This study investigates the impact of women's political and economic empowerment on social protection expenditures across 27 European Union countries (2006–2023).
Using fixed-effects panel regressions with Driscoll–Kraay standard errors, we explore both linear and conditional (interaction and non-linear) effects of selected dimensions of women's empowerment on shaping total social welfare spending and welfare-chosen functions.
Findings reveal subtle differences in the overall impact. When considering linear effects, greater women's political and managerial empowerment, and narrowing gender pay gap, are associated with increased total welfare spending, and health and elderly functions, while only the gender pay gap significantly explains spending on the family/children function. Uncovered U-shaped relationships suggest that political and economic women's representation must reach a threshold before exerting a positive effect on social spending. Moreover, political empowerment interacts significantly with income inequality, indicating that its effect is stronger in more unequal societies.
Overall, the study contributes to a more refined understanding of how political and economic empowerment influences welfare systems in diverse ways, highlighting the importance of considering empowerment as a multidimensional and context-sensitive factor in shaping inclusive and equitable social protection policies.
1. Introduction
In the context of the global drive for gender equality, the relationship between women's empowerment and social protection has become a central topic in social policy research. Women's empowerment serves not only as a pathway to gender equality but also influences the priorities and structure of social protection policies (Ennser-Jedenastik, 2017; Hessami and da Fonseca, 2020; Karaalp-Orhan et al., 2024). Women's economic and political empowerment significantly impacts spending on family welfare, childcare and health services, with countries that exhibit higher levels of women's empowerment demonstrating greater inclusivity and fairness in these functional areas (Allik, 2016; Bolzendahl and Brooks, 2007).
Although previous studies offer preliminary empirical support for links between women's empowerment and social protection expenditures, findings remain inconsistent and vary by region and context. Some studies argue that empowerment's impact on social protection expenditures is stronger in low-income countries, where more female political representation improves public services, such as education and healthcare, than in high-income countries, where the effect is weaker (Hessami and da Fonseca, 2020; Weeks and Baldez, 2015). The influence of female labour force participation (FLFP) on social protection expenditures is also debated. While some studies find that rising FLFP increases welfare spending on family support and childcare (Ennser-Jedenastik, 2017), others show a non-positive correlation in certain regions (Karaalp-Orhan et al., 2024). Most existing studies emphasize linear relationships, but these relationships may be more complex and context-dependent than previously assumed.
To address these gaps, this study aims to determine the differences in the understanding of the women's empowerment/social protection expenditures nexus by (1) examining the linear impact of different dimensions of women's empowerment on social protection expenditures, both in total and by selected functions, (2) testing the interaction on women's empowerment with income inequality and (3) verifying non-linear relationships. We focus on the EU countries as a research area, based on their shared values, including gender equity, and common institutional background (Deininger et al., 2020; Sharaunga et al., 2019), using macro panel data covering all 27 EU countries from 2006 to 2023. We conceptualise women's empowerment as a multidimensional construct, encompassing three dimensions: political empowerment, economic empowerment and gender pay gap, in line with previous studies indicating their mutual complementarity (Deininger et al., 2020).
2. Literature review
2.1 Women's empowerment: welfare state context
The welfare state remains a core institution of contemporary society, redistributing resources and delivering public services through state intervention to address market instability, social inequality and systemic risk, thereby promoting equity and social stability (Garland, 2015). The typology of Esping-Andersen (1990) provides a crucial framework for understanding cross-national variations in social policy systems, classifying welfare regimes by levels of decommodification and social stratification, but is criticised for neglecting the gender dimension. Feminist scholars argue that it often assumes similar welfare outcomes for men and women, neglecting the gendered division of labour, caregiving roles and women's specific needs in social policy (Laperrière and Orloff, 2018).
The gendered division of labour, within the labour market and the family, is a fundamental aspect of welfare state organisation (Schmitt and Trappe, 2010). In most Western countries, FLFP has increased, and dual-earner households are the norm, significantly transforming intra-household labour arrangements. This structural shift has altered patterns of unpaid labour, with men and women spending less time on housework and caregiving responsibilities, thereby challenging social policies based on male breadwinning. The growing demand for publicly-provided childcare, healthcare and elderly care services has led to a substantial expansion in social protection spending by the state (Bolzendahl and Brooks, 2007; Ennser-Jedenastik, 2017). Women's entry into paid employment has driven welfare state development, making gender both influenced by, and influential in, welfare institutions; thus, prompting growing scholarly attention to its interaction with women's empowerment.
The Human Development Report of the United Nations Development Programme (UNDP) identifies women's empowerment as a key dimension of gender equality in the Gender Inequality Index (GII). Women's empowerment is a complex, multidimensional process that is shaped by social and cultural contexts (Besnier, 2023; Gaweł et al., 2024; Sharaunga et al., 2019). A process that increases women's independence (Hornset and De Soysa, 2022; Persson et al., 2022), empowerment is the power and agency (Sharaunga et al., 2019) to control resources, act on greater choice and resist control by others (Persson et al., 2022). By increasing the ability and skills to make choices, empowerment converts those choices into effective actions that lead to desired outcomes (Sharaunga et al., 2019). This process encompasses various dimensions: political, economic, social, familial, legal and psychological empowerment of women are the aspects that are most often discussed, each contributing uniquely to the empowerment landscape (Bayissa et al., 2018; Mahato et al., 2023). We focus on political and economic empowerment of women, which are frequently discussed, but with an unclear relationship between them. It is noteworthy that some studies suggest complementarity (Deininger et al., 2020), while others point to weak interdependence (Bayissa et al., 2018).
