Welfare Spending and Female Labor Force Participation

Labor force participation is essential for economic security, yet women around the world continue to face challenges to active employment. On average, women are more likely to engage in unpaid work, earn less than men, be underrepresented in senior decision-making positions within companies, encounter uneven barriers to entrepreneurship, and find themselves in vulnerable employment in most countries (Iversen and Rosenbluth Reference Iversen and Rosenbluth2006; UNDP 2015). Some of the challenges to women's equal participation in the labor force stem from gendered assumptions about care. Cultural norms and expectations about who should provide care for children and other household members result in women more frequently providing care within families (Abramovitz Reference Abramovitz1996; Fraser Reference Fraser1987; Rosenbluth, Salmond, and Thies Reference Rosenbluth, Salmond and Thies2006).

These gendered assumptions result in women on average performing three times more unpaid work than men, encompassing a range of care activities (UNDP 2015). This uneven care burden means that women spend a disproportionate amount of time providing care as compared to men, a trend which holds across both wealthy and poorer countries (Bittman et al. Reference Bittman, England, Folbre, Sayer and Matheson2003).Footnote ² This results in mothers having some of the lowest rates of participation in the labor force, even in states in the global North (Daly Reference Daly2000; Marino, Romanelli, and Tasso Reference Marino, Romanelli and Tasso2013). When childcare or elder care is supported by the state, women will have more time to participate in the labor force (Orloff Reference Orloff2002). Thus, one direct impact of the allocation of more funds to social programs could be an increase in the extent of women's paid employment in the total labor force. For instance, previous research suggests that social spending on childcare and family-related support in Scandinavian countries during the 1980s and 1990s was instrumental for many women to reconcile family obligations and have more paid employment outside of the home (Kolberg Reference Kolberg1991).

Similarly, public spending on healthcare both allows people to be healthy enough to work as well as relieves some of the care burden that many women face. Several studies have found that spending on health initiatives is associated with better health outcomes (Anyanwu and Erhijakpor Reference Anyanwu and Erhijakpor2009; Baldacci, Guin-Siu, and de Mello Reference Baldacci, Guin-Siu and de Mello2003; Farag et al. Reference Farag, Nandakumar, Wallack, Hodgkin, Gaumer and Erbil2013). For example, a study of 47 African countries found that higher amounts of health spending reduce infant and under-five mortality rates (Anyanwu and Erhijakpor Reference Anyanwu and Erhijakpor2009). Farag et al. (Reference Farag, Nandakumar, Wallack, Hodgkin, Gaumer and Erbil2013) had similar findings regarding improvements to infant and child health in their examination of public expenditures in 133 low- and middle-income countries. Specifically related to women's health, Frankenberg and Thomas (Reference Frankenberg and Thomas2001) illustrate that public financing of a midwife program in Indonesia between 1993 and 1997 was related to a significant increase in body mass index for women of reproductive age. Additionally, the presence of a village midwife during pregnancy was associated with increased infant birthweight. Both of these outcomes are regarded as beneficial for the mothers and children in the study.Footnote ³

Overall, the existing research indicates that public expenditures on health are positively associated with improvements to public health, particularly for children and pregnant women. Women's improved health conditions would enable them to become more active participants in the economy. Further, since women tend to disproportionately be central caregivers for children, healthier children would mean less demand on their time, which might be filled with paid employment in the labor force. Therefore, health spending likely contributes to healthier women and healthier children, both of which would enable women's labor force participation.

Better health also enables the sustained and effective use of the knowledge and skills that individuals acquire through education (Schultz Reference Schultz1999). Education is essential for gaining the skills necessary for individuals to be competitive in most job markets. Gupta, Verhoeven, and Tiongson (Reference Gupta, Verhoeven and Tiongson2002) show that both the level and the composition of spending on education are important determinants of enrollment rates, persistence rates through grade 4, and primary school dropout rates. Likewise, Baldacci et al. (Reference Baldacci, Clements, Gupta and Cui2008) find that an increase in education spending of just one percentage point of GDP is associated with three more years of schooling on average and raises the annual growth of GDP per capita by 1.4 percentage points in 15 years.

