The authors would like to thank the referees of the Revue de la régulation and the organizers of the Séminaire d’économie politique and discussants for their very useful comments and suggestions. All errors and omissions, of course, remain the responsibility of the authors.
1The European Union, like most countries, recognizes that women’s participation in the labor market is a factor of growth and that equal rights for men and women should exist (World Bank Group, 2018). The principle of equality between women and men is inscribed in different EU treaties and in recent years progress has been made: women have reaped the benefits of education and significantly increased their presence in the labor market. The reason is not just a desire for more equality; it is also economic. Better access to the labor market for women and more equal employment opportunities have been identified as key factors for achieving growth and a sustainable social system in Europe, in particular in view of the aging population (European Commission, 2018).
2However, two questions arise. First, when a crisis of the magnitude of the Great Recession hits economies, what happens to these principles and what is the recession’s impact on gender equality in the labor market? Second, the European Union is not a homogenous zone despite the existence of a single market and a monetary union. A diversity of national models persists due to divergent historical, institutional, political, and cultural conditions in the individual member states. Economic, social and political institutions differ considerably from one country to another despite European integration, the free movement of goods and services within the common market, and different strategies aimed at achieving high levels of economic harmonization. Under these circumstances, does the evolution of gender inequalities differ in member states or groups of countries?
3A number of studies have examined how the economic and financial crisis and ensuing policies have affected gender inequalities. Among the most important questions are the disparities in the labor market, even though they differ according to countries, regions, and sectors (see Karamessini & Rubery, 2013; and Eydoux et al., 2014, for early important contributions). One of the widely acknowledged gendered effects of the recession has been coined “He-cession” and “She-austerity”, although many European countries followed their own particular paths (Périvier, 2014; 2018). Overall, male employment was severely affected during the recession phase, while female employment contracted because of subsequent austerity and fiscal policies. These gendered effects resulted from gendered sectorial segregation in terms of employment combined with the sectorial dimensions of the recession and of austerity policies.
4In addition, labor market evolutions by gender encompass not only the direct effects of the recession or austerity policies, but also indirect effects, which have resulted from “gender-blind” measures and were implemented following the crisis and a decade of an overall gender oriented European Employment Strategy (Smith & Villa, 2014). Thus, the overall gender recession effects include “foregone employment” by gender: that is, the gender specific loss in employment when compared with the prevailing trend before the crisis.
5These findings also fit the perspective of welfare state models (Esping-Andersen, 1990; Esping-Andersen & Myles, 2009), and gender regimes (eg Lewis, 1992; Hobson, 1994). Because welfare states regimes articulate different degrees of both “defamilialization” and “decommodification”, combinations of social protection, public service and family policy provision, labor market regulation, and fiscal policy affect gender inequalities in the labor market. Some authors have attempted to “gender the varieties of capitalism” (eg Estévez-Abe, 2005; Soskice, 2005; Rubery, 2009 for a critical view), while others have called for cautiously combining welfare state, gender and production regimes typologies, including labor market regulation (O’Reilly, 2006; Rubery, 2011).
6Analyzing the complementary nature of institutions helps explain the interaction, coherence and potential evolution of gender-based differences in the labor market in a dynamic way. Institutions reflect, structure and reinforce gendered patterns and gendered modes of interaction, so that seemingly neutral institutional processes and practices are, in fact, gendered. Many of them, such as working conditions, social protection programs, pension schemes, and public services for children and elderly persons, are fundamental variables in shaping gender inequalities in the labor market. Flowing out of the historical, economic, political, and social history of each country, whether it be a question of social protection, childcare, other family-oriented policies, or employment protection laws, the European systems have not actually begun reconciling their organizational modes and therefore remain stamped by quite distinctive institutional features. According to Morel (2007), “cross fertilization” between institutionalist approaches and feminist economics can shed light on the influence of institutions and their complementarity in shaping gender inequality.