Social protection, a core welfare state function, is essential for reducing poverty, promoting equity and supporting economic resilience. Programmes such as pensions, healthcare, unemployment insurance and family support provide basic financial security, and foster social cohesion (Ferrera et al., 2023; Spasova and Ward, 2019). The priorities and structure of social protection expenditures vary across countries, as they are shaped by politics, economic capacity, demographics and culture (Karaalp-Orhan et al., 2024; Spasova and Ward, 2019). The EU provides a shared institutional context in which social protection systems are increasingly shifting from nationally-led models toward forms of supranational governance, with social rights progressively embedding in national policy frameworks (Ferrera et al., 2023). Within this evolving landscape, examining how women's empowerment, particularly political and economic, influences social protection expenditures has become particularly salient, shifting focus from how institutions shape gender roles to how empowerment drives welfare policy.
2.2 Women's political/economic empowerment shapes social protection expenditures
Women's empowerment has become a global policy priority, with key policy frameworks emphasising its dual objective: empowering women as a goal in itself, and as a catalyst for broader development processes (Sundström et al., 2017). Politics dictates resource allocation, often privileging certain groups, as those in power shape individual behaviour, influence social institutions and codify norms into law (Franceschet and Piscopo, 2014; Sundström et al., 2017). Women's political empowerment remains contested (Huang and Lin, 2022), being defined as a process of increasing women's capacity to exercise choice, agency and meaningful participation in societal decision-making (Sundström et al., 2017). Effective participation requires adequate representation in formal political positions, often measured by women's representation in government; therefore, international organisations such as the UN, and the EU, have called on member states to ensure that women hold at least 30% of elected positions (Sundström et al., 2017), based on assuming that increased political representation can influence policy agendas and decision-making processes.
Policy preferences influence legislative behaviour and resource allocation (Funk and Gathmann, 2015; Ozdamar, 2017). Greater female political representation often shifts policy agendas toward caregiving domains like childcare, health and family welfare (Britt and Egerer, 2024; Högström and Lidén, 2024; Steklov et al., 2023). Detraz and Peksen (2018) and Allik (2016) find that women's presence in parliament links to higher welfare spending, including childcare and education (Park, 2017), public healthcare (Clayton and Zetterberg, 2018), housing and elderly care (De Siano and Chiariello, 2022). However, the empirical evidence on the policy consequences of women's political empowerment remains mixed and context-dependent. Studies affirm female representatives increase social protection expenditures; others suggest their symbolic influence, with limited impact on overall spending patterns (Ozdamar, 2017). In developing countries, women's political representation improves public goods provision, particularly education and healthcare (Hessami and da Fonseca, 2020). In developed countries, while total spending may not rise, female representation may influence policy content and debates, leading to gender-sensitive innovations such as expanded public childcare, often non-discernible in spending statistics. Local-level studies report that a 1% increase in women's parliamentary participation means higher safety expenditures, lower administrative costs, but mixed effects elsewhere (Andreoli et al., 2022).
Economic empowerment has complex effects on welfare policy (Karaalp-Orhan et al., 2024; Serban et al., 2020), but its relationship with social protection expenditures is understudied. As women gain economic independence, their reliance on certain welfare subsidies may decrease (Deininger et al., 2020; Hessami and da Fonseca, 2020). Higher FLFP raises demand for work-life balance services like childcare and family support (Ennser-Jedenastik, 2017; Karaalp-Orhan et al., 2024). Economically empowered women advocate more for family and community welfare; but, economic independence does not diminish welfare reliance (Camilletti, 2021). Focacci (2023) notes that in many European countries, prioritising pension expenditures often reduces funding for youth and family welfare, especially where women's empowerment is higher. This suggests a potential trade-off in social protection allocations, where resources directed toward elderly welfare may limit funding for childcare and family support.
Economic empowerment has a complementary nature (Deininger et al., 2020). Economic empowerment often enhances women's political influence, enabling them to support policies aligned with their needs. Women in local leadership are particularly effective in advancing family-related policies (Steklov et al., 2023). When combined, political and economic empowerment amplified women's impact on social protection systems. But, is women's empowerment becoming central to the maintenance of social policy (Bolzendahl and Brooks, 2007)? Moreover, existing studies often focus narrowly on specific expenditure categories, produce inconsistent and context-dependent findings, and frequently overlook the interaction between political and economic empowerment.
2.3 Critical mass theory and its limits in welfare policy research
Building on the interplay between political and economic empowerment, the critical mass theory (CMT) offers a lens with which to understand how having enough empowered women can change policy priorities. Introduced by Kanter (1977), CMT has become a foundational, yet contested, framework in the study of women's political representation. It assumes that when women make up less than 15% of a male-dominated group, their roles are symbolic and shaped by stereotypes; thus, the threshold of female representation is needed for real policy influence. But, despite its popularity, the CMT remains theoretically underdeveloped and empirically inconclusive for welfare policymaking (Ozdamar, 2017).
Empirical studies suggest various thresholds (15%, 20% or 30%) as points at which female representation begins to shift policy agendas. Giles-Sims et al. (2012) associate a 20% female presence in U.S. state legislatures with better elderly care policies, while Swiss et al. (2012) and UN agencies suggest a 30% benchmark. However, evidence for specific thresholds is weak. Park (2017) finds a positive association between women's legislative presence and education and childcare spending, even at 10%, but no significant effect on healthcare, regardless of whether representation exceeded conventional thresholds, challenging the idea of a fixed “critical mass”. Ozdamar (2017) argues that the policy influence of female politicians, especially on family benefits, becomes visible only after reaching a critical threshold, although female politicians often align with median voter preferences, rather than gender-specific interests.