Female deprivation in many parts of the world has been likened to a “vicious cycle” wherein girls are not educated because they are not expected to contribute financially to a family (Summers Reference Summers1994, 1). Family resources are not dedicated to girls’ education, as they are not regarded as being a good investment for the economic prospects of the family. However, it is likely because they are uneducated that they cannot gain the skills necessary to access many jobs. This results in a self-fulfilling prophesy and the vicious cycle continues (Sabarwal, Sinha, and Buvinic Reference Sabarwal, Sinha and Buvinic2010; Summers Reference Summers1994). Public funding for education takes some of the burden off of families in paying for the education of both sons and daughters. More women may therefore be able to gain an education and the benefits that go along with it, including being more competitive for paid jobs. The skills that accompany education are a vital factor in whether or not many women will be able to enter the formal labor market (Edin and Lein Reference Edin and Lein1997).

In sum, we argue that public funding for welfare programs reduces some of the barriers to women's participation in the labor force. Social programs that lessen some of the care burden that many women face improve their ability to seek and hold jobs.Footnote ⁴ Additionally, spending on health and education help many women to be healthy enough to work and have skills that make them more competitive in the labor market. For the reasons outlined above, we offer the following hypothesis:

Welfare Spending and Women's Involvement in Politics

Women also tend to be underrepresented in politics, on average holding many fewer elected national positions than men. While globally there is a good deal of variation in women's presence in legislatures, states overwhelmingly fail to achieve gender equity in these bodies. According to the World Bank (2016), the global percentage of parliamentary seats in a single or lower chamber held by women has increased from 17% in 2006 to 22.9% in 2015, which is still a relatively low figure. Indeed, the United Nations General Assembly passed a resolution on women's political participation in 2011, which noted that “women in every part of the world continue to be largely marginalized from the political sphere, often as a result of discriminatory laws, practices, attitudes and gender stereotypes, low levels of education, lack of access to healthcare and the disproportionate effect of poverty on women” (UN General Assembly 2011, 2).

The last three factors in the UN's list—lack of education, access to healthcare, and economic security—are three of the areas in which welfare spending is often targeted. As discussed above, social spending aids women by removing barriers to their participation in public life. For instance, education spending enables women to gain skills and have expanded opportunities (Edin and Lein Reference Edin and Lein1997; Sabarwal, Sinha, and Buvinic Reference Sabarwal, Sinha and Buvinic2010; Summers Reference Summers1994). A basic example of this is the role of education in providing simple literacy skills necessary to stand for public office (Shvedova Reference Shvedova, Ballington and Karam2005). Existing work also shows that education is specifically connected to increases in political participation (Burns, Schlozman, and Verba Reference Burns, Schlozman and Verba2001). Education allows for training in the analysis of and participation in political debates. It also potentially includes engagement in voluntary opportunities like school organizations that may foster leadership experience, bureaucratic or organization skills, as well as cooperation and negotiation skills, each of which is useful for participation in national elections.

Additionally, education can provide networking channels that can manifest in political recruitment for national elections. There is some evidence that women's participation in education initiatives in particular can result in more active political engagement. Examples from Venezuela, Nicaragua, Uganda, and India illustrate that state-funded public education campaigns resulted in a sharp increase in the number of women participating in local and national elections, along with engaging in protests and participating in associations and organizations within civil society (Goetz Reference Goetz2003).Footnote ⁵ Hence, state spending on education initiatives increases the likelihood that women who participate in them might gain skills, confidence, or networks that can be translated into more active presence in national politics, including national parliaments (Shvedova Reference Shvedova, Ballington and Karam2005).

Likewise, spending that results in increased labor force participation and increased economic security is another important component of political participation. Poverty results in strains on resources to actively participate in politics. It is more challenging for people struggling with economic insecurity to participate actively in political processes (Berg-Schlosser and Kersting Reference Berg-Schlosser and Kersting2003; Holzner Reference Holzner2007). Studies find that, in particular, social security transfers and public health spending significantly reduce poverty (Brady Reference Brady2005). This means that women who face poverty may have some of their economic burden lessened through social spending, thus enabling them to assume more active roles in national-level politics. According to Razia Faiz, a former member of Parliament in Bangladesh,

While this does not apply to all women equally, it does indicate that the economic marginalization that many women face is a challenge to participation in national parliaments. Social spending that removes some of this economic marginalization, while also potentially facilitating network connections and other education benefits through education spending, can thus remove some of the significant barriers that women face to running for and holding seats at the national level.

Finally, public spending removes obstacles to women's involvement in the formal job sector by providing skills, healthcare, and care-supporting services. As public spending allows more women to participate in the labor force, particularly in well-organized public sectors, this increases the level of organization of women and their potential for political mobilization (Huber and Stephens Reference Huber and Stephens2001). For example, women who are employed by the state may be more involved in political networks, prompting their more active engagement in politics (Rosenbluth, Salmond, and Thies Reference Rosenbluth, Salmond and Thies2006). At the same time, the increased access to skills and time provided by improved education and care services enables women to translate political awareness into political engagement.

