The Adoption of Paid Sick Leave in US States

Abstract

Paid sick leave, or the ability to remain home from work in the event of illness and receive compensation, has risen in prominence after the COVID-19 pandemic. Among the OECD countries, all but two, the United States and South Korea, have national paid sick leave (PSL) policies. Yet despite federal inaction in the United States, states have been adopting PSL, with 15 plus the District of Columbia having done so by the end of 2020. In the absence of federal policy, what drives states to adopt PSL mandates? In this article, we investigate two possible explanations – women in politics and jurisdictional competition. In the former, we suggest that increases in female representation in state-level governance make it more likely that a state will adopt a PSL policy. In the latter explanation, we suggest that jurisdictional competition in the form of cities or counties adopting municipal PSL policies creates pressure on the state-level government to enact statewide policies to harmonize policy, in a process of “bottom-up” federalism. To evaluate our hypotheses, we create a dataset of all state and municipal PSL policies in the United States. We find strong support for the gender representation argument, but not for the jurisdictional competition argument.

The COVID-19 pandemic and the ensuing economic downturn strained social safety nets around the world. As the pandemic spread, one such policy garnered heightened attention: paid sick leave (PSL). PSL came to be seen as not only a standard component of the modern welfare state, but as a public health policy with important consequences for the spread of disease. Studies on the effectiveness of PSL as a public health measure in Germany and the United States indicate that the rate of influenza-like illness is lower in states where the government mandates PSL (Pichler and Ziebarth 2017). In the H1N1 pandemic of 2009–10, individuals without access to PSL were more likely to report symptoms (Kumar et al. 2012), whereas those with PSL were more likely to take sick days when they felt unwell (Zhai et al. 2018). The availability of PSL reduces the incidence of presenteeism, with workers instead opting to remain home from the workplace when feeling sick at a higher rate, and thus slows the spread of illness (Colla et al. 2014; Schneider 2020).

In this discussion, the United States stands out as one of only two countries in the Organization for Economic Cooperation and Development (OECD) (the other is Korea) without statutory national PSL policies (Organization for Economic Cooperation and Development 2020). Across the OECD, excluding the United States and Korea, national PSL policies on average replace 70% of an employee’s income for up to 4 weeks per year. The United States and Korea scrambled to provide benefits as the pandemic spread, but those measures were temporary and limited to COVID illnesses.

Although the outlier status of the United States in terms of national paid leave is well known, it is less widely recognized that in the face of federal inaction many states have been adopting PSL policies. Starting with Washington, D.C., in 2008 and Connecticut in 2011, 15 US states, plus the District of Columbia, passed PSL legislation by the end of 2020.¹ In less than a decade, states with a cumulative population of 123 million have adopted PSL, meaning that subnational governments within the United States have expanded this benefit to a population greater than that of France and Spain combined. This is no small policy reform, yet it remains unexamined by political scientists.

This article begins to fill this gap by exploring the state-level dynamics that shape the likelihood of adoption of PSL policies using a new dataset of all subnational PSL policies in the United States. We focus on two possible drivers of state-level adoption of PSL. First, we argue that women have distinctive preferences for PSL policies and draw on the substantive representation literature to hypothesize that the likelihood of a state adopting PSL increases as the number of women serving in the state legislature rises. Second, we argue that the logic of jurisdictional uniformity implies that when municipalities begin to pass PSL legislation, the state government becomes more likely to adopt PSL to provide regulatory consistency for businesses across the state. Following the literature on “bottom-up federalism” (Shipan and Volden 2006), we expect a snowball effect to occur when a significant number of municipalities begin to pass local PSL legislation. To evaluate these hypotheses, we create a dataset of all state and municipal PSL policies in the United States. We find strong support for the gender representation argument, but not for the jurisdictional competition argument. The gender representation finding is stable to a battery of sensitivity tests, including Oster’s test of vulnerability to omitted confounding variables (Oster 2019), and is consistent with case narratives of the passage of PSL.

We first discuss the spread and effect of PSL policies, and then turn to elaborating our two hypotheses for explaining variation across states and time in the adoption of PSL. We then introduce the data and report the results of the discrete-time Cox models, followed by a discussion of several cases before concluding.

PSL Policies in the United States

Although there is no federal law mandating PSL, efforts have been made to pass legislation in the past. The most recent attempt came in March of 2019 with S.840, the Healthy Families Act, introduced by Senator Patty Murray (D-WA). The bill failed to move out of committee and died on December 31, 2020. In light of the coronavirus pandemic, Congress did include PSL provisions as part of the Families First Coronavirus Response Act (H.R. 6201, enacted 18 March 2020). The legislation includes provisions that require certain employers to offer up to two weeks of sick leave at an employees’ regular pay rate and an additional two weeks of leave at two-thirds of the employee’s regular rate of pay. However, the benefit was restricted to employees or their families that are dealing specifically with COVID-19, and expired at the end of the year. President Biden proposed a comprehensive parental, family, and medical leave program as part of his American Families Plan, but although parts of the plan were passed in other legislation, the PSL provisions did not move forward.

With continuing lack of action at the federal level, several US states have enacted their own sick leave policies: 15 states, plus the District of Columbia, adopted PSL by the end of 2020. In 2011, Connecticut became the first state to pass a state-level PSL policy. At the municipal level, San Francisco led the way with a PSL policy taking effect in 2007. Although we do not analyze the differences in generosity and coverage of policies in this article, benefits do vary across states and municipalities. For example, Massachusetts’ policy applies to non-federal public and private employers with 11 or more employees, whereas Nevada’s policy applies to private employers with 50 or more employees. Several states award one hour of accrued time for every 30 hours worked (California, Maryland, and Massachusetts, for example), whereas employees in Connecticut accrue one hour for every 40 hours worked, and in Vermont employees must work 52 hours for every hour accrued. Most states cap sick leave accrual at 40 hours per year, with some states allowing these hours to be carried over from one year to the next.