7This paper proposes an empirical contribution to the literature. First, we set up an empirical framework, which tries to bring together a comparative, feminist, and institutionalist approach following Rubery (2003) and Grimshaw et al. (2017), with Régulation theory approaches (Amable et al., 2005, 2011). For this purpose, we analyze gendered effects of social protection and welfare state regimes, together with labor and product market regulation. Second, we study the evolution of several gender gaps in 14 European Union countries before and after the outbreak of the crisis. In this way, we are able both to obtain original aggregate results as well as conclusions that are more familiar.
8In particular, we find, in line with previous studies, that a greater number of jobs in the service sector and in the public sector contributes to decreasing gender gaps in employment and unemployment rates. We also obtain results concerning institutions. On the labor market side, we highlight the positive role of unions in promoting gender equality with respect to participation, employment, and, unemployment. A minimum wage also lowers the part-time gap. We find that employment protection legislation has different effects. The protection of permanent job contracts is beneficial for employment equality, but widens the part-time gap. The protection of temporary contract workers, on the other hand, is associated with greater employment inequalities. As for increased product market regulation, it is associated with higher participation and employment gaps. Public policies, taxation, and public spending also influence gender gaps. We find negative associations between unemployment replacement rate, participation tax rate and social protection spending, and respectively participation, part-time and unemployment gaps. Incentives associated with fiscal policies seem to exert differentiated effects on aggregate female and male behaviors.
9The rest of the paper is organized as follows. Section I reviews the related literature background. Section II outlines our empirical methodology and our data. Section III presents the results and discusses their main implications.
10We successively review the mechanisms and findings from different approaches concerned with gendered effects of the recession that started in Europe in 2008-2009.
- 1 This approach also puts a particular emphasis on the effect of children on women’s labor force part (...)
11Standard economic analysis has relied mainly on the neoclassical approach (labor supply models, wage equation estimations, à la Becker, 1964, or Mincer, 1958) to study various women’s labor market behaviors, with a focus on the behavior of women seen through the added worker effect or the discouraged worker effect.1 Added worker effect is broadly defined as an increase in the spousal participation in the labor market in response to an income shock. The increase in the female participation rate may reflect a distress response to the crisis (Lundberg, 1985; Cerrutti, 2000; Sabarwal et al., 2011), when wives enter the labor force as a response to spousal job loss. In contrast, the discouraged worker effect describes a situation where unemployed individuals stop looking for jobs when they anticipate their search has a low probability of success. In particular, married women may become discouraged when the unemployment rate is high, leading them to leave the labor force. Both effects exist and are responses to the worsening of the macroeconomic environment. The question is which one is dominant in time of crisis, since they have opposite effects on the participation of women in the labor force (Gong, 2011).
12A second approach relates to firms' behavior, again with two contradictory effects. On the one hand, management may be tempted to hire women because they provide cheaper labor than male workers do. We refer to this as the substitution effect (Bruegel, 1979; Rubery & Tarling, 1988). On the other hand, women can be considered as a flexible labor reserve, operating on the secondary labor market (Piore, 1978), and a decrease in employment opportunities can thus lead to more women being laid off or otherwise leaving the labor force. This is the buffer effect, leading to fewer women being part of the labor force during a downturn (Rubery, 1988).
13Finally, gender segregation is a third approach through which shifts in the participation of women in the labor force can be analyzed. Men and women do not occupy the same jobs and are not present in the same sectors (Bettio & Verashchagina., 2009). Workforce job data is limited in that it does not indicate patterns of occupational vertical or horizontal segregation within sectors. However, trends at the level of the industrial sector make it possible to analyze whether gender segregation of men and women in different sectors influences patterns of male and female job losses (Smith & Villa, 2014). Men and women appear to be in “protected” or “buffer” jobs based on their job losses within sectors being proportionate or disproportionate to the level of sector representation.
14Another important result taught by the empirical literature refers to country heterogeneity in, for example, child penalty on women’s employment. This sheds light on the importance of taking into account the variety of national regimes (family policies, fiscal arrangements, child care systems). In particular, comparative welfare state analysis might be fruitfully combined with feminist approaches to highlight varieties of “gendered welfare systems” in which interactions between women’s employment patterns and family structure and culture vary (see Anxo et al. 2007). In countries where a majority of women has been working and part of the labor force for decades, the “discouraged worker effect” and the “added worker effect” are different from those in countries where men are still viewed as the main breadwinners.
