduced the share of workers who reported working while sick in the month prior to the survey by 8 percentage points. However, at least over this short-term follow-up period, the law did not serve to reduce work–life conflict for Washington workers. These estimates indicate the effectiveness of Washington State’s paid sick leave law; however, they also provide valuable, plausibly causal estimates of the effect of paid sick leave on presenteeism and work–life conflict. Although prior research found evidence of strong associations,3,4 my approach diminishes concern about bias from unobservables.


This research is subject to important limitations. The estimates are based on worker reports collected before the law’s implementation and then 2 to 7 months following implementation. They summarize, then, very short-term impacts of implementation. In this respect, they suggest that many employers in Washington State acted quickly to implement paid sick time and that employee awareness of the benefit was substantial. The strongest effects were on the most proximate outcomes: availability of paid sick leave and not working while sick; there was little effect on the more distal outcome of work–life conflict. It is possible that such more-distal outcomes will be more strongly affected by the law over time as employees make repeated use of the benefit. However, it is important to bear in mind that the estimates are limited to retail and food service workers, who may be among the most likely to experience change as a result of the law.

Additionally, the sample was constructed using an innovative approach that takes paid advertisements on Facebook as both the sampling frame and recruitment device. Although the data are internally consistent and meet benchmarks from gold standard sources,25 they represent a nonprobability sample. To the extent that sample composition is biased on a variable that conditions the estimated effect size, the average treatment estimates may also be biased. However, these data do allow insight into the employee-side experience of paid sick leave regulations that is otherwise difficult to obtain.

Finally, the data offer limited capacity to assess the parallel-trends assumption that is fundamental to the difference-in-difference model. Although there is no evidence of significant divergence in trends in the 2 periods before implementation, there is no longer-run pretrend to assess. Similarly, the data offer limited capacity to assess the extent to which other contemporaneous changes in labor law specific to Washington State might contaminate the study’s estimates. Although Seattle passed secure scheduling legislation in this period, workers in Seattle were excluded from the analysis, and although Washington State has also passed a paid family and medical leave law, this law does not take effect until 2020. There was a statewide increase in the minimum wage over the observation period. Although this change could confound the effects of paid sick leave, given the outcomes examined, this seems unlikely. Specifically, there is no theoretical basis for hypothesizing that when employers are mandated to increase wages, they might also increase paid sick leave. It is possible, though, that if workers were paid more, they might opt not to work while sick even in the absence of paid sick leave.

Public Health Implications

I present early evidence on the effectiveness of Washington State’s paid sick leave law. In the first year of implementation, there were large and significant effects of the law on workers’ reported access to paid sick leave and reductions in working while sick for a particularly disadvantaged segment of the state’s labor force. These results support the idea that policy levers that may be perceived as quite distal to health outcomes, such as labor policy, can have important effects on public health by shaping important socioeconomic determinants of health and well-being. In this sense, the broader constellation of labor regulations that are designed to “raise the floor” on job quality, such as minimum wage, paid parental leave, and secure work scheduling laws, can all be seen as a way of advancing health, perhaps particularly for the most disadvantaged workers and families in the United States.