More broadly, it is also important to recognize patterns across sectors in access to PSL from private employers.² Among US employees, about 45% of full-time workers have no PSL, and coverage rates are particularly low for part-time workers, young employees, and service sector workers (Susser and Ziebarth 2016). The divide in the availability of sick leave benefits tends to fall between those at higher levels of income and those at the lower end: “86% of workers in the highest wage decile have access to PSL, compared to just 19% of those in the lowest page decile” (Gould, Filion, and Green 2011). Across sectors, the service industry lags with lower coverage rates than other sectors (Ingraham 2020). Even more concerning, the Centers for Disease Control and Prevention has reported that “1 in 5 food service workers have reported working at least once in the previous year while sick with vomiting or diarrhea” (Ingraham 2020). State-level mandates have significant scope to fill in gaps in coverage from private employers and to provide a more equitable distribution of PSL benefits.

Studies have investigated whether access to sick leave reduces presenteeism (the practice of showing up to work while ill) as well as its impact on the spread of illness. Research has found that states with PSL have a lower rate of reported influenza-like illness (Pichler and Ziebarth 2017). Furthermore, studies at the state and local level have indicated that the availability of sick leave decreased presenteeism in San Francisco (Colla et al. 2014) and Washington state (Schneider 2020). Moreover, the benefits of PSL seem to come without the negative labor market effects (reduced employment levels or lower wages) feared by many (Pichler and Ziebarth 2020). Overall, substantial evidence suggests that access to PSL reduces the spread of illness and worker stress levels with few, if any, negative labor market effects.

Explanations: Women in politics and jurisdictional competition

Substantive representation and PSL

Our first hypothesis is that the likelihood of passing PSL legislation increases as the number of women in the state legislature rises. The argument proceeds in three steps. First, we posit that women have distinctive preferences over PSL that differ from those of men. Second, drawing on the literature on descriptive representation, we assert that female legislators place greater emphasis on turning these preferences into policies than male legislators do. Third, we argue that women need to be represented in the legislature in nontrivial numbers to turn these preferences into policies. From these three premises, it follows that the likelihood of passing PSL legislation will be increasing in the number of women occupying seats in the state legislature.

We support our assumption of gendered preferences over PSL by appealing to the large literature documenting gendered preferences over a range of social policies, gendered patterns of childcare and leave taking in the United States, and survey research. A common argument in favor of increased gender diversity among elected representatives is that women bring a different set of attitudes, interests, and beliefs to policymaking. Distinct preferences arise from women’s experiences in gendered societal positions, responsibilities related to caring for children, and occupational differences between men and women. Indeed, a large literature has found that women possess distinctive preferences over a range of social policies (Diamond 1977; Dodson 1998; Sapiro 1981; Thomas and Wilcox 2005; Wangnerud 2000; Yildirim 2022). Assuming that gendered preferences similarly apply to PSL makes sense because social and economic behavior patterns insure that paid sick days are of particular interest to women. Although men’s and women’s preferences about reconciling work and family life might, in principle, be similar, women’s shared life and work experiences, as well as prevailing social norms, impart a more intense interest in PSL policies. Women function as the primary caregivers in most families, and research has shown strong gendered patterns in leave-taking from work (Henreksson and Person 2004) and established that women miss more work than men to care for sick children (Angelov, Johansson, and Lindahl 2013; Smith and Schaefer 2012). Moreover, the benefits of PSL for women, in particular, have been well documented (DeRigne and Stoddard-Dare 2019; Miller and Williams 2013). Finally, the only scholarly analysis (as opposed to simple poll results) of the determinants of support for PSL bears out this pattern: gender is a strong and significant predictor of support for PSL even after controlling for an extensive battery of other variables (Lindemann, Houser, and White 2015).

However, why should women’s representation in state legislatures be required to translate these gendered preferences into policies? Here, we draw on the literature on descriptive representation. Even though male legislators should be equally responsive to their constituents’ needs, and thus women’s preferences about PSL policies, a large body of evidence has documented major discrepancies between constituents’ demands and their representatives’ priorities (Bartels 2008; Gilens 2012). In terms of closing this representation gap, it is well established that policymakers’ personal traits and life experiences influence their behavior in office (Canon 1999; Carnes 2012; Gelpi and Feaver 2002; Narud and Valen 2000; Washington 2008). Several studies on female representation find that when women have distinct preferences about policy, they prioritize these issues (relative to men) when they enter politics. Indeed, research on gender differences in legislators’ attitudes suggests that women tend to take more progressive stances and express more liberal preferences in areas traditionally considered as key interests of women, such as social policy (Lovenduski and Norris 2003; Lyn 2001; Narud and Valen 2000; Swers 2002). Many studies also find that women officeholders are much more likely to prioritize female constituents’ demands and preferences, and that female legislators view representing women’s interests as their duty (Diaz 2005; Reingold 2000; Swers 1998; Thomas 1994; Thomas and Welch 1991). In the United States, research finds that female legislators often prioritize issues of particular importance to female voters (Swers 2013), and, in turn, that female voters expect female legislators to pursue distinctive agendas and show heightened awareness of their actions in office (Jones 2014).