15Finally, the varieties of capitalism literature and the Régulation school approach have put forward the role of the institutional framework in defining national regimes, pointing to potential complementarities of production systems, including educational, labor and product market institutions, as well as welfare regimes in shaping differentiated integration of women into labor markets (see Thévenon, 2015).
16None of the aforementioned typologies (welfare regime, gender regime, production regime) perfectly coincides with each other, highlighting the importance of national specificities, as well as some methodological differences; nonetheless, they identify institutions that matter in shaping gender inequalities on the labor market, given country heterogeneity as the result of interactions between gender, welfare and regulation regimes. In what follows, we examine how these varieties of institutions affect gender gaps on the labor market during an economic recession.
17Gender equality in the labor market encompasses a variety of dimensions. When analyzing on a cross-country level, the literature often focuses on the participation gap and the employment gap. Some studies have also looked at the unemployment gap, while the evidence on the part-time gap is scarce. We successively review the available evidence from econometric studies on the gender effects of the recent crisis.
18Looking at eight European countries most affected by the financial or austerity crises, Karamessini and Rubery (2014) highlight the sectorial dimension of gendered effects of the recent crisis on employment, the so-called “He-cession” and “She-austerity” phenomena. Men are over-represented in the industry or construction sectors, which were the first hit by the crisis and have witnessed a massive destruction of jobs. This leads to a decrease in the employment gender gap. On the other hand, since women are overrepresented in the service and the public sectors and are more at risk of bearing the negative consequences of public spending cuts, austerity and fiscal consolidation policies might lead to increases of gender gaps. Périvier (2014, 2018) shares the same overall conclusions, but highlights country specificities and discrepancies concerning the extent of the role of sector segregation in the gendered pattern of employment during the crisis. As an example, some countries (eg Sweden) did not exhibit a strong sector segregation related gendered pattern because sector segregation by gender was relatively low before the crisis, and/or because austerity measures in the public sector were moderate.
19Education and labor market institutions mitigate or exacerbate these variations. In particular, male part-time has increased on average, while female part-time has remained stable in most countries (Karamessini & Rubery, 2013). From the standard point of view employment protection is thought to favor “male insiders”, but in fact deregulation has often been directed at secondary segments, where it has mostly degraded women’s employment conditions. In addition, education might not guarantee women’s access to quality employment as much as it did before the crisis. However, when looking at the employment gap in EU 28, Jaba et al. (2015), find that GDP per capita had a positive effect on the gender gap in EU 15 countries, and that the education gap is positively associated with the employment gap. Kushi and McManus (2018) estimate female and male unemployment equations for 28 OECD countries, and find that employment protection is more favorable to women (as well as expected industry and service sectors effects).
20Kushi and McManus (2018) also devote particular attention to the gender effects of the welfare regime type: they find that the Nordic and Liberal models equally protected women and men from unemployment, albeit with fundamentally different underlying mechanisms, while Continental and Southern welfare regimes failed to do so, with women being the most impacted. They point to public social spending and public sector employment dynamics as the primary explanation of the gendered regime outcomes.
21Adding a gender aspect to varieties of capitalism, Kang (2017) finds that women’s overall employment is less penalized in CMEs (Coordinated Market Economies, with high regulation of the labor and product markets) than in LMEs (Liberal Market Economies, with low regulation of the labor and product markets). However, longer parental leaves are associated with higher employment gaps in CMEs due to delayed skill acquisition and career interruption for women. Going beyond the duality between CMEs and LMEs, Thévenon (2015) incorporates welfare state variety into the analysis of female employment outcomes of family policies, and confirms that their effects are context dependent.
22Summing up, empirical studies of labor markets in times of crisis exhibit explicit gendered effects associated with employment sectorial shares, employment protection, education, and GDP per capita levels. In addition, other institutions, related either to the labor market (unionization, minimum wage policy, unemployment replacement rate), the product market (product market regulation) or fiscal policy (participation tax rate) affect female and male behavior and perspectives on the labor market differently. Because of the implementation of gender blind policies during the financial crisis and the period of austerity that followed, in addition to the abandonment of previous gender equality policies, an end has been put to the gender equalizing trends prevailing before the crisis (Smith & Villa; 2014; Kushi & McManus, 2018).