  1. Bureau of Labor Statistics. 2017Employee benefits survey: paid leave benefits. Table 32: leave benefits: access, private industry workers, March 2017. Available at: Accessed November 16, 2019. Google Scholar
  2. Heymann J, Rho HJ, Schmitt J, Earle AContagion Nation: A Comparison of Paid Sick Day Policies in 22 CountriesWashington, DCCenter for Economic and Policy Research2009Google Scholar
  3. DeRigne L, Stoddard-Dare P, Quinn LWorkers without paid sick leave less likely to take time off for illness or injury compared to those with paid sick leaveHealth Aff (Millwood). 2016;35(3):520527CrossrefMedlineGoogle Scholar
  4. Clemans-Cope L, Perry C, Kenney G, Pelletier J, Pantell MSAccess to and use of paid sick leave among low-income families with childrenPediatrics2008;122(2):e480e486CrossrefMedlineGoogle Scholar
  5. Stearns J, White CCan paid sick leave mandates reduce leave-taking? Labour Econ2018;51:227246CrossrefGoogle Scholar
  6. Marotta J, Greene SPaid Sick Days: What Does the Research Tell Us About the Effectiveness of Local Action? Washington, DCUrban Institute2019Google Scholar
  7. Carre F(In)Dependent Contractor MisclassificationWashington, DCEconomic Policy Institute2015. EPI Briefing Paper #403. Google Scholar
  8. Galvin D. Deterring wage theft: alt-labor, state politics, and the policy determinants of minimum wage compliancePerspect Polit2016;14(2):324350CrossrefGoogle Scholar
  9. Bernhardt A, Spiller M, Polson DAll work and no pay: violations of employment and labor laws in Chicago, Los Angeles and New York CitySoc Forces2013;91(3):725746CrossrefGoogle Scholar
  10. Weil D, Pyles AWhy complain? Complaints, compliance, and the problem of enforcement in the US workplaceComp Labor Law Policy J2005;27:5992Google Scholar
  11. Colla C, Dow W, Dube A, Lovell VEarly effects of the San Francisco paid sick leave policyAm J Public Health2014;104(12):24532460LinkGoogle Scholar
  12. Romich JLocal mandate improves equity of paid sick leave coverage: Seattle’s experienceBMC Public Health2017;17(1):60CrossrefMedlineGoogle Scholar
  13. Appelbaum E, Milkman RNo Big Deal: The Impact of New York City’s Paid Sick Days Law on EmployersNew York, NYCenter for Economic and Policy Research2015Google Scholar
  14. Lindemann D, Britton DEarned Sick Days in Jersey City: A Study of Employers and Employees at Year One Center for Women and WorkPiscataway, NJRutgers, the State University of New Jersey School of Management and Labor Relations2015Google Scholar
  15. Romich J, Bignell W, Brazg Tet al. Implementation and Early Outcomes of the City of Seattle Paid Sick and Safe Time OrdinanceSeattle, WACity of Seattle, Office of City Auditor2014Google Scholar
  16. Hill HPaid sick leave and job stabilityWork Occup2013;40(2):143173CrossrefGoogle Scholar
  17. Pichler S, Ziebarth NLabor market effects of US sick pay mandatesJ Hum Resour2018;0117-8514R. [E-pub ahead of print October 16, 2018.] Google Scholar
  18. Ahn T, Yelowitz AThe short-run impacts of Connecticut’s paid sick leave legislationAppl Econ Lett2015;22(15):12671272CrossrefGoogle Scholar
  19. Lohaus D, Haberman WPresenteeism: a review and research directionsHum Resour Manage2018;29:4358Google Scholar
  20. Maume DCan men make time for family? Paid work, care work, work–family reconciliation policies, and gender equalitySoc Currents2016;3(1):4363CrossrefGoogle Scholar
  21. Washington State Dept of Labor & Industries. Employer resource center for paid sick leave2019. Available at: Accessed November 16, 2019. Google Scholar
  22. Osterman P, Schulman BGood Jobs America: Making Employment Better for EveryoneNew York, NYRussell Sage Foundation2011Google Scholar
  23. Schneider D, Harknett KConsequences of routine work-schedule instability for worker health and well-beingAm Sociol Rev2019;84(1):82114CrossrefGoogle Scholar
  24. Wing C, Simon K, Bello-Gomez RDesigning difference in difference studies: best practices for public health policy researchAnnu Rev Public Health2018;39(1):453469CrossrefMedlineGoogle Scholar
  25. Schneider D, Harknett KWhat’s to like? Facebook as a tool for survey data collection. Sociol Methods Res. 2019. Available at: Accessed December 13, 2019. Google Scholar


The Shift Project gratefully acknowledges funding from the Bill and Melinda Gates Foundation, National Institutes of Health (NICHD), the Robert Wood Johnson Foundation, The James Irvine Foundation, the W.T. Grant Foundation, and support from the Institute for Research on Labor and Employment at UC Berkeley. The views expressed here do not necessarily reflect the official views of the sponsors.

The authors gratefully acknowledge the work of Megan Collins and Connor Williams in the preparation of this report.

Conflicts of Interest 

The author reports no conflicts of interest.

Human Participant Protection

This study was approved by the University of California, Berkeley’s Committee for the Protection of Human Subjects.