Our third premise is that passing PSL legislation becomes more likely as the number of women in the state legislature increases. This perspective posits that pushing forward a female-friendly agenda and overcoming obstacles, such as gender biases, male dominance in political arenas, and difficulties combining a political career with caring responsibilities requires collaborative efforts by a caucus of female legislators. Several national-level studies have found that increases in women’s presence in legislative bodies have a positive effect on the passage of female-friendly legislation in democracies (Kittilson, 2008; Schwindt-Bayer and Mishler, 2005). A regression discontinuity study using the adoption of gender quotas as a representation shock found that a sudden influx of women into the legislature shifts government spending priorities toward historically feminized policy areas, such as social policy (Clayton and Zetterberg 2018). Although scholars debate whether or not a critical threshold of women in the legislature is necessary (Beckwith and Cowell-Myers 2007), numerous studies support the general contention that as the number of women in politics increases, the emphasis on women’s issues increases (Barnes 2012; Bratton and Ray 2002; Franceshet and Piscopo 2008; Schwindt-Bayer 2010; Williamson and Carnes 2013).

Taken together, these three premises imply that the likelihood of passing PSL legislation will be increasing in the number of women occupying seats in the state legislature, leading to our first hypothesis:

H1. As the percentage of women in a state legislature increases, the likelihood of the state adopting paid sick leave increases.

Pressure for jurisdictional consistency and PSL

Just as individual states have passed PSL mandates in the absence of federal law, so too have local jurisdictions (cities, towns, and counties) in the absence of statewide legislation. For example, municipalities in New Jersey began passing their own PSL policies beginning in 2013, with Jersey City leading the way. By the time the state passed PSL legislation in 2018, 13 municipalities in New Jersey had already adopted PSL, covering approximately 15% of the state’s population. Individual cities or groups of municipalities that pass sick leave rules may create a snowball effect that induces the state to harmonize sick leave policies across the state in the interest of regulatory uniformity across the state’s economy. Furthermore, the adoption of PSL ordinances in one municipality may draw the attention of other municipalities to do the same, creating a sort of jurisdictional competition to craft and pass sick leave policies. By passing sick leave ordinances, municipalities may also create a learning effect by demonstrating the feasibility of PSL to other cities and the state government. We posit that as more local governments pass PSL legislation, increasing the percentage of the population covered by PSL, state governments will become more likely to pass state-level legislation to maintain consistency across the state and harmonize policy.

H2. As the share of a state’s population that is covered by municipal sick leave law increases, it becomes more likely that the state will adopt paid sick leave.

This hypothesis posits the spread of policies from lower to higher levels in a federal system, a dynamic dubbed “bottom-up federalism” (Shipan and Volden 2006). Given that states have long been considered laboratories of democracy, many studies have looked at bottom-up dynamics, but have primarily focused on the spread of policies from states to the federal level, rather than from local governments to the state level (see, e.g., Mooney 2001). Focusing on antismoking policies, Shipan and Volden (2006) demonstrate the existence of a snowball effect within states, whereby increasing numbers of local-level ordinances induce the state to act, either through a learning effect or a desire to impose policy uniformity across the state. Shipan and Volden find that the snowball effect is more likely to operate in states with professional legislatures because they have the staff, time, and expertise to dedicate to researching policies and their effects. In some issue areas, such as climate policies, activists have explicitly attempted to use the logic of bottom-up federalism to trigger action from higher levels of government (Engel 2006; Selin and VanDeever 2009).

Several examples from debates over PSL policies suggest that policymakers are aware of the types of dynamics described in the bottom-up federalism literature. More specifically, 17 states have passed laws to preempt municipalities from passing paid leave legislation in an apparent effort to stop the snowball effect from beginning and to ensure policy uniformity across the state at a low level of provision. For example, in 2011, Wisconsin Governor Scott Walker signed into law Senate Bill 23, passed by the Republican-controlled legislature, which prohibits cities and counties from mandating sick leave or paid family leave. At that time, Milwaukee had already passed a PSL ordinance in 2008. By enacting Senate Bill 23, the state government voided the city’s policy and imposed uniformity across substate jurisdictions. Recent research finds broader empirical support for the practice of preemption at the state level. Barber and Dynes (2023) “find that municipal officials are more likely to report preemption by the state government when their city is ideologically incongruent with their state legislature” (2). They find that this effect occurs in both Republican and Democratic state legislatures, but both tend to preempt more liberal cities and counties.

Data, methods, and results

To test these hypotheses, we rely on time-series cross-sectional data from all 50 states plus Washington, D.C., from 2007 to 2020, yielding a total of 651 state-year observations. We start our analysis in 2007 because PSL legislation at the subnational level emerged as a serious policy option around this time, with the first policy going into effect in San Francisco at the start of 2007. Our outcome of interest, the adoption of PSL at the state level, is a dichotomous variable which takes a value of 1 in the year a state passes a PSL law and 0 otherwise; states drop out of the analysis after the year of passage. Although the policies adopted by states vary in their details, we focus on understanding the drivers of PSL adoption rather than variations in generosity of benefits. Data on subnational PSL policies were collected from the National Conference of State Legislatures (2020) and the National Partnership for Women and Families (2020). These data cover local jurisdictions as well as states, allowing us to test the bottom-up federalism hypothesis by creating a variable recording the percentage of each state’s population living in a local jurisdiction with PSL each year. To test the substantive representation hypothesis, we draw on information from the Center for American Women and Politics (2020) at the Eagleton Institute of Politics at Rutgers University to create yearly values for the percentage of seats occupied by women in every state legislature, as well as indicator variables for the gender of the governor, lieutenant governor, and attorney general.³

In Figure 1, we plot a histogram of the distribution of values for women in the legislature covering 50 states from 2000 to 2020. The bulk of observations fall in the range of 15–30%. At the end of the right tail, the only state legislature to feature a female majority – Nevada from 2018 to 2020 – is represented. The average has been slowly but steadily growing over time, from 22% in 2000 to 29% in 2020. These data feature both substantial cross-state and overtime variations, underlining the importance of both longitudinal and cross-sectional analyses when exploring the link between the growth of women’s representation and the expansion of PSL.

