Occupational welfare: economic and normative explanations

The occupational welfare literature tends to explain employer provision as economically motivated, i.e., as the result of profit-maximising cost-benefit analyses (Greve, Reference Greve2007, Reference Greve2018; Farnsworth, Reference Farnsworth2013; Sinfield, Reference Sinfield2013). This perspective can be summarised as a rational–choice theoretical approach (DeMenezes & Kelliher, Reference DeMenezes and Kelliher2011; Estévez-Abe et al., Reference Estévez-Abe, Iversen, Soskice, Hall and Soskice2001; Hart, Reference Hart2010). In this view, employers provide occupational welfare to increase profits by improving hiring efficiency or retention of highly skilled employees (Naczyk, Reference Naczyk, Natali, Pavolini and Vanhercke2018; Riva & Rizza, Reference Riva and Rizza2021), increasing productivity (Beauregard & Henry, Reference Beauregard and Henry2009; Konrad & Mangel, Reference Konrad and Mangel2000), or preventing costly confrontations with organised labour (Seeleib-Kaiser, Reference Seeleib-Kaiser2008; Trampusch, Reference Trampusch2013). Rational–choice theory is the dominant theoretical approach in studies of occupational welfare by sector or industry (Den Dulk & Groeneveld, Reference Den Dulk and Groeneveld2012; Wiß, Reference Wiß2015), employee skills and shortages (Berg et al., Reference Berg, Appelbaum, Bailey and Kalleberg2004; Mares, Reference Mares, Hall and Soskice2001), number of employees (Farnsworth, Reference Farnsworth and Greve2018; Osterman, Reference Osterman1995), degree of unionisation (Budd & Mumford, Reference Budd and Mumford2006; Jacobi, Reference Jacobi2023) and share of female employees (Kelly et al., Reference Kelly, Moen and Tranby2011; Seyler et al., Reference Seyler, Monroe and Garand1995).

Of special interest to social policy scholars is the relationship between the pillars in the social division of welfare (Mann, Reference Mann2009; Titmuss, Reference Titmuss1956). However, particularly in explaining the relationship between social and occupational welfare, rational-choice theoretical explanations are increasingly insufficient. These predict that employers provide little support in polities with generous social provision, as this is not an efficient use of employer resources, and because public provision decreases economic pressure on employers vis-à-vis their employees. This mechanism – greater public provision in conjunction with less occupational welfare – is referred to as ‘crowding-out’ (Etzioni, Reference Etzioni1995; Wiß & Greve, Reference Wiß and Greve2020). However, the empirical evidence increasingly suggests that generous social welfare is often associated with widespread occupational welfare (Natali et al., Reference Natali, Keune, Pavolini and Seeleib-Kaiser2018), a dynamic termed ‘crowding-in’ (Chung, Reference Chung2019; Wiß & Greve, Reference Wiß and Greve2020). This is especially true in studies examining the relationship between public and employer family policies (Den Dulk et al., Reference Den Dulk, Peters and Poutsma2012, Reference Den Dulk, Groeneveld, Ollier-Malaterre and Valcour2013; Mun & Jung, Reference Mun and Jung2018). Here, neoinstitutionalist explanations of employer behaviour are becoming increasingly relevant.

Neoinstitutionalism explains employer-family policy as the organisational response to normative pressures, such as changing perceptions of what is an appropriate level of support employees can expect from their employers (Lewis & Smithson, Reference Lewis and Smithson2001; Powell, Reference Powell, Powell and DiMaggio1991). The theoretical premise is that the availability of generous public family policies legitimises employees’ claims for support (Scott, Reference Scott, Powell and DiMaggio1991; Tomaskovic-Devey & Avent-Holt, Reference Tomaskovic-Devey and Avent-Holt2019). This creates normative pressure for employers to offer family policies to gain normative legitimacy from their environment (DiMaggio & Powell, Reference DiMaggio and Powell1983, Reference DiMaggio and Powell1991). Neoinstitutionalist perspectives better explain the evidence suggesting that public family policy crowds-in employer family policy in Europe (Chung, Reference Chung2019; Den Dulk et al., Reference Den Dulk, Peters and Poutsma2012; Wiß & Greve, Reference Wiß and Greve2019).