Despite the popularity of the CMT, little evidence supports any specific threshold, and the exact critical point remains unclear. A key concern is that an overemphasis on numeric thresholds might unintentionally suggest that women with lower levels of representation are less relevant to policy. The key question is: Under what conditions does women's political and economic agency lead to social protection outcomes? While the CMT is often cited in gender and politics research, few studies have examined its applicability to social protection expenditures. This study addresses this gap by re-evaluating threshold effects in the EU context and their relationship to social protection expenditure, advancing understanding of the link between women's legislative representation and welfare policy.
3. Research methodology
3.1 Research assumptions
The paper examines the impact of women's economic and political empowerment on social protection benefit expenditures: (1) effects of different aspects of women's empowerment and (2) responses of various social functions. We made the following assumptions. Expenditures on social protection, both in total term and per function, are the consequence of (1) political decisions related to rules and laws of benefits, which reflect the welfare state regime, and are implemented in the legal framework (De Siano and Chiariello, 2022), as well as (2) the number of beneficiaries of the social protection support. The impact of women and their empowerment on social protection expenditures can be (1) direct effect reflecting their influence on social policy through their participation in the national parliament, (2) indirect impact from social pressure related to their position in the society, and the values and norms of a given country. Another possible indirect impact of women's empowerment is through the number of beneficiaries of social protection support, as women's positions affect their need for social help.
3.2 Data and method
We designed/conducted empirical studies. We focus on EU countries, based on publicly available data from the EU statistical office, Eurostat. We selected data for all 27 EU countries (Austria, Belgium, Bulgaria, Croatia, Cyprus, Czechia, Denmark, Estonia, Finland, France, Germany, Greece, Hungary, Ireland, Italy, Latvia, Lithuania, Luxembourg, Malta, Netherlands, Poland, Portugal, Romania, Slovakia, Slovenia, Spain and Sweden), and years 2006–2023, based on data availability. By implementing macro panel data from several countries, we follow the research approach in previous studies, while assessing the social impact of women's empowerment (Foudjo and Keneck-Massil, 2024; Hornset and De Soysa, 2022; Park, 2023).
Public expenditures on social protection benefits, dependent variables in our study, reflect the redistributive process of the welfare state, aiming to equalise opportunities and improve living conditions across the society. We considered two perspectives. Firstly, we analysed the total expenditure on social protection benefits as the dependent variable. Secondly, we examined specific functions of social protection, in order to understand the differences in their interactions. Among these, we focused on aspects that are commonly associated with female responsibilities. This assumption is supported by findings showing that female parliament members are more likely to participate in areas such as childcare and family policy (Rozanova and Mikheev, 2020). Consequently, we investigated how women's empowerment influences functions: the sickness/healthcare function, the elderly-related function, and the family/children function. Therefore, we used four measures of social protection benefits expenditures as dependent variables (SP, SHC, DIS and FC in Table 1).
List of variables and descriptive statistics of raw data for years 2006–2023
| Variable | Explanation | Mean | Standard deviation | Minimum | Maximum |
|---|---|---|---|---|---|
| Dependent variable | |||||
| Social protection benefits (SP) | Expenditures on all social protection benefits functions, the share of gross domestic product. (GDP) (in %) | 22.32 | 5.64 | 10.28 | 35.10 |
| Sickness and health care expenditures (SHC) | Expenditures on the sickness/health care function, the share of GDP (%) | 6.27 | 1.64 | 3.27 | 10.27 |
| Expenditures on the elderly-related function (OLD) | Expenditures on the elderly-related function, the share of GDP (%) | 9.16 | 2.56 | 3.16 | 15.55 |
| Expenditures on the family and children function (FC) | Expenditures on the family and children function of social protection benefits, the share of GDP (%) | 1.98 | 0.79 | 0.76 | 4.31 |
| Independent variables | |||||
| Women's political empowerment (WPE) | Seats held by women in national parliaments, the share of women in national parliaments, both chambers (lower house and upper house, if applicable) (%) | 26.68 | 10.18 | 8.70 | 49.60 |
| Women's managerial empowerment (WME) | The share of women as management board members in the largest publicly listed companies (%) | 18.52 | 10.44 | 1.00 | 46.10 |
| Gender pay gap (GPG) | Gender pay gap in unadjusted form: the difference between average gross hourly earnings of male and female paid employees as a percentage of male earnings (%) | 13.77 | 6.28 | −0.90 | 30.90 |
| Control variables | |||||
| Early leavers from education (EL) | The share of 18- to 24-year-olds with at least a secondary education, not involved in education or training in the past four weeks (%) | 10.39 | 5.54 | 2.00 | 38.50 |
| Unemployment rate (UE) | The share of unemployed people in the total labour force, aged 20–64 years (%) | 8.00 | 4.25 | 2.00 | 27.60 |
| The Gini index (Gini) | The relationship between cumulative population shares, ordered by equalised disposable income, and their cumulative share of total disposable income (from 0 to 100) | 29.70 | 3.91 | 20.9 | 40.80 |
Women's empowerment (independent variable), is a multidimensional construct represented by multiple measures (Besnier, 2023; Gaweł et al., 2024). The GII of the UNDP measures women's empowerment via parliamentary seats, and women with, at least, a secondary education. Recognising the complexity of empowerment measures (Sharaunga et al., 2019), we investigate both the economic and political empowerment (Deininger et al., 2020) from two aspects: power positions, and the disadvantaged situation. Since power positions are linked to women's participation in parliament and business, especially on management boards, we accepted measures that reflect this aspect, specifically, the proportion of women among national parliament members and management board members. We assumed that the gender pay gap indicates a disadvantaged situation; therefore, we also accepted measures of empowerment. By doing this, we presented two extremes and included three measures of women's empowerment as independent variables (WPE, WME and GPG in Table 1). Compared to the GII, we include political empowerment but exclude women's education, as higher education in the EU often shows a reversed gender gap. Instead, we incorporate economic aspects of empowerment, in order to capture its complexity better.