Figure 1. Women in state legislatures, 2000–2020.

In Figure 2, we depict the spread of PSL laws at the local level. Since San Francisco adopted PSL in 2007, 37 substate jurisdictions in 10 states followed its example by the end of 2020. The bars, plotted against the left-hand axis, show the number of states, for each year, in which at least one local jurisdiction had adopted PSL. Among these states, the percentage of the state’s population covered by local laws has ranged from a low of 2.1% (when San Francisco was the only jurisdiction in California offering PSL) to a high of 48.3% (in New York after Westchester County joined NYC in offering PSL in 2019); these lows and highs are shown by the red and orange lines plotted against the right-side axis. In states with at least some local-level coverage, the average share of the population covered has gradually increased from 2.1% to nearly 26% in 2020, a trend plotted with the yellow line.

Figure 2. The spread of local-level paid sick leave laws.

In addition to these two variables of interest, we also include seven control variables in the baseline models (Table 1). Real per capita GDP comes from the Bureau of Economic Analysis at the US Department of Commerce (Bureau of Economic Analysis various). Across the OECD, higher levels of per capita income are associated with more social policies, yielding the expectation that richer states may be more likely to pass PSL. Population data come from the US Census Bureau (United State Census Bureau 2021; United States Census Bureau 2023); larger states may have greater incentives to standardize policies across substate jurisdictions. To proxy for the demand for PSL, we use the states’ labor force participation rate under the assumption that the more women that are active in the labor market, the higher the demand for PSL. Union density is included as an indicator of the strength of organized labor, which has consistently been a supporter of PSL policies. Caughey, Warshaw, and Xu (2017) find that states with legislatures and governorships controlled by Democrats enact more liberal policies, particularly in recent decades; accordingly, we create a binary variable for unified Democratic control of state government.⁴ We include a variable for possible geographical learning effects influencing PSL adoption: following the procedure in Shipan and Volden (2006), we calculate the percentage of neighboring states that have passed PSL legislation. Finally, we use an updated version of Squire’s legislative professionalism index (Squire 2007) to tap the idea that states with a more professional legislature may show a greater capacity for learning and innovation, yielding a greater likelihood of adopting PSL once the idea begins to spread.

Table 1. The passage of paid sick leave policy in US states, 2007–2020

Notes: Entries are coefficients with robust standard errors clustered by state in parentheses. *p < .1,**p < .05,***p < .01.

Results

To explore the relationship between the representation of women in parliaments, local coverage, and PSL policy, we analyze the timing of the adoption of legislation creating PSL from 2007 to 2020. Our coding rule identifies 15 events at the state level; in addition, 38 substate jurisdictions adopted PSL ordinances between 2007 and 2020 (the largest being New York City and the smallest Emeryville, California). The former provides our primary outcome variable, whereas we use the latter to create a “local coverage” variable that records the percentage of a state’s population living in local jurisdictions that have adopted PSL for each year.

We model the probability of the adoption of PSL policy as a function of time, women in the state legislature, local-level coverage, the battery of seven core control variables, and a range of additional variables in robustness checks. Discrete-time hazard models provide a useful framework for exploring the relationship between our variables of interest and the likelihood of PSL adoption. We use Cox proportional hazard models, and also explore logistic regressions with time trend variables as a robustness check (reported in the Supplementary Materials, Appendix). Having no strong theoretical priors about the shape of the hazard function, we chose the semi-parametric Cox proportional hazard model for our baseline models. With the Cox model, we do not have to assume a specific probability distribution about the time until events occur (survival time). States that have never adopted PSL legislation are treated as right censored. In the baseline models, we include three states in which PSL was passed by a ballot initiative after efforts in the legislature stalled: Washington, Arizona, and Massachusetts. Results do not depend on this decision; in the Supplementary Materials, Appendix, we report models dropping the initiative states.

The results of Cox survival models for the adoption of PSL legislation are in Table 1, with standard errors clustered by state; results for the same models using logistic regressions are reported in the Supplementary Materials, Appendix.⁵ The key takeaway is that the substantive representation hypothesis receives strong support, whereas the bottom-up federalism hypothesis does not. In models without controls, the percentage of seats held by women is significant at better than the 1% level. The local coverage variable is also significant at the 5% level in these models.

However, when we add the seven control variables, the coefficient on the local coverage variable becomes indistinguishable from zero, whereas the percent of legislators who are female remains significant at better than the 1% level. Among the control variables, real GDP per capita, population size, union density, and female labor force participation have no discernible relationship with the outcome. Unified Democratic control of the state government (a trifecta) is strongly associated with a greater likelihood of passing PSL. The variable for possible geographical effects in PSL adoption is significant at the 10% level. We also include a variable for the professionalization of the state legislature. The effect of the professionalization index is positive at the 5% level: states with more professional legislatures are more likely to pass PSL.