23The following table partly draws on Rubery (2011) and summarizes the main variables and mechanisms that can affect labor market gender gaps.
Table 1. Potential Gendered Economic Impact of Macroeconomic, Fiscal, and Institutional Variables
- 2 Average effective tax rate, for a transition into full-time work for persons not entitled to unempl (...)
|
Variable
|
Potential gendered economic impact
|
|
GDP per capita
|
An increase in GDP per capita is normally associated with a decrease in the labor market participation gap.
|
|
GDP growth rate
|
Higher women’s participation in the labor force boosts GDP (OECD, 2012), which does not mean a growth in GDP automatically leads to an increase in the participation rate of women.
|
|
Education level
|
An increase in the education level is associated with a decrease in the participation gap (Steinberg & Nakane, 2012).
|
|
Employment protection legislation (EPL),
|
According to the standard view, employment protection legislation reduces the participation and employment of women (who are seen as outsiders). However, EPL can stabilize the employment of women, and other institutions or policies can mitigate the outsider/insider divide.
|
|
Fertility
|
Some studies show a positive correlation between employment and fertility across EU Member States (Del Boca et al., 2005). In countries where women have had a high participation rate in the labor market over a few decades, higher fertility can accompany higher levels of employment. Other studies find that for individual countries there is evidence of a negative relationship between fertility and women’s participation in the labor force (Bloom et al., 2009).
|
|
Minimum wage policy
|
According to the standard view, a minimum wage reduces the level of employment, and women are particularly affected due to their lower relative productivity on the labor market. However, this effect on employment is put into question in alternative non-competitive models. Moreover, a minimum wage may increase women’s participation rate and decrease the earnings gap.
|
|
Participation tax rate2 (PTR)
|
PTR is supposed to discourage activity, and the labor supply of women is viewed as more sensitive to fiscal incentives. However, its impact on the labor market depends on labor market conditions and labor market institutions.
|
|
Product market regulation
(PMR)
|
The product market environment has an impact on economic growth, on living standards, and on employment (Amable et al., 2005). Complementarities between product and labor market regulations have been documented (Nicoletti & Scarpetta, 2005; Amable et al., 2005).
|
|
Sectorial employment
|
Male and female workers are not equally present in all sectors due to occupational segregation in the jobs market along gender lines. Jobs in sectors where men are predominantly employed can be affected by the crisis at a different time than sectors that are the key source of employment for women.
|
|
Social spending (share of GDP)
|
According to the standard view, higher social spending should lower women’s participation in the labor force and increase gender gaps. However, social protection can also narrow gender gaps if it is directed towards lowering women specific barriers to employment or participation.
|
|
Unemployment net replacement rate (NRR)
|
According to the standard approach, net replacement rates constitute a disincentive for women. But this assumes that women have the same NRR as men.
|
|
Unionization
|
In the standard view, unionization is detrimental to women, who are viewed as outsiders (they have less access to regulated/high wage sectors). However, unions reduce individual bargaining and discretionary discrimination so that women may be the biggest beneficiaries of higher levels of unionization.
|
Source. Author's
In the next section, we propose an empirical approach that encompasses different types of institutional variables as well as macroeconomic conditions.
24The main database used is The European Labour Force Survey (EU LFS), provided by Eurostat. Eurostat and the OECD provide macroeconomic and fiscal indicators. Our sample is made up of 14 European countries. These countries are: Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, Netherlands, Portugal, Spain, Sweden, and the United Kingdom. We study the period between 2003 and 2013.
25We estimate models of four dependent variables, namely the participation gap, the employment gap, the part-time employment gap, and subsequently the unemployment gap. We use GDP per capita, growth rate, female education level, fertility rate, unionization, sectorial employment, employment protection legislation, product market regulation, participation tax rate, net replacement rate, public social spending (as share of GDP), and hourly minimum wage as covariates in the four models. Moreover, we control the female and male unemployment rates in the participation gap equation, and add the female and male participation rates in the unemployment gap equation. This allows us to test how unemployment and inactivity gaps might be related, to assess potential discouraged or added worker effects, and to test if labor market institutions have gendered effects.