The relevance of the crowding-in dynamic has so far rarely been applied to analyses of the United States (for an exception, see Daiger von Gleichen, Reference Daiger von Gleichen2022), most likely because public family policy is far less generous than in Europe (Daiger von Gleichen & Seeleib-Kaiser, Reference Daiger von Gleichen, Seeleib-Kaiser and Shaver2018; O’Connor et al., Reference O’Connor, Orloff and Shaver1999). Yet there has been considerable growth in U.S. public family policy at the sub–federal level, especially in the realm of childcare (Cohen-Vogel et al., Reference Cohen-Vogel, Sadler, Little, Merrill and Curran2020; Daiger von Gleichen & Parolin, Reference Daiger von Gleichen and Parolin2020; Friedman-Krause et al., Reference Friedman-Krause, Barnett, Garver, Hodges, Weisenfeld, Gardiner and Jost2022). This suggests that in the United States, where rational-choice explanations for occupational welfare dominate (Hacker, Reference Hacker2002, Reference Hacker2006; Lynch, Reference Lynch, Béland, Morgan and Howard2014), normative explanations might be underappreciated. Moreover, a burgeoning literature suggests that female leaders – who are particularly prevalent in the United States – are increasingly driving normative change in organisations.

Female leaders: from cogs in the machine to change agents

In this journal, Kowalewska (Reference Kowalewska2020) argued that analyses of the gendered character of welfare states should include considerations of women’s representation in corporate leadership. This argument was born out of a shift in how the sociological literature views the role of women leaders in organisations. Female leaders have been ‘cogs in the machine’, i.e. displaying masculine traits (Miliopoulou & Kapareliotis, Reference Miliopoulou and Kapareliotis2021), bias in favour of men (Dutton, Reference Dutton2018), or lacking power to affect gender-positive change (Kanter, Reference Kanter1977).

Increasingly, however, research suggests that women leaders can be effective ‘change agents’ (Fuwa, Reference Fuwa2021; Kelan & Wratil, Reference Kelan and Wratil2018), i.e. willing and able to weaken gender–biased aspects of organisational culture and procedures (Gould et al., Reference Gould, Kulik and Sardeshmukh2018; Jouber, Reference Jouber2022; Khushk et al., Reference Khushk, Zengtian and Hui2023). This is evidenced when female leaders are not anomalies (Miliopoulou & Kapareliotis, Reference Miliopoulou and Kapareliotis2021) but part of a critical mass of women leaders (Kowalewska, Reference Kowalewska2020, Reference Kowalewska2021). While acknowledging the economic power they wield, the research stresses female leaders’ ability to apply normative pressure for change (Gould et al., Reference Gould, Kulik and Sardeshmukh2018; van Mensvoort et al., Reference van Mensvoort, Kraaykamp, Meuleman and van den Brink2021).

The role of change agents is an important aspect of neoinstitutionalism, which posits that claims-makers within organisations leverage normative legitimacy bestowed by generous public policy (Sauer et al., Reference Sauer, Valet, Shams and Tomaskovic-Devey2021; Tomaskovic-Devey & Avent-Holt, Reference Tomaskovic-Devey and Avent-Holt2019). In this view, employer policy is the result of normative, not economic pressure. Studies of female leadership and employer family policy (Flüter-Hoffman & Seyda, Reference Flüter-Hoffman and Seyda2006; Heywood & Jirjahn, Reference Heywood and Jirjahn2009) explain the generally positive associations as a response to both economic and normative pressures (Ingram & Simons, Reference Ingram and Simons1995; Pasamar & Alegre, Reference Pasamar and Alegre2015; Wiß, Reference Wiß2017).