Percent female, a variable measuring the share of seats held by women, is substantively and statistically significant in all permutations of these models. The coefficients are very similar in size and significance in models featuring any and all combinations of subsets of the control variables. We experimented with different versions of the female percentage of seats variable, and a linear additive specification is preferred. There is no evidence of a nonlinear relationship (tested with the inclusion of squared and cubic terms), or of threshold/critical mass effects, or of interactive effects with other variables. Moreover, we sequentially dropped each state that has passed PSL legislations from the analysis, and the results remained unperturbed (that is, there do not appear to be any influential outlying observations – see the Supplementary Materials, Appendix for further discussion). The effect of the female representation variable is strong and robust. By contrast, the positive result for substate coverage disappears in the full model. Why? The inclusion of any one of three of the control variables renders local coverage indistinguishable from zero: GDP per capita, population size, or Democratic control of government. We surmise that these are functioning as confounders, with each related to both local coverage and the likelihood of passing PSL; after including any of the three, the relationship between local coverage and PSL accordingly disappears.

To explore further the relationship between women in politics and the passage of PSL legislation, in Table 2, we report the models featuring a range of indicators of female prominence in state institutions, including two “placebo” indicators that should have no direct influence on the passage of PSL legislation but should be correlated with the general “progressiveness” of a state. All the models reported in Table 2 include the control variables from Table 1; however, we do not report their coefficients in the table to focus on the gender variables.⁶ First, in model (1), we add a variable for the gender of the governor. Governors play a gatekeeping role for the passage of legislation via their veto powers, as well as often driving legislative initiatives. In accordance with our argument about substantive representation, the gender of the governor should matter. In addition, indeed, a female governor is associated with the passage of PSL legislation at better than the 5% level. Moreover, even though the gender of the governor is highly significant, the effect of female percentage of the legislature remains both substantively and statistically significant when this variable is included. In columns 2 and 3, we separate out the lower and upper houses of state legislatures.⁷ Having a larger proportion of women in either chamber increases the likelihood of PSL passage, although the substantive effect is smaller than when there is a high percentage in both chambers. We then create a variable for a female “trifecta,” which we define as a female governor plus a high level of representation of women in both chambers (we define “high” as being in the top 25th percentile of observations for percentage of women). This variable is highly significant at better than the 1% level. Having more women in any key policymaking position matters, but having more women in all of the key positions matters most.

Table 2. Unpacking the presence of women in state institutions

Notes: Entries are coefficients with robust standard errors clustered by state in parentheses from Cox proportional hazard models. *p < .1, **p < .05, ***p < .01.

The variables featured in models 1–4 of Table 2 all measure female representation in key policymaking positions in the process of proposing and passing PSL legislation. However, not all state-level positions have a causal link to the making of social legislation. For example, whereas the governor plays a key role in the legislative process, lieutenant governors typically play little, if any, role in the legislative process; their main function is to fill in for the governor if she becomes incapacitated or is out of the state. However, important positions, such as lieutenant governor are nonetheless probably more likely to be filled by women in more progressive states compared to less progressive states. This creates the possibility of a placebo test: although a female lieutenant governor may proxy for progressive politics by indicating that a state’s electorate is open to women in high offices, it should have no direct influence on the policy process. The same argument applies to state attorney general, whose role as chief legal officers of the state should have no direct influence on the policy process. Accordingly, in columns 5 and 6, we report the results of including variables for a female lieutenant governor or attorney general, treating them as “placebo” variables.⁸ As we would expect from a placebo treatment, neither has any discernable relationship to the likelihood of passing PSL laws. When women are in positions with a direct causal link to social legislation, such as the governorship or state legislature, the likelihood of passing PSL laws increases; however, when women are in positions of power with no link to the policy process, there is no discernable effect. If the election of women were merely proxying for the “progressiveness” of a state, we would not expect to see this pattern. The non-result for the “placebo” variables strengthens our confidence that the relationship between women in power and PSL is causal.

We further explore the possibility that strong female representation in the state legislature and the adoption of PSL are both products of liberal ideology in the model reported in column 1 of Table 3. In our baseline models reported in Table 1 (and all subsequent models), we included a variable for Democratic control of state government; the Democratic party both fields more female candidates than the Republican party and is also associated with progressive policies. Here, as an alternate control for the ideology of states, we draw on Richard Fording’s updated data on government and citizen ideology at the state level, a measure originally introduced in Berry, Ringquist, Fording and Hanson (1998). These data only go up to 2017 for government ideology and 2016 for citizen ideology, causing the loss of 161 observations and several positive cases of PSL adoption; for this reason, we did not include it in the models in Table 1.⁹ Column 1 reports the results using government ideology as measure of the “progressiveness” of a state’s politics. As expected, more liberal ideology is strongly related to an increased likelihood of PSL. Nonetheless, the female representation variable is still significant at better than the 1% level, and the estimated coefficient increases when the ideology measure is included. Results are very similar if citizen ideology is substituted for government ideology, although with the loss of another year of data.

Table 3. Additional variables and observations

Notes: Entries are coefficients with robust standard errors clustered by state in parentheses from Cox proportional hazard models. *p < .1, **p < .05, *** p < .01.

As a final strategy for evaluating the sensitivity of our results to unobserved confounders (with progressive ideology chiefly in mind), we followed the sensitivity analysis developed by Oster (2019). Oster’s procedure estimates how strong any omitted confounders would have to be, relative to included controls, to account for the estimated association between female representation and the passage of PSL.¹⁰ Oster suggests a threshold of 1.0 for the key parameter δ, with values smaller than that indicating cause for concern. Running the test for models (1) and (2) from Table 1 yields δ values of 4.85 and 2.24, indicating that it is very unlikely unmeasured confounders would eliminate the relationship between female representation and PSL.