26Before presenting the econometric model, a few preliminary remarks concerning the data and the indicators are in order. Indeed, there are some limits to the comparability of labor market indicators across countries. For instance, the size of the informal sector varies considerably from one country to another. Another limit to the comparability of labor force indicators is related to the existence or not of a military draft in each country. In addition, the length of the workweek for full-time positions varies, and subsequently the part-time indicator of people working less than “full-time” (which is usually set at 30 hours per week). In addition, the participation and unemployment rates do not account for workers who become discouraged, stop looking for work, or are not available right away to take a job. This is important because in times of crisis the number of discouraged workers excluded from the unemployment rate is even higher than in normal times. Unemployment figures do not include people who worked only a few hours but who want full-time employment. Also, the employment rate does not account for workers who are underemployed and wish to work more. Despite the shortcomings and limits of labor market cross-country time-series data and indicators, a lot can be learned from such an empirical analysis.
27All labor market gender gaps are defined for the working-age population (15-64 years old). The participation, employment and unemployment gaps are defined as male rates minus female rates, whereas the part-time gap is defined as female rates minus male rates. Subsequently, we run an econometric analysis of the participation, employment, part-time, and unemployment gaps.
28We use the panel structure of our data by assuming two-way fixed effects (Baltagi, 2013). This allows us to deal with individual heterogeneity, like constant through time unobserved country specific factors, as well as with year specific shocks common to all countries.
29More precisely, for country i and year t, we model each dependent variable Yit as follows (1):

- 3 We leave to future research a full account of path dependency by including the lagged dependent var (...)
where Xit is a (1 x k) vector of exogenous or predetermined variables, ci is a country specific (and time invariant) effect, ƒt is a time specific effect (common to all countries), and εit ~ iid N (0, σ2), i = 1, …, N and j = 1,…, T3.
We implement the previous model as a least square dummy variable model (LSDV), that is we estimate by ordinary least squares the following equation (2):

Where Ci is a dummy variable associated with country i, with associated coefficients υi, and Ft is a dummy variable associated with year t, with associated coefficients γt.
30In order to avoid, or at least minimize, endogeneity concerns, which may arise due to reverse causality, or simultaneity, between some explanatory variables and labor market gaps, we use one period lags of the most suspicious variables, namely the growth rate, the fertility rate, GDP per capita and social spending measured as a share of GDP. Gendered patterns in the labor market may indeed have simultaneous reciprocal relations with each of the four variables previously mentioned: they influence fertility decisions (see eg Bloom et al., 2009; Da Rocha and Foster, 2006; Currie and Schwandt, 2014), and growth rates or levels of development (Dollar & Gatti, 1999; Onaran, 2107). Because social spending includes family and housing policies, among other things, it is also partly determined by the gendered outcomes in the labor market. In addition, we use gendered lagged unemployment rates in the participation gap equation: in this way we aim to take into account added or discouraged worker effects. Finally, we also include gendered lagged participation rates in the unemployment gap equation in order to avoid simultaneity between participation and unemployment rates. It seems reasonable to think that labor market gendered features have no direct contemporaneous effects on the other explanatory variables.
31In all estimations, we compute robust standard errors (with the Huber/White correction), and we allow country clustering in models (2). This procedure ensures that statistical significance tests are valid whenever disturbances εit are correlated across time and for each country.
- 4 Public social expenditure are provided by general government, and includes cash benefits, direct in (...)
32For each dependent variable, we run two series of estimations. In a first step, we only introduce time (year) dummies and country fixed effects (Model 1). This allows us to identify time trends in gender gaps, as well as clusters of countries in terms of labor market gender gaps. We then add macroeconomic variables (per capita GDP level and growth, employment sectorial shares), women-specific variables such as fertility rate and the share of women having superior or intermediate education levels, labor market variables (EPL, unionization, hourly minimum wage, participation tax rate and unemployment replacement rate), product market regulation index, and public social expenditures’ share of GDP4 (Model 2).