Since social protection benefit expenditures depend on factors beyond women's empowerment, control variables that reflect the socio-economic situation, which may affect beneficiaries, and societal values were included. We used three measures: early leavers from education, unemployment rate, and the Gini index (EL, UE and Gini in Table 1). The first two measures relate to the number of people in need, as potential beneficiaries, while the Gini index directly reflects societal inequality, indirectly indicating potential beneficiaries. Unemployment (control variable) was also used in other studies on welfare spending (De Siano and Chiariello, 2022).
From 2006 to 2023, across all 27 EU countries, social protection benefits expenses consistently accounted for 22.3% of gross domestic product (GDP). Of these expenditures, spending on the elderly was the highest at 9.16% of GDP, followed by healthcare (6.27%) and family and children (1.98%).
Looking at women's empowerment, on average in EU countries, women held 26.68% of seats in national parliaments (WPE), and they are 18.52% of management board members in listed companies (WME). These average values of women's empowerment show that, despite the progress made in EU countries toward gender equality (Rios et al., 2023), especially in some countries, there is still room for improvement. The correlation results show that political and economic empowerment are moderately and positively related (0.625). The gender pay gap (GPG) has a very weak correlation with both WPE (0.034) and WME (−0.100).
Our aim is to examine how different dimensions of women's empowerment affect social protection expenditures, while recognising potential reciprocal causality. Social protection systems, particularly those promoting childcare, healthcare and income security, may also enhance women's political and economic agency. To address this, we used panel regression with fixed effects, which controls for time-invariant and individual-specific factors, ensuring it isolates the effect of the studied variables from unobserved differences. However, we do not claim to identify causal effects in a strict econometric sense. Instead, the analysis aims to uncover patterns of association and interaction. A similar attitude is also accepted in other research dedicated to women's studies (Dumauli, 2019; Gao and Tian, 2023). To linearise the relationships, all raw data were transformed into natural logarithms. The specifications of the models were based on the initial equation:
where:
lnSPit – expenditures on social protection benefits as dependent variable converted into natural logarithms in t period and in i country, successively: expenditures on all functions (lnSPit), sickness and health care function (lnSHCit), elderly-related function (lnOLDit) and family/children function (lnFCit),
lnWEit, – independent variables of women's empowerment in t period and i country converted into natural logarithms, successively: women's political empowerment (lnWPEit), women's managerial empowerment (lnWMEit), and gender pay gap (lnGPGit),
lnELit, lnUEit lnGiniit, – control variables in t period and in i country converted into natural logarithms,
β0 … β4 – vectors,
i – the selected EU countries (1, 2, …, 27),
t – years (2006, 2007, …, 2023),
εit – error term.
We estimated models using total expenditures on social protection benefits (lnSP) and spending on selected functions as dependent variables, each with three independent variables. The variance inflation factor (VIF), which assesses multicollinearity among variables, indicated all VIFs were below 2.0, allowing their inclusion in the subsequent analysis. The initial models were tested for heteroscedasticity using the Wald test, for autocorrelation with the Wooldridge test, and for cross-sectional dependence via the Pesaran cross-sectional dependence (CD) test. When heteroscedasticity, autocorrelation or CD were detected, a fixed-effects panel regression with Driscoll-Kraay standard errors was used to control for these issues. To ensure robustness, we conducted sensitivity checks, including alternative specs with random and fixed effects with standard errors, plus subsample analyses of different periods. Since results remained consistent, we proceeded with the research.
To deepen our understanding, we first examine the linear effects of women's political and economic empowerment on social protection benefits expenditures. Then, we introduce interaction terms with income inequality (Gini coefficient), to assess whether inequality influences these relationships. Additionally, we test for non-linear effects using quadratic terms and segmented regression.
4. Results
The first set of models with the expenditures on social protection benefits (lnSP) as a dependent variable (Table 2) shows that all three measures of women's empowerment are statistically significant in explaining the social protection benefits expenditure. Women's political and managerial empowerment impacts the social protection benefits directly, the gender pay gap inversely. The rise of social protection benefit expenditures is influenced by the increase in women's political and managerial empowerment, and the decline of the gender pay gap. When comparing the absolute values of the parameters, women's political empowerment shows the strongest impact (β = 0.08*), followed by the gender pay gap with a moderate effect (β = −0.05**), and women's managerial empowerment with the weakest influence (β = 0.04**).