We explore other alternative variables and specifications in Table 3. In the baseline models, we included labor force participation as a proxy for demand for PSL. Given that low-income workers disproportionately lack sick leave, the percentage of state workers earning the minimum wage might be a better proxy for demand. Model (2) in Table 3 reports the result; a higher percentage of minimum wage workers is related to the likelihood of PSL passage, but negatively, not positively. This may be because a higher percentage of minimum wage works also correlates with the size of the small business and fast-food sectors. As our case studies show, small business and fast-food lobbying organizations are typically the strongest opponents of PSL laws. If this variable is proxying for the strength of those sectors, then the negative coefficient makes sense.¹¹

Next, we explore the possibility of an interaction effect between legislative professionalism and the local coverage variable. Shipan and Volden (2006) suggest that a learning effect from the experience of municipalities is more likely to occur in states with a professional legislature. Surprisingly, the interaction effect is negative, the opposite of expected. This pattern suggests that legislative professionalism and local coverage are functioning as substitutes, not complements. However, an exploration of possible outliers reveals that this result is driven by Washington state; when Washington is removed from the sample, neither the interaction term nor the local coverage term is significant. Washington is one of the states that passed PSL by initiative, making it less likely that legislative professionalism played a role in the passage. Without this one state, there is no evidence of either a local coverage effect or an interaction.

We explored further the sensitivity of the results to individual states in a series of tests. First, as column 4 in Table 3 reports, we added Washington, D.C. to the sample. Although not a state, Washington, D.C. is prominent in the history of PSL in the United States, having passed PSL in 2008, and features an executive-legislature government structure in which the Council of the District of Columbia functions as an equivalent of a state assembly. The addition of the District of Columbia makes no substantial difference for results. On the other hand, exploration of possible outliers using DFBETAS flags Arizona as a possible influential observation for the female percentage variable. Accordingly, in the model reported in column 5, we exclude Arizona. The coefficient on female representation shrinks a bit, but it is still significant at better than the 1% level. In addition, we deleted every state which passed PSL in turn, and all three initiative states together (a result reported in the Supplementary Materials, Appendix), and the female percentage of legislature remained significant at better than the 1% level.

Case narratives

To examine the mechanisms by which PSL passed, we created case narratives for all 15 states. Although our focus was primarily on the role played by female legislators and the possible relevance of local laws, several other patterns emerged quite strongly from our investigation of the process by which PSL was passed. First, in nearly every case, the coalition pushing for PSL included female-focused NGOs, such as 9to5, MomsRising Together, or Planned Parenthood, alongside labor organizations. In Washington, endorsements for Initiative 1433, which passed in 2016, came from a range of organizations, including the League of Women Voters, the Washington Academy of Family Physicians, and the Children’s Alliance. It was also supported by labor unions and workers’ rights advocates (Faught 2016). Similarly, in Maryland, which enacted PSL in 2018, support for the measure came from groups like the United Food and Commercial Workers Union as well as the Women’s Law Center of Maryland, and in Colorado 9to5 has long been at the forefront of the PSL movement.

Whereas social justice organizations focusing on women’s and family issues and unions were uniformly a part of pro-PSL coalitions, small business associations, fast food associations, and chambers of commerce were almost universally opposed to these measures. The National Federation of Small Businesses and state Chambers of Commerce opposed PSL in all 15 states, with a notable role often played by restaurant associations (In Rhode Island, Dunkin’ Donuts of Rhode Island was particularly active in opposing PSL). For example, opposition in Washington was spearheaded by the Washington Restaurant Association and the Washington Retail Association. Although research shows that PSL policies have minimal impact on employment and wages (Ahn and Yelowitz 2015; Pichler and Ziebarth 2020), these groups argued that the measures would be a detriment to employers, particularly small businesses. This opposition sometimes crossed party lines, particularly in Rhode Island. Democratic State Senator Leonidas Raptakis, the owner of a pizza restaurant, voted against the state’s PSL legislation on the grounds that it would harm small businesses.¹²

In all the states that have adopted PSL, female legislators played a major role in promoting the cause of PSL laws.¹³ In 12 of the 15 states, women were lead sponsors, or accounted for more than half of the lead sponsors, when PSL legislation was introduced into the state legislature. Nearly every state that ultimately passed PSL laws saw at least one earlier, ultimately unsuccessful, attempt, and these initial efforts were also disproportionately led by female politicians. For example, even though PSL was eventually passed by an initiative in Massachusetts in 2014, the issue first rose to prominence through the repeated introductions of PSL legislation, starting in 2005, by Representative Kay Kahn and Senator Patricia Jalen.

In six of the states that passed PSL, substate jurisdictions had already passed their own PSL policies: California, Maryland, New Jersey, New York, Oregon, and Washington. In these cases, including, in particular, New Jersey, debates about whether state law should override local laws sometimes pitted advocates of the policy against each other, with some supporters wanting to allow local jurisdictions the freedom to enact more generous standards, whereas others wanted to achieve harmonization across the state. In California, New York, Oregon, and Washington, the PSL laws do not preempt local ordinances. In California and New York, employers are to follow whichever regulation is more generous. In Oregon, a special exception was permitted for Portland’s PSL regulations, and Seattle’s slightly more generous provisions are permitted under Washington’s PSL law. In contrast, New Jersey and Maryland’s PSL mandates both preempted local ordinances as a way to make their policies uniform across the state. In most states passing PSL laws, advocates of local flexibility in benefit levels won out over those hoping to harmonize standards across the state.