Table 2. Gender gaps for EU-14 (2003-2015)
|
2003-2007
|
2008-2015
|
|
Variable
|
Mean
|
sd
|
Min.
|
Max.
|
N
|
Mean
|
sd
|
Min.
|
Max.
|
N
|
|
Participation
|
|
|
|
|
|
|
|
|
|
|
|
Participation rate
|
71.73
|
5.09
|
61.6
|
80.60
|
75
|
73.08
|
4.93
|
62
|
81.7
|
120
|
|
Men’s participation rate
|
78.94
|
3.26
|
72.60
|
84.60
|
75
|
78.7
|
3.53
|
72.2
|
85.3
|
120
|
|
Women participation rate
|
64.49
|
7.81
|
48.7
|
77
|
75
|
67.48
|
6.92
|
51.1
|
79.90
|
120
|
|
Participation gap
|
14.45
|
6.33
|
3.4
|
26
|
75
|
11.22
|
4.92
|
2.8
|
23.4
|
120
|
|
|
|
|
|
|
|
|
|
|
|
|
Employment
|
|
|
|
|
|
|
|
|
|
|
|
Employment rate
|
66.77
|
5.42
|
56.1
|
77.40
|
75
|
66.12
|
6.75
|
48.8
|
77.90
|
120
|
|
Men employment rate
|
74.01
|
3.76
|
67.09
|
82.2
|
75
|
71.23
|
5.72
|
57.9
|
83.2
|
120
|
|
Women employment rate
|
59.51
|
8.27
|
42.8
|
73.40
|
75
|
61.03
|
8.43
|
39.9
|
74.10
|
120
|
|
Employment gap
|
14.51
|
6.95
|
2.7
|
28.3
|
75
|
10.19
|
5.14
|
1.5
|
25.8
|
120
|
|
Part-time employment
|
|
|
|
|
|
|
|
|
|
|
|
Part-time rate
|
18.62
|
9.31
|
4.1
|
46.3
|
75
|
21.2
|
9.48
|
5.4
|
50
|
120
|
|
Men part-time rate
|
7.25
|
4.8
|
1.6
|
22.5
|
75
|
9.58
|
5.07
|
2.6
|
26.5
|
120
|
|
Women part-time rate
|
32.79
|
15.36
|
7.5
|
75
|
75
|
34.81
|
15.35
|
9.80
|
77.10
|
120
|
|
Part-time gap
|
25.55
|
11.93
|
5.5
|
53.2
|
75
|
25.23
|
11.87
|
5
|
52.7
|
120
|
|
Unemployment
|
|
|
|
|
|
|
|
|
|
|
|
Unemployment rate
|
7.07
|
2.1
|
3.8
|
11.5
|
75
|
9.56
|
5.36
|
3.4
|
27.5
|
120
|
|
Men unemployment rate
|
6.34
|
1.9
|
3
|
11.4
|
75
|
9.47
|
5.14
|
3
|
25.6
|
120
|
|
Women unemployment rate
|
8.04
|
3.08
|
4
|
16.3
|
75
|
9.72
|
5.91
|
3.7
|
31.4
|
120
|
|
Unemployment gap
|
1.7
|
2.56
|
-1.2
|
9.6
|
75
|
0.25
|
2.43
|
-7.2
|
7.10
|
120
|
|
|
|
|
|
|
|
|
|
|
|
Source. Eurostat, author's calculations
33Table 2 shows that average labor market gender gaps decreased between 2003 and 2015: participation and employment rate gaps were both over 14 points on average in the 2003-2008 period; they declined respectively by over 3 and 4 points. Looking at figures 1.a and 1.b, the decline in the participation and employment gaps between 2003 and 2015 was continuous, and more pronounced during 2008-2010. Meanwhile, the average unemployment rate gap fell from 1.7% to 0.25% (figure 2). However, different gender patterns in terms of labor market outcomes account for these changes. While participation and employment gaps have continuously declined, part-time and unemployment gaps have nonlinear patterns. The narrowing of the participation gap is in fact rooted in an increase in women's activity rates, while men's participation has been roughly stable. The employment gap decrease is due to both a fall in the employment rate for men and an increase in female employment.