Models of expenditures on social protection benefits (lnSP)
| Variable | Model 1 | Model 2 | Model 3 |
|---|---|---|---|
| Constant | 3.49*** (0.33) | 3.64*** (0.23) | 3.89*** (0.24) |
| Independent variables | |||
| lnWPE | 0.08* (0.04) | ||
| lnWME | 0.04** (0.01) | ||
| lnGPG | −0.05** (0.01) | ||
| Control variables | |||
| lnEL | 0.004 (0.03) | 0.03 (0.03) | 0.004 (0.04) |
| lnUE | 0.12*** (0.03) | 0.12*** (0.03) | 0.11*** (0.03) |
| lnGini | −0.27** (0.09) | −0.29** (0.08) | −0.27** (0.08) |
| Fit statistics | |||
| No. of countries | 27 | 27 | 27 |
| No of observations | 480 | 480 | 463 |
| Wald test | χ2(27) = 1645.57, p = 0 | χ2(27) = 2082.06, p = 0 | χ2(27) = 1737.54, p = 0 |
| Wooldridge test | F(1, 26) = 299.90, p < 0.001 | F(1, 26) = 288.84, p < 0.001 | F(1, 26) = 236.98, p < 0.001 |
| Pesaran CD test | z = 32.78, p < 0.001 | z = 32.57, p < 0.001 | z = 27.05, p < 0.001 |
| F | F(4, 26) = 11.30, Prob > F = 0.000 | F(4, 26) = 8.13, Prob > F = 0.0002 | F(4, 26) = 16.01, Prob > F = 0.000 |
| Within R2 | 0.16 | 0.17 | 0.17 |
Note(s): ***p < 0.001, **p < 0.01, *p < 0.05; DKSE – Driscoll-Kraay standard errors in brackets
The next models (Table 3) use expenditures on selected social benefit functions as the dependent variables. As with total expenditures, political and managerial empowerment proportionally affect expenditures on sickness and health care function, while the gender pay gap is inversely proportional. The impact of women's political empowerment (β = 0.143**, Model 4) is again the strongest, followed by the moderated impact of gender pay gap (β = −0.06**, Model 6), and the weakest is women's managerial empowerment (β = 0.04*, Model 5).
Models of expenditures on social benefit functions
| Variable | Model 4 | Model 5 | Model 6 | Model 7 | Model 8 | Model 9 | Model 10 | Model 11 | Model 12 |
|---|---|---|---|---|---|---|---|---|---|
| Dependent variable | lnSHC | lnOLD | lnFC | ||||||
| Constant | 1.88** (0.55) | 2.23*** (0.45) | 2.51*** (0.40) | 2.46*** (0.21) | 2.57*** (0.21) | 3.10*** (0.25) | 3.22*** (0.75) | 3.34*** (0.62) | 3.14*** (0.61) |
| Independent variables | |||||||||
| lnWPE | 0.13** (0.04) | 0.11** (0.04) | 0.01 (0.05) | ||||||
| lnWME | 0.04* (0.01) | 0.08*** (0.01) | −0.01 (0.01) | ||||||
| lnGPG | −0.06** (0.02) | −0.09*** (0.01) | −0.04* (0.02) | ||||||
| Control variables | |||||||||
| lnEL | 0.05 (0.04) | 0.05 (0.03) | 0.03 (0.04) | −0.07* (0.03) | −0.01 (0.04) | −0.06 (0.04) | 0.02 (0.04) | −0.002 (0.05) | 0.06 (0.05) |
| lnUE | −0.03 (0.03) | −0.04 (0.03) | −0.04 (0.03) | 0.16*** (0.03) | 0.16*** (0.02) | 0.15*** (0.03) | 0.03 (0.03) | 0.03 (0.04) | 0.03 (0.03) |
| lnGini | −0.16 (0.15) | −0.17 (0.12) | −0.16 (0.13) | −0.24** (0.07) | −0.27*** (0.07) | −0.25*** (0.07) | −0.82*** (0.22) | −0.81*** (0.22) | −0.78*** (0.22) |
| Fit statistics | |||||||||
| No. of countries | 27 | 27 | 27 | 27 | 27 | 27 | 27 | 27 | 27 |
| No of observations | 480 | 480 | 463 | 480 | 480 | 463 | 480 | 480 | 463 |
| Wald test | χ2(27) = 2172.31, p = 0 | χ2(27) = 1469.91, p = 0 | χ2(27) = 1309.82, p = 0 | χ2(27) = 2377.11, p = 0 | χ2(27) = 4382.52, p = 0 | χ2(27) = 1769.44, p = 0 | χ2(27) = 13,791.9, p = 0 | χ2(27) = 14,627.9, p = 0 | χ2(27) = 12,243.8, p = 0 |
| Wooldridge test | F(1, 26) = 111.76, p < 0.001 | F(1, 26) = 130.55, p < 0.001 | F(1, 26) = 90.44, p < 0.001 | F(1, 26) = 222.71, p < 0.001 | F(1, 26) = 229.87, p < 0.001 | F(1, 26) = 205.54, p < 0.001 | F(1, 26) = 131.27, p < 0.001 | F(1, 26) = 130.76, p < 0.001 | F(1, 26) = 123.26, p < 0.001 |
| Pesaran CD test | z = 29.61, p < 0.001 | z = 31.21, p < 0.001 | z = 26.64, p < 0.001 | z = 24.99, p < 0.001 | z = 21.87, p < 0.001 | z = 25.22, p < 0.001 | z = 11.37, p < 0.001 | z = 12.06, p < 0.001 | z = 10.95, p < 0.001 |
| F | F(4, 26) = 8.15, Prob > F = 0.0002 | F(4, 26) = 23.19, Prob > F = 0.000 | F(4, 26) = 8.24, Prob > F = 0.0002 | F(4, 26) = 11.99, Prob > F = 0.000 | F(4, 26) = 54.41, Prob > F = 0.000 | F(4, 26) = 57.50, Prob > F = 0.000 | F(4, 26) = 12.49, Prob > F = 0.000 | F(4, 26) = 16.77, Prob > F = 0.000 | F(4, 26) = 7.41, Prob > F = 0.000 |
| Within R2 | 0.07 | 0.05 | 0.06 | 0.26 | 0.31 | 0.30 | 0.04 | 0.05 | 0.05 |
Note(s): ***p < 0.001, **p < 0.01, *p < 0.05; DKSE – Driscoll-Kraay standard errors in brackets
When considering the elderly-related function, the results align with previous models. All three empowerment measures are statistically significant: an increase in this spending is driven by the rise in women's political and managerial empowerment, and the narrowing gender pay gap. Once again, the impact of women's political empowerment (β = 0.11**, Model 7) is the strongest among these three functions; the effect of the gender pay gap is moderate (β = −0.09***, Model 9), while the influence of women's managerial empowerment is the weakest (β = 0.08***, Model 8).