In the following sections, we further detail the process of PSL adoption in New Jersey, which featured the most interesting dynamics about local-level jurisdictions, and in the states that ultimately passed PSL by initiative, with a focus on Arizona, which provides an example of how, in the initiative states, blockage at the statehouse of strong momentum toward PSL diverted efforts to the ballot box.

New Jersey

Efforts to pass PSL in the Garden State began in 2014, and a bill sponsored by State Senator Loretta Weinberg (D-Bergen) was introduced the following year.¹⁴ The bill, S785, provided that “workers would accumulate an hour of sick time for every 30 worked. Companies with 10 or fewer workers would have to allow workers to earn at least 40 hours of sick leave. Those with more than 10 would have to offer at least 72 hours” (Friedman 2015). By the time the bill was introduced in the Senate, eight municipalities in New Jersey had already passed their own versions of sick leave, including the state’s two largest cities: Jersey City and Newark. A ninth municipality, New Brunswick, would also adopt a paid leave policy in July of 2015. In addition to opposition from the business community, some supporters of PSL objected to the bill’s provision to preempt local laws. Analila Mejia, executive director of the New Jersey Working Families Alliance, argued that by applying uniform standards across the state, the law would undercut municipalities’ ability to respond to the local community (Friedman 2015). In contrast, bill sponsor Loretta Weinberg argued that preemption was necessary to provide uniformity of business conditions “so there aren’t 562 separate local laws.”

Disagreements between the New Jersey Senate and Assembly over the bill’s preemption of local PSL provisions derailed this first effort, and S785 failed to advance during the 2015 legislative session. Similar efforts during the 2016–2017 legislative session also came to naught. In a pattern typical of debates over PSL legislation, many business groups in New Jersey lobbied strongly against the measure, even after amendments were made to accommodate their concerns (Symons 2014). Even if PSL legislation were to pass the Senate and Assembly, it was widely expected that Governor Chris Christie (R) would have vetoed it. Christie vetoed an expansion of paid family leave passed by the state legislature in 2017 (A4927), citing the additional cost to businesses in the state (Marcus 2017).

The 2017 gubernatorial election ushered in Democrat Phil Murphy, unblocking a pathway toward progressive legislation. A renewed PSL measure (A1827) was introduced in the Assembly by several sponsors and cosponsors, including a leading role by Pamela Lampitt (D-Camden),¹⁵ with the effort in the Senate again spearheaded by Senator Loretta Weinberg. Lampitt argued that “guaranteeing workers the ability to earn paid sick days would help ensure workers do not have to choose between their health and their economic security” (quoted in Livio 2018). This proposal was passed in both chambers and signed by Governor Murphy in May 2018, going into effect in October of that year, with differences over preemption resolved by an agreement to harmonize PSL across New Jersey at a level higher than any local jurisdiction (Ferinot 2018). The final text of the law mandates that employees receive 1 hour of paid leave for every 30 hours worked, with a maximum of 40 hours earned in a year. The new law standardized New Jersey jurisdictions at a more generous level of provision than that provided by any municipality, thus satisfying organizations like the New Jersey Working Families Alliance, who had opposed the initial action because it superseded municipal laws that, in some cases, were more generous than the proposed state ordinance.¹⁶

By the time the PSL law was passed in 2018, 13 municipalities in New Jersey had already adopted local paid leave regulations, covering about 14.6% of the state’s population. At the municipal level, Jersey City was the first to pass a local policy in 2013, with Newark following the next year. From this perspective, it does appear to have been a cascading effect, with seven other localities passing paid leave legislation in 2015, three in 2016, and one in 2017. Although there is no systematic relationship between local coverage and passage of PSL, complicated bottom-up federalism dynamics were clear in the New Jersey case. In some ways, the existence of municipal laws made passage more difficult, as some natural supporters of PSL opposed the state provision when it appeared that it would achieve uniformity at a lower benefit level. On the other hand, supporters such as Loretta Weinberg were clearly motivated by the desire to achieve uniform standards across the state economy, and the drive toward harmonization ultimately won out. In the New Jersey case, the pathway to passage may have been rendered more difficult, but the ultimate result was a more generous state law than initially proposed as the local laws acted, as something of an upward ratchet on the generosity of statewide provisions.

Initiative states

In three of the 15 states that currently have sick leave provisions, voters at the ballot box were the ones who approved the policies. In all three initiative states, female legislators took the lead in the earlier efforts to pass PSL: Representative Laurie Jenkins in Washington, Representative Kay Kahn and Senator Patricia Jehlen in Massachusetts, and Senator Katie Hobbs in Arizona. In all three states, blockages by Republicans controlling a veto point and strong lobbying by the business community caused the effort to spillover from the statehouse to initiative drives. In Washington, for example, HB 1356 was able to pass the House and enjoyed the support of Democratic Governor Jay Inslee, but was stalled in the Republican controlled Senate where the bill was introduced and retained five times in 2015 and 2016. Massachusetts was the first state to mandate PSL through a ballot initiative in 2014, and both Arizona and Washington passed measures in 2016 elections.

The case of Arizona is emblematic of these three states: although Democrats had been making steady inroads in state politics, the state legislature, and governorship remained with Republicans. In terms of female state legislators, Arizona has hovered in the low to mid-30s since 2000, and in 2016 was at 36%. No local jurisdictions in Arizona passed PSL, although in 2015, Councilwoman Regina Romero introduced a PSL ordinance in Tucson. Responding to the threat of such ordinances, Republican Governor Doug Ducey signed into law a bill that restricted local jurisdictions’ ability to implement measures like PSL, providing an example of the preemption of more generous benefits.