While it is true that some women who choose to run for public office might already be wealthy enough to not necessarily benefit directly from means-tested assistance programs (Shvedova Reference Shvedova, Ballington and Karam2005), state spending on social security, health, and education are often publicly available regardless of socioeconomic status. This means that even women who would not necessarily enroll in programs due to being economically marginalized might still gain some benefits from them in the form of having healthier children and reduced burdens on their time, which could have a positive impact on their ability to run for office. For economically marginalized women these benefits would likely be even greater. Additionally, there is evidence to suggest that women in key economic and political positions tend to strongly advocate for social spending once in office (Bolzendahl Reference Bolzendahl2009; Bolzendahl and Brooks Reference Bolzendahl and Brooks2007; Bratton and Ray Reference Bratton and Ray2002; Caiazza Reference Caiazza2004; Misra Reference Misra2003; O'Regan Reference O'Regan2000). Therefore, it is likely that social spending increases the chances for women to participate in politics, where they are likely to support social spending, which results in even more women getting involved in politics. This parallels Andrea Campbell's (Reference Campbell2005) findings in her influential work on the relationship between increases in benefits from Social Security and Medicare, and heightened civic participation among senior citizens in the United States.

To summarize, we assert that social spending enables women to participate in the political sphere by offering more women the opportunities to gain the skill sets and economic stability necessary to be competitive in national elections. It also relieves the care burden of those women who are tasked with caring for children or older relatives, which enables them become more active in politics. Based on the above discussion, we offer the following hypothesis:

Outcome Variables

The first outcome variable, Female labor force participation, is the percentage of women in the total labor force. It is a widely used proxy that captures women's overall labor force participation among the economically active population. The labor force data are from the World Development Indicators (World Bank 2015). The second outcome variable, Women's participation in national parliaments, is the proportion of seats held by women in national parliaments. The variable specifically accounts for female participation in the single chamber unicameral parliaments and the lower chamber in bicameral parliaments. The data for female parliamentarians are from the World Development Indicators (World Bank 2015) and Women in National Parliaments (Inter-Parliamentary Union 2015) datasets.

Explanatory Variables

To examine the impact of welfare spending on women's economic and political status, the model includes the welfare spending variable. It captures governments’ overall fiscal commitments to various social needs in three major areas, including education, health, and social security. The variable is measured as a percentage of total public spending.Footnote ⁶ Education expenditures include all government investments and expenditures in public education. Health expenditures include all the basic and hospital health care spending. Social security spending refers to such social protection expenses as family and children, unemployment, and housing. There is no guarantee that devoting money to certain social welfare areas will automatically result in improvements on those areas. Yet the share of welfare spending in total public expenditures is an indication of priorities set by governments. Further, there is already evidence indicating that allocation of more money and other resources is positively associated with advancements in social services, thus spending serves as a good proxy for whether people have access to public goods (Brown and Hunter Reference Brown and Hunter2004). The spending data come from the Relative Political Capacity (RPC) dataset (Kugler and Tammen Reference Kugler and Tammen2012). The RPC dataset originally gathered the spending data from the United Nations Statistics Division and the International Monetary Fund's Government Finance Statistics database.Footnote ⁷

The model also includes a battery of control variables to account for all the major covariates of women's economic and political empowerment. The natural log of gross domestic product (GDP) per capita income controls for the possible positive impact of economic wealth on women's wellbeing. Economic prosperity might help women improve their economic and political status by helping them gain more economic freedom, personal autonomy, and socioeconomic empowerment (Detraz and Peksen Reference Detraz and Peksen2016; Duflo Reference Duflo2012; Peksen Reference Peksen2011; Richards and Gelleny Reference Richards and Gelleny2007).

The CEDAW ratification variable controls for the expectation that countries that ratified the Convention on the Elimination of All Forms of Discrimination against Women (CEDAW) are less likely to tolerate gendered discrimination in politics and the marketplace (Cole Reference Cole2012; Murdie and Peksen Reference Murdie and Peksen2015). It is coded one for all the years since the ratification of the CEDAW convention, and zero otherwise. We also control for the percentage of women in total population (female population) since the higher proportion of women in total population might be positively associated women's involvement in the economy and politics.