The models examining the impact of women's empowerment on family and children's care expenditures show that the effects of two measures, women's political empowerment (Model 10) and managerial empowerment (Model 11), are statistically insignificant, while the reduction of the gender pay gap increases such expenditures (β = −0.04*, Model 12).
At the second research stage, we examined the different effects of women's political and economic empowerment on expenditures for social protection benefits (lnSP). Building on the linear effects (models 1 and 2, Table 2), we add interaction terms of both empowerment dimensions with income inequality (Gini coefficient) to assess whether inequality influences these relationships, and we also explore their non-linear effects (Table 4).
Interaction and non-linear models of expenditures on social protection benefits (lnSP)
| Variable | Model 13 | Model 14 | Model 15 | Model 16 |
|---|---|---|---|---|
| Constant | 7.16*** (1.08) | 4.82*** (0.34) | 3.88*** (0.85) | 3.64*** (0.23) |
| Independent variables | ||||
| lnWPE | −1.12*** (0.31) | −0.98*** (0.11) | ||
| lnWPE*lnWPE | 0.18*** (0.02) | |||
| lnWME | −0.05 (0.28) | −0.06 (0.03) | ||
| lnWME*lnWME | 0.02** (0.01) | |||
| Interactions | ||||
| lnGini | −1.36*** (0.31) | −0.23* (0.09) | −0.36 (0.26) | −0.26** (0.08) |
| lnGini*lnWPE | 0.36*** (0.10) | |||
| lnGini*lnWME | 0.03 (0.08) | |||
| Control variables | ||||
| lnEL | 0.01 (0.03) | 0.02 (0.03) | 0.03 (0.03) | 0.03 (0.03) |
| lnUE | 0.12*** (0.03) | 0.12*** (0.03) | 0.11*** (0.03) | 0.12*** (0.03) |
| Fit statistics | ||||
| No. of countries | 27 | 27 | 27 | 27 |
| No of observations | 480 | 480 | 480 | 480 |
| Wald test | χ2(27) = 2607.44, p = 0 | χ2(27) = 1659.86, p = 0 | χ2(27) = 2086.67, p = 0 | χ2(27) = 2290.6, p = 0 |
| Wooldridge test | F(1, 26) = 301.18, p < 0.001 | F(1, 26) = 295.92, p < 0.001 | F(1, 26) = 283.99, p < 0.001 | F(1, 26) = 285.04, p < 0.001 |
| Pesaran CD test | z = 30.84, p < 0.001 | z = 27.58, p < 0.001 | z = 32.62, p < 0.001 | z = 31.01, p < 0.001 |
| F | F(5, 26) = 193.97, Prob > F = 0.000 | F(5, 26) = 25.96, Prob > F = 0.000 | F(5, 26) = 11.05, Prob > F = 0.000 | F(5, 26) = 6.93, Prob > F = 0.0003 |
| Within R2 | 0.18 | 0.21 | 0.17 | 0.18 |
Note(s): ***p < 0.001, **p < 0.01, *p < 0.05; DKSE – Driscoll–Kraay standard errors in brackets
When considered alone (Model 1 in Table 2), women's political empowerment (lnWPE) has a positive effect (β = 0.08*), indicating that, without controlling for other factors, a 1% increase in women's political empowerment is associated with an average 0.08% rise in social protection expenditures. However, in Model 13 (Table 4), when interactions with income inequalities (lnGini) are included, the effect turns negative and is highly significant (β = −1.12***), suggesting that WPE alone may not directly increase social protection expenditures, unless other factors are considered. Income inequality, consistently, has a negative impact on social protection expenditures across all models, meaning that higher inequality leads to lower expenditures. When an interaction between women's political empowerment and income inequality (lnGini*lnWPE) is added, this interaction term is positive and highly significant (β = 0.36***), indicating that in societies with greater women's political empowerment, the negative effect of inequality on social protection expenditures is lessened. This suggests that in a high inequality context, women's political empowerment can help offset the negative effects of the Gini coefficient, and the influence of WPE becomes more significant when inequality is high.
The linear effect of women's economic empowerment (lnWME) on social spending is found to be significantly positive (β = 0.04***, Model 2 in Table 2), indicating that greater women's economic empowerment leads to increased social protection expenditures. However, when the interaction of women's economic empowerment and income inequality (lnGini) is implemented in estimations (Model 15 in Table 4), all women's economic empowerment (lnWME), income inequalities (lnGini) and their interaction (lnGini*lnWME) became insignificant effects on social protection expenditures, suggesting that income inequality does not significantly interact with women's economic empowerment in shaping social protection expenditures.
Next, we investigated the non-linear relationship between women's empowerment and expenditures on social protection benefits (Table 4). Since the function includes a quadratic term (lnWPE*lnWPE), which is statistically significant and positive (β = 0.18***, Model 14), it implies the U-shape relationship between women's political empowerment and expenditures on social protection benefits. Then, we tested the non-linear impact of women's economic empowerment and, despite the fact that the impact of lnWME is negative but marginally significant (β = −0.06, p = 0.087), the evidence of a U-shaped relationship between women's economic empowerment and social protection benefit expenditures was found, based on the positive value of lnWME*lnWME coefficient (β = 0.02**, Model 16).
For the U-shape relationship, we calculated the thresholds of functions 14 and 16, based on the formula of the vertex of a parabola, and estimated the segmented panel regressions for observations below and above thresholds (Table 5).