Also in 2015, a PSL bill introduced in the Arizona legislature failed to advance despite polling showing that a majority of voters supported the policy. With no prospect of overturning Republican control of state government in the short term, advocates turned to the initiative process. In a pattern of support typical for PSL legislation, unions, education groups, and women’s advocacy groups led the coalition supporting Prop 206, alongside strong support from prominent politicians who had seen their efforts in the legislature blocked. Many state legislators actively supported Prop 206, as did the mayors of Phoenix and Tucson. Leaders of stalled legislative efforts, such as Senate Minority Leader Katie Hobbs (D-Phoenix) and Representative Celeste Plumlee (D-Tempe), were prominent in the initiative campaign. Writing in support of the initiative, Senate Minority Leader Katie Hobbs noted that the initiative “will impact over a million hard-working people in our state. It will lift up hard-working Arizonans, building healthy and strong families” (Arizona Secretary of State 2016). US Representative Ann Kilpatrick also played a prominent role in the Prop 206 campaign.

Patterns of opposition to Prop 206 were also typical of battles over PSL in other states. “The state’s most influential business organizations, representing large corporations, small businesses and restaurants, opposed” the measure (Rau 2016). As is typical, the chief among these organizations were the Arizona Chamber of Commerce and the Arizona Restaurant Association.

In November 2016, Arizona voters approved Proposition 206, the Fair Wages and Healthy Families Act with a margin of 58.3–41.7%. The measure provides that employees of businesses with more than 15 employees accrue one hour of sick leave for every 30 hours worked. Employees can earn up to 40 hours of PSL per year. Employees in businesses with fewer than 15 employees must also accrue one hour of leave for every 30 hours worked; however, employers at small firms are only required to provide 24 hours of PSL per year.

Conclusion

The COVID-19 pandemic focused attention on PSL policies as a key part of the social safety net. The effects of PSL policies in slowing the spread of disease, as well as easing work-family balance tradeoffs, are well-established in epidemiological and sociology research. An increasing body of economics research suggests that these benefits come with few, if any, negative labor market effects. Given the importance of this issue, political scientists’ relative lack of attention to the politics of PSL is disappointing. Empirically, we know the effects of PSL policies, but we do not know their causes. We begin to rectify this situation both theoretically and empirically. Empirically, we have created a new dataset of all state and municipal PSL legislation in the United States, and conducted the first systematic analysis (to the best of our knowledge) of the adoption of PSL policies across US states. Theoretically, we build on the descriptive representation literature and the “bottom-up federalism” literature to offer hypotheses about the spread of PSL. We argue that the likelihood of adopting PSL policies increases as the share of seats in the state legislature occupied by women increases. Specifically, leave policies will be placed on the agenda as larger numbers of women are represented in state legislatures. We also argue that a “snowball effect” is likely to occur when municipal governments begin passing PSL at the local level, so the likelihood of the passage of state legislation increases with the share of the population covered by municipal leave policies. Empirically, we find strong and robust support for the substantive representation hypothesis, whereas the snowball effect does not appear to operate with any regularity in this policy area.

Our work has implications for both research and policy advocates. We advance the research on PSL policy at the subnational level, by utilizing both cross-sectional and longitudinal variations across US states. This allows us to pin down the relationship between women in legislatures and PSL by showing that changes in legislation tend to follow the expansion of the number of seats occupied by women. For the descriptive representation literature, we show that having women in politics is important not only for placing women’s issues on the agenda, but also for translating those preferences into policy outcomes. Much of this research focuses on the activities of female politicians; we show that as larger numbers of women enter legislatures, they have a collective impact on policy outcomes. Finally, our research suggests that policy advocates hoping to encourage the passage of more generous PSL legislation should pay attention to the pipeline of prospective candidates and new politicians. Having a substantial bloc of women in Congress may be something akin to a necessary condition for the passage of progressive PSL policies; getting a larger number of women into politics by encouraging their entry into the candidate pipeline is a crucial first step to fulfilling that condition (Thomsen 2017). The ongoing surge of women into Congress, with a record 28% of seats held by women in the 118th Congress, may eventually open the door to the US joining its OECD peers in adopting national-level PSL policies.

Our research also points to an area warranting further attention. Those who work in the growing gig economy are less likely to receive the same benefits as those who work in either full- or part-time employment. Because many of these workers are considered independent contractors or freelancers, companies may not be required to provide the benefits associated with other forms of employment, and they are typically not covered by the new PSL laws. This had led to a situation that Kathleen Thelen (2019) refers to as the rise of the “precariat,” as the typical employment arrangement, which used to include longer term contracts with benefits and predictable hours, breaks down and employees are increasingly subject to greater levels of risk that would have previously been taken on by employers. This reflects a general trend in the US labor market, wherein more risks are shifted from the employer to the employee (Hacker, 2004). The availability of sick leave for those in more secure employment positions reflects the dualism of the labor market more broadly, where “workers are divided between those with permanent contracts that include valuable benefits and extensive labor market protections and those who work under contingent contracts or no contract at all. This latter group receives few or no labor market protections and lower levels of social benefits” (Rueda et al. 2015, 89). Even as PSL is gradually extended to a larger number of US employees via state-level legislation, a rapidly increasing segment of the job market is being left behind.