The acknowledgement that both pressures are at play is the theoretical frontier of female leadership and employer family policy. This is likely because the predicted effects of both forms of pressure are the same: both normative and economic pressure make employers more likely to offer family policy. The contribution of this study is to show a case when the association between female leadership and employer provision is better explained by normative versus economic pressures. To do so, the study exploits the landscape of public sub-federal and employer childcare in the United States.

Public and employer childcare in the U.S., and hypotheses

Public family policy in the United States is liberal in terms of targeting low-income families (Gornick & Meyers, Reference Gornick and Meyers2003; O’Connor et al., Reference O’Connor, Orloff and Shaver1999), and financial assistance to families is ungenerous in comparison to Western European countries (Daiger von Gleichen & Seeleib-Kaiser, Reference Daiger von Gleichen, Seeleib-Kaiser and Shaver2018; Woods, Reference Woods2012). Moreover, the United States is the only OECD country that does not have paid maternity leave at the national level (Engeman, Reference Engeman2020). The United States has thus been labelled a ‘laggard’ in public family policy expansion (Daiger von Gleichen & Seeleib-Kaiser, Reference Daiger von Gleichen, Seeleib-Kaiser and Shaver2018).

Nevertheless, several dimensions of public U.S. family policy have expanded in recent decades. The most prominent has been paid leave, currently offered in eleven states (Engeman, Reference Engeman2020). But state-level public childcare expansion has also been extensive. As of 2020, seventeen states had so-called ‘universal’ pre-kindergarten programs in which age and residency were the only requirements for eligibility (Stanford, Reference Stanford2023). An additional thirty-four states offer programs with eligibility tied to an income cap (Friedman-Krause et al., Reference Friedman-Krause, Barnett, Garver, Hodges, Weisenfeld, Gardiner and Jost2022). This leads to the first hypothesis:

Furthermore, there are two categories of employer childcare in the United States that differ by how they can be conceptualised as responses to economic versus normative pressures. FSAs are economically efficient in that they are both low-cost (Galinsky et al., Reference Galinsky, Bond, Sakai, Kim and Giuntoli2008; Kelly, Reference Kelly2003) and targeted towards high–skill, high-income employees (Hipp et al., Reference Hipp, Morrissey and Warner2017). They enable employees to annually devote up to $5,000 pre-tax dollars, administered by their employer, to childcare expenses. Yet setting up an FSA requires that the full sum for the whole year be paid up front, which deters low-income employees from participation (Hipp et al., Reference Hipp, Morrissey and Warner2017). The low cost to employers and greater take-up by high–income employees suggests that FSAs are likely a response to economic pressure.

In contrast, other forms of employer support for and provision of childcare tend to be both highly expensive for employers (Ratnasingam et al., Reference Ratnasingam, Spitzmueller, King, Rubino, Luksyte, Matthews and Fisher2012; Ruddy, Reference Ruddy2020) without being especially attractive to high-skill employees (Feldman & Schultz, Reference Feldman and Schultz2001; Morrissey & Warner, Reference Morrissey and Warner2009). For example, childcare centres have high fixed costs (Feierabend & Staffelbach, Reference Feierabend and Staffelbach2016), and take-up is stronger among lower–income employees (Hipp et al., Reference Hipp, Morrissey and Warner2017; Morissey & Warner, Reference Morissey and Warner2009). Childcare subsidies are furthermore tied to specific childcare providers, limiting parental choice (Sosinsky, Reference Sosinsky, Lloyd and Penn2012), a criterion more determinate of high- as compared to low-income families’ use (Shlay et al., Reference Shlay, Tran, Weinraub and Harmon2005). Such non-FSA forms of childcare are thus less likely a response to economic, but rather to normative pressure on employers. This leads to the following hypothesis pair:

Data, measures, and method

Measures

Dependent variables

The NSE queried respondents about seven forms of employer childcare: information to help locate childcare, childcare vouchers or subsidies not funded by payroll taxes but by the employer directly, childcare at or near the work site, childcare during school vacations, back-up when regular childcare fails, childcare for sick children, and FSAs that pay for childcare with employees’ pre-tax income matched by employer subsidies from payroll tax deductions. Of these seven forms of childcare, six were included in the analysis. Information to help locate childcare was excluded because this sets a very low bar for employer involvement, including for example a simple internet search or mentioning the existence of a nearby crèche. This low cost to employers combined with uncertainty as to how valuable such information is to employees (who can likely obtain it themselves) made it theoretically ambiguous, which is why it was excluded.

From the remaining six categories, I created two binary dependent variables, operationalised as ‘not provided/provided’: FSAs and childcare other than FSAs (‘non-FSAs’). The variable FSAs captures whether employers provide flexible spending accounts, while the variable non-FSAs captures whether the employer provides any of the other five forms of childcare. Thus, the dependent variables capture overlapping samples: employers providing FSAs may or may not additionally provide non–FSA forms of childcare, and vice versa. Correlation between the two forms of childcare is not significant. The distributions are shown in Table 1.

Table 1. Descriptive statistics of dependent variables

Note: Unweighted data. Data weighted for employer size suggest that FSAs are provided by 56%, non-FSAs by 13% of employers in the United States.

Independent variables

The explanatory variables are state-level public childcare, the prevalence of female employees, and industry skill profile. State-level public childcare is binary, indicating whether the state in which the employer is located offers public pre–kindergarten not subject to a family income cap, but accessible to children with the appropriate age and residency requirements independent of family income. This operationalisation ensures that public pre–kindergarten represents an option for employees irrespective of income. The prevalence of female leaders is coded as an ordinal variable, the only one provided by the NSE. The categories are ‘no female executives,’ ‘women on some executive levels,’ and ‘women on all executive levels.’ The skill–profile variable is binary, categorising employers’ industry as ‘high-general skills’ or ‘not high–general skills,’ the latter combining low–general and mixed-skills industries.Footnote ⁴

At the organisational level, the models control for the number of employees, difficulty filling vacancies, share of female employees, and presence of unionised employees. To capture additional state-level distinctions, the models control for state GDP per capita, state-level female labour force participation, and an attitudinal variable capturing the percentage of people per state who prefer a male breadwinner model. Table 2 shows the distributions of the explanatory variables, and those for the control variables are given in Table A3 of the appendix.

Table 2. Descriptive statistics of explanatory variables for each model

Findings

The findings displayed in Table 3 show repeatedly that patterns of FSA and non–FSA provision are governed by different logics. Public childcare without an income cap is associated with a 5% greater likelihood that an employer will provide non-FSA childcare but is insignificant for FSAs. This confirms H₁ regarding non-FSAs, but not FSAs. Similarly, employers dependent on high-general skills (HGS) are 13% more likely to provide FSAs, whereas HGS are not predictive of non-FSA provision (confirming H₂). Meanwhile, female executives are positively associated with both forms of employer childcare (confirming H₃), but the effect size differs by policy depending on the prevalence of female leaders across executive levels. Specifically, employers with women on some executive levels are 12% more likely to offer FSAs and 6% more likely to offer non-FSA childcare than employers lacking female executives. In comparison to employers lacking female executives, those with women on all executive levels are 16% more likely to offer FSAs and 19% more likely to offer non-FSAs.Footnote ⁶

Table 3. Average marginal effects based on logistic regression models

Note: Original calculations. Models control for number of employees, presence of unionised employees; as well as state-level gdp per capita, female labour force participation and gendered breadwinner/carer preferences. Full models presented in Table A6 of the appendix. *p < 0.05, **p < 0.01, ***p < 0.001.