Segmented models of expenditures on social protection benefits (lnSP)
| Variable | Model 14.1 Below threshold | Model 14.2 Above threshold | Model 16.1 Below threshold | Model 16.2 Above threshold |
|---|---|---|---|---|
| Threshold | lnWPE* = 2.73, WPE* = 15.34% | lnWME* = 1.45, WME* = 4.28% | ||
| Constant | 3.82*** (0.68) | 7.16*** (1.08) | 3.91** (0.76) | 3.47*** (0.26) |
| Independent variables | ||||
| lnWPE | −0.04 (0.05) | 0.16** (0.06) | ||
| lnWME | 0.006 (0.02) | 0.05*** (0.01) | ||
| Control variables | ||||
| lnGini | −0.38 (0.18) | −0.08 (0.07) | −0.35 (0.23) | −0.24* (0.09) |
| lnEL | 0.06 (0.05) | 0.002 (0.03) | −0.08* (0.02) | 0.03 (0.04) |
| lnUE | 0.18*** (0.03) | 0.10*** (0.03) | 0.30*** (0.04) | 0.11*** (0.03) |
| Fit statistics | ||||
| No of observations | 68 | 412 | 25 | 455 |
| F | F(4, 8) = 35.12, Prob > F = 0.000 | F(4, 25) = 15.06, Prob > F = 0.000 | F(4, 5) = 56.41, Prob > F = 0.0002 | F(4, 26) = 5.05, Prob > F = 0.004 |
| Within R2 | 0.60 | 0.15 | 0.71 | 0.14 |
Note(s): ***p < 0.001, **p < 0.01, *p < 0.05; DKSE – Driscoll–Kraay standard errors in brackets
The relationship between women's political empowerment and expenditures on social protection benefits can be expressed as the function lnSP = 4.82-0.98lnWPE+0.18lnWPE2+controls (Model 14). The threshold of lnWPE, based on the formula of the vertex of a parabola, is lnWPE* = 2.73 (lnWPE* = = = 2.73). Later, after converting lnWPE into the actual WPE by exponentiation, it yields the threshold in WPE as 15.34%. Next, we estimated segmented regressions, based on initial model 1, separately below (Model 14.1) and above the threshold (Model 14.2). Results of both models support the U-shaped relationship; below the 15% threshold of women's participation in national parliaments, women's political empowerment does not significantly influence social protection expenditures, suggesting that low levels of political empowerment are insufficient to impact policy decisions. Above the 15% threshold, women's political empowerment has a significant positive impact on social protection expenditures (β = 0.16**), indicating that higher political empowerment correlates with increased social protection expenditures.
A similar procedure was applied to examine the non-linear impact of women's economic empowerment and expenditures on social protection benefits based on Model 16, expressed as the function lnSP = 3.64-0.06lnWME+0.02lnWME2+controls. From this, we calculated the threshold of lnWME* = 1.45 (lnWME* = = = 1.45), and by exponentiating this value, we found the WME threshold to be 4.28%. Next, segmented regressions were used to estimate models below (Model 16.1) and above (Model 16.) the lnWME* threshold of 1.45. Their results support U-shape effect, with women's economic empowerment having no significant impact on social protection spending below the threshold of 4.28% women's participation in management boards. However, above this threshold, women's economic empowerment positively affects social protection expenditures (β = 0.05**). This means that higher economic empowerment increases social protection expenditures. Overall, this means that once women's participation in management boards exceeds about 4.28%, increased economic empowerment leads to higher social protection spending.
The comparison between the interaction and non-linear models reveals both complementarities and divergences in explaining how women's empowerment influences social protection expenditures. For political empowerment, models 13 (interaction with inequality) and 14 (U-shaped specification) offer consistent insights. Both suggest that women's political empowerment does not uniformly increase spending, but becomes effective under specific conditions, either in more unequal societies or after surpassing a threshold level of women's political representation. In contrast, the results for women's economic empowerment are less aligned. While model 16 supports a U-shaped relationship, indicating that economic empowerment becomes impactful only beyond a certain threshold, model 15 shows no significant interaction with inequality. This suggests that the economic empowerment of women only becomes effective in impacting social protection expenditures after passing a certain threshold of women's participation in management positions, regardless of broader structural inequality.
5. Discussion
To evaluate the influence of women's empowerment on social protection spending, both overall and by function, we analysed macro panel data from 27 European Union countries, focusing on three aspects of women's empowerment related to power roles and inequalities. This study advances previous research by offering a more comprehensive and a detailed look at the complex and multifaceted impacts of women's empowerment, both political and economic, on social protection expenditures. Instead of assuming a straightforward relationship between empowerment and welfare growth, our findings reveal complex and varied effects influenced by empowerment dimensions, types of welfare functions, and socio-economic factors. While we mainly interpret these effects from the perspective of empowerment affecting expenditure, it is also possible that social protection spending, especially when aimed at reducing structural inequality, may help boost women's empowerment. Therefore, our results should be viewed in light of potential reciprocal relationships.
The findings from the first-stage analysis demonstrate the differentiated effects of women's empowerment across its multiple dimensions on social protection expenditures. Women's political empowerment exerts the most consistent and significant positive influence, particularly in the domains of healthcare and old-age benefits. These results echo earlier studies (Allik, 2016; Detraz and Peksen, 2018), which suggest that greater female representation in politics tends to promote gender-sensitive social protection expenditures, especially in care-related sectors. This “care-oriented” prioritisation reflects the persistent gendered structure of social reproduction (Ferree, 2010; Few-Demo and Allen, 2020), whereby female policymakers are more likely to champion inclusive welfare provisions.