Fig 1.a.a. Participation rates of women and men

Source. Eurostat, authors’ calculations
Fig 1.a.b. Participation gap

Source. Eurostat, authors’ calculations
Fig 1.b.a. Employment rates of women and men - EU-14 (2003-2013)

Source. Eurostat, authors’ calculations
34The change in the unemployment gap is explained by a sharper increase in male unemployment compared to the rate for women, which can be seen in Figure 2. This shows the unemployment gap became negative in 2009. In other words, when looking at unemployment, the fact that the gap between rates for men and women has been narrowing stems mainly from the worsening of men's position.
Fig 1.b.b. Employment gap - EU-14 (2003-2013)

Source. Eurostat, authors’ calculations
Fig 2.a. Unemployment rates of women and men - EU-14 (2003 – 2015)

Source. Eurostat, authors’ calculations
Fig. 2.b. Unemployment gap - EU-14 (2003 – 2015)

Source. Eurostat, authors’ calculations
35The mean part-time employment gap for the two periods under study is similar, but Figure 3 shows a sharp rise in the gap before the outbreak of the crisis and then a decline: while a much higher percentage of women are still working part-time compared with men, the part-time employment rate for men increased sharply, rising to 9.58%.
Fig 3.a. Part-time rates of women and men - EU-14 (2003-2015)

Source. Eurostat, authors’ calculations
Fig. 3.b. Part-time gap - EU-14 (2003-2015)

Source. Eurostat, authors’ calculations
36We gain insights about countries and groups of country specificities by looking at their labor market and institutional background (Table 3 in the Appendix). For this purpose, we refer to an “extended” welfare regime typology (Esping-Andersen, 1990; Ferrera, 1996; Tridico & Paternesi Meloni, 2018). In line with conventional wisdom, we observe that Scandinavian countries have low participation, employment and unemployment gaps, while the part-time gap is moderate. The share of total employment in the public and social services reaches nearly a third, and government social spending represents 26% of GDP, the highest level in our European sample. We find both social welfare state and “universal breadwinner” regime features (Anxo et al., 2007). The labor market combines rather low regulation, and a high degree of unionization, as well as no centralized minimum wage policy. Product market regulation is intermediate. The Mediterranean group of countries exhibits the highest gender gaps in terms of participation, employment and unemployment, while the part-time gap is the lowest, matching Anxo et al.’s “exit or full-time” typology. The institutional background corresponds to a highly regulated labor and product market, with low levels of public social spending, public employment, as well as low levels of minimum wage and union coverage. The Anglo-Saxon and Continental groups are rather close in terms of gender gap behavior, with intermediate participation and employment gaps and high part-time gaps. They are also quite similar regarding the extent of public sector employment, but the share of government social spending is lower in the Anglo-Saxon group. In addition, the labor and product markets are more regulated in the Continental group (while overall regulation in the Anglo-Saxon group represents the sample’s lowest levels).
37When we only take into account yearly and country fixed effects (Table 4 in the Appendix), the results show a narrowing of the participation gap and of the employment gap over time relative to 2004. The results are not significant when it comes to the part-time gap. Concerning the unemployment gap, the results are significant during the crisis and after, with a narrowing of the unemployment gap starting in 2008. These changes reflect the order in which “male” and “female” sectors were affected during the crisis. Industries that were predominantly male, such as construction, were hit first, reducing the unemployment gap by bringing the male rate closer to the female rate. However, industries that were predominantly female, such as the service sector and public sector, were subsequently affected, slowing the reduction in the unemployment gap.