The democracy variable controls for the possible positive relationship between democracy and women's wellbeing. The variable ranges from −10 (most authoritarian) to + 10 (most democratic) and comes from the Polity IV dataset (Marshall, Jaggers, and Gurr Reference Marshall, Jaggers and Gurr2012). Because democratic governments are more constrained by systems of checks and balances as well as the possibility of losing office through elections, they have more incentives to prevent systematic gendered, and other forms of discrimination against their citizens (Beer Reference Beer2009; Norris and Inglehart Reference Norris and Inglehart2001; Poe, Tate, and Keith Reference Poe, Tate and Keith1999). Women might therefore have a stronger presence in the economy and politics in democracies than nondemocratic regimes.

Earlier research suggests that more economic globalization is positively associated with women's economic and political status (Gray, Kittilson, and Shandholtz Reference Gray, Kittilson and Sandholtz2006; Neumayer and De Soysa Reference Neumayer and Soysa2007). As economic globalization promotes better living conditions through increased economic growth and prosperity, women might have more economic opportunities that they can use for economic independence and self-empowerment. To take this argument into account, the model includes economic openness, which is the natural log of total trade flows (exports plus imports) as a percentage of GDP. The data for economic development, economic openness, and female population are from the World Development Indicators dataset (World Bank 2015)

Methodological Concerns

To correct for autocorrelation (i.e., temporal dependence in the outcome variables), we lag both outcome variables by one year and then include them—Female Legislator_t-1 and Female Labor_t-1—on the right-hand side of the equation. We first report models with random effects. We also run models with country-fixed effects to ensure that the results are not biased by any unobserved country-specific, time-invariant factors. Earlier research suggests that the extent of female participation in politics and the economy might affect the level of welfare spending by the government (Bolzendahl Reference Bolzendahl2009; Bolzendahl and Brooks Reference Bolzendahl and Brooks2007). Thus, there might be a mutual interdependence (i.e., endogeneity) between the extent of welfare spending and women's involvement in the economy and politics.

To account for this possible endogenous process, we adopt two strategies. First, we lag the welfare spending variable one year in the models to make sure that they temporally precede the outcome measures. While this is not a perfect solution to address the question of reciprocal causation, it at least temporally allows the main explanatory variables to come first. Second, we estimate two-stage least squares (2SLS) models with an instrumental variable (IV) to account for the possibility of endogeneity. It is a challenging task to find instrumental variables, as it requires finding strong predictors of welfare spending that are theoretically and statistically exogenous to women's participation in national parliaments and the labor force. We use the following indicators in the 2SLS models: ethnic tensions, population size, the share of agriculture in GDP, and past welfare spending.Footnote ⁸

Ethnic tensions are also likely to decrease the extent of public services as rival ethnic groups might see general public services as less beneficial to their own groups and thus ethnic groups who have significant influence in the state might be less inclined to invest in social services (Alesina, Baqir, and Easterly Reference Alesina, Baqir and Easterly1999; Alesina and Spolaore Reference Alesina and Spolaore1997; Easterly and Levine Reference Easterly and Levine1997). The ethnicity data are from the International Country Risk Group (ICRG) dataset (Knack and Keefer Reference Knack and Keefer1998). The population size of a country also appears to have a significant negative impact on the amount of public spending (Easterly and Rebelo Reference Easterly and Rebelo1993). This is in part because it becomes more challenging for the state to provide social services for a large populace. Countries that heavily depend on agriculture as one of their main sources of income tend to have higher public spending, as those are also the countries with less liberalized economies and therefore face less pressure to limit public spending (Easterly and Rebelo Reference Easterly and Rebelo1993; Tanzi Reference Tanzi, Goldin and Alan Winters1992). The data for population size and the GDP indicator come from the World Development Indicators dataset (World Bank 2015). The extent of welfare spending in the recent history of a country is another key predictor of the current welfare spending trends. However, the spending in the immediate past might still be endogenous to women's well-being. We thus use a three-year lagged welfare spending variable to estimate the instrumental variable.Footnote ⁹

Additional Analyses and Robustness Checks

To make sure that our results are robust to the operationalization of the welfare spending measure, we used an alternative operationalization of the spending variable, welfare spending as a percentage of GDP. We find no major change in the main findings in the models estimated with the alternative spending variable. This finding should not be very surprising, as the correlation score between the two spending measures is considerably high, 0.82. The results from the models with the alternative spending variable appear in the supplementary material (Table A3). In addition to the fixed-effects models reported in the main analysis that account for any unobserved country-specific effects, we estimated models controlling for the dominant religion as well as the extent of ethnic or religious fractionalization. Our main findings remain unaltered in these additional models. This further confirms our theoretical claims that government's fiscal commitments to welfare policies might have a positive effect on women's involvement in the total labor force and national legislatures