Women's managerial empowerment shows a positive effect on social protection expenditures, but it is weaker compared to political empowerment, indicating that representation in corporate leadership has a different weight in shaping broader macro-level welfare policy as political representation. Political empowerment aligns more closely with the mechanisms that drive public welfare investment, in line with Deininger et al. (2020), who argue for the complementarity between economic and political empowerment, while also resonating with Bayissa et al. (2018), who caution that economic empowerment alone, particularly when symbolic or tokenistic, may lack sufficient institutional leverage, with which to alter redistributive patterns. Additionally, the inversely proportional relationship between the gender pay gap and welfare expenditures is confirmed, meaning that lower income inequality between men and women may support collective investment in social protection.
The analysis reveals interesting patterns across different social protection functions. For both sickness and healthcare expenditures and elderly-related expenditures, all three dimensions of women's empowerment show effects that are similar to those observed in total social protection spending. In contrast, expenditures on the family/children function are not significantly influenced by political or managerial empowerment, but are inversely associated with the gender pay gap, indicating that greater income equality fosters greater investment in family-related benefits. This contributes to the discussion about relying on market-based or private alternatives to public family support (Hessami and da Fonseca, 2020; Focacci, 2023), and the gender equality in the labour market/public investment in family care nexus (Risman, 2017).
The second-stage analysis adds further depth by revealing conditional and non-linear effects of empowerment. Firstly, income inequality significantly moderates the relationship between women's political empowerment and social protection expenditures. In countries with higher levels of inequality (as measured by the Gini coefficient), the positive effect of political empowerment on social protection expenditures is amplified. This supports the idea that, in more unequal societies, women's political empowerment may function as a redistributive corrective mechanism (Foudjo and Keneck-Massil, 2024; Rios et al., 2023), particularly in contexts where vulnerable populations have limited institutional support.
Secondly, the U-shaped non-linear relationship suggests that the empowerment/welfare nexus does not progress linearly. At low levels of empowerment, increases may initially correspond with decreased social protection expenditures, possibly due to limited political influence or institutional inertia. Only after reaching a critical threshold does empowerment begin to have a strong positive effect. This supports the CMT (Högström and Lidén, 2024), and emphasises the importance of threshold effects in the EU context. As Britt and Egerer (2024) point out, the policy impact of empowerment is often delayed, and early-stage gains may be symbolic, rather than substantive, unless they are embedded within institutional frameworks.
Combined, these findings reinforce two core insights. Firstly, women's empowerment must be conceptualised as a multidimensional and context-dependent process. Secondly, its effects on social protection expenditures are not automatic, but mediated by broader structural and cultural conditions. For policymakers, this implies that advancing gender equality alone may not be sufficient to drive welfare transformation. Instead, it is essential to understand and leverage the thresholds, institutional channels and cultural resistances that shape the translation of empowerment into inclusive and equitable welfare outcomes.
5.1 Research limitations
Despite important insights into the relationship between multidimensional women's empowerment and social protection expenditures in the EU, several limitations should be acknowledged. Firstly, the use of macro level panel data captures national trends, but may obscure subnational variations. Future research could consider more granular data to reflect within-country differences. Secondly, establishing causality is challenging. While fixed-effects models and Driscoll-Kraay errors address some endogeneity and serial correlation, they cannot rule out that social protection expenditures may also promote women's empowerment, raising the risk of bidirectionality. Future studies could strengthen causal inference through instrumental variables or natural experiments. Thirdly, focusing solely on EU countries limits generalisability. The findings may not apply to contexts with different welfare structures and gender norms, particularly low- and middle-income countries. Comparative research beyond the EU could illuminate whether similar patterns persist elsewhere.
5.2 Practical and societal implications
The study's findings offer implications for research, policy, education and the society. Firstly, they advance existing theoretical debates by highlighting multidimensional and nonlinear relationships between women's empowerment and social protection expenditures, inviting future research into interactive and contextual factors. Secondly, policymakers should consider that, while supporting family, elderly, and healthcare services might improve welfare, care must be taken to avoid reinforcing gender stereotypes. Instead, transformative policy should actively promote gender equality. Increasing women's political representation and reducing gender pay gaps can be strategically leveraged to enhance care-based welfare provisions, and broaden equality-focused and resilience social policies. Recommended actions include binding gender quotas, more women in policy and budgetary decision-making bodies, pay transparency and EU-level coordination on gender-responsive budgeting to ensure measurable impacts. Thirdly, this research offers teaching material for public policy, economics, and gender studies, enriching discussions on welfare and empowerment. Finally, by clarifying how welfare systems shape gender outcomes, these findings can raise public awareness and foster support for inclusive, equity-oriented reforms.
6. Conclusion
This study examined how women's empowerment, measured by political representation, managerial presence and the gender pay gap, relates to social protection expenditures across EU countries. Greater political and managerial empowerment, along with a narrowing gender pay gap, contribute to increased social protection spending, especially for total, healthcare and elderly expenditures. However, expenditures on the family/children function appear responsive only to income inequality. Importantly, the impact of women's empowerment on welfare spending is context-dependent. Interaction models show that income inequality shapes this relationship, and a U-shaped link suggests nonlinear dynamics. These suggest that the empowerment/expenditure nexus operates through institutional, social and political thresholds, and not solely linear accumulation. Thus, higher empowerment links to greater welfare investment under some conditions. We acknowledge the challenge of causal direction. While we examine how empowerment affects spending, social protection programmes (especially those enhancing care, income and service access) may also foster women's agency, a potential feedback loop suggested by prior research and our findings. Finally, our findings raise normative questions: Should welfare reinforce traditional roles as empowerment, or aim to enhance women's autonomy through redistributive policies? Both coexist in policy. For real progress, social protection must address structural barriers, and embed gender equality beyond symbolic measures.