38We can split the sample into three groups of countries (with the UK being our base category), according to their similarities in terms of the four gender gaps’ behavior, once common time trends are accounted for. The first group comprises Austria, Belgium, Germany, and the Netherlands. We find that in this group participation gaps tend to be smaller, but other gaps increase, or are at best left unchanged (employment gaps for Austria and Germany). In these countries, less participation inequality between genders is associated with a lower rate of employment, more part-time, and more unemployment for women as compared with men. This reflects a modified “Continental model” group, whose model includes more part-time work, especially for mothers, due to childcare and parental leave policies. In the second group, where we find Denmark, Finland, France, Portugal and Sweden, all gender gaps narrow (except in the French case where the part-time gap remains steady), except for the unemployment gap. This group of countries corresponds to an extended Scandinavian or Nordic model. In the third group of countries (Greece, Ireland, Italy, and Spain) the participation gap is higher, as well as the employment and the unemployment gaps (with the exception of Ireland), while there is a decline in the part-time gap. We find countries from the Mediterranean model, to which Ireland is added, where the work model for women seems to be “exit” or “full time” and confirms the findings of Anxo et al. (2007). Notice that this clustering remains essentially the same if we take into account path dependency and include lagged dependent variables (the results are not shown here but are available upon request). Of course, these country differences reflect cultural, institutional, social and economic discrepancies between countries, some of which the subsequent models try to disentangle.
39We now turn to the results of the full model (Table 5 in the Appendix). Starting with the macroeconomic variables, we note that higher (lagged) per capita GDP levels are associated with lower participation, employment and unemployment gaps, while we show the positive effect of economic growth on the employment and unemployment gaps. The growth rate effect is somewhat ambiguous and reflects sector-leading growth characteristics and potential corresponding labor shortages. For example, if growth is service sector oriented, one might expect higher labor market participation and employment for women; on the contrary, if growth is strong in male-dominated sectors, one might observe an increase in the participation gap, due to higher male participation in the workforce. Good times appear overall more beneficial to men than to women. Employment sectorial shares indicate that the larger the service sector, the smaller the employment and unemployment gaps. In addition, a large public sector is also associated with a smaller employment gap. We observe a gendered discouraged worker effect, as both the female and male (lagged) unemployment rates have a significant effect on the participation gap, with a stronger effect for men.
40Turning to women and family specific variables, we find that the (lagged) fertility rate is significantly associated with the employment and unemployment gaps with a negative sign. These results raise some endogeneity concerns, despite using the lagged fertility variable. However, these results are in line with previous findings showing that higher women’s unemployment decreases (postpones) fertility decisions (Da Rocha & Fuster, 2006; Currie & Schwandt, 2014). Moreover, when a larger number of women have degrees in higher education, there is a decrease in the part-time gap: all else being equal, higher education seems to be one of the key factors fostering access to full-time employment for women. With respect to the participation gap, middle education share appears to make a difference.
41Fiscal and social protection policies exert mixed effects on gender gaps. The participation tax rate is found to decrease the part-time rate gap. This result is at odds with the expected standard effect of taxation on the labor supply, and it might reflect other omitted institutional effects. We also note that there is no clear-cut identified pattern in the overall OCDE data between the participation tax rate and women’s behavior (Nikolka, 2016). Because we use one measure of the participation tax rate that refers to a family with two children, we might also be picking up effects related to special tax allowances, or tax credits for families with children, which are not otherwise accounted for in our model. Social spending narrows the unemployment gap at given participation rates. This effect could indicate that active labor market policies may be more beneficial to women. We note otherwise no disincentive effects of social spending. However, these results call for further analysis of the kinds of social spending.
42Looking at institutional features, our results show that labor market institutions are not gender neutral, and point to possible public policies that could promote more gender equality. Indeed, it appears that unionization is associated with narrower participation, employment and unemployment gaps. The standard analysis of the labor market associates unions with lower employment and higher unemployment rates because unions demand higher real wages. Our results indicate that this mechanism seems to be more effective for men, and the wage returns of unionization for women may be lower. This could be the result of an asymmetric presence of unions across sectors or an uneven enforcement of labor market institutional settings. Another approach underscores the collective and anti-discriminatory roles of unions, which would benefit women the most.
43Employment protection legislation has contrasted effects: permanent contract protection legislation has significant negative effects on employment, as well as unemployment, but is associated with a higher part-time gap. This tends to indicate that contra-cyclical effects of employment protection legislation on regular contracts are more effective for women than for men. In other words, in the absence of protective labor legislation, women are more at risk of being laid off—the so-called buffer effect. However, protection of temporary workers, which measures the strictness of regulation on the use of fixed-term and temporary contracts, increases the employment gap, penalizing female employment. We interpret this evidence as a “status” segregation effect, since women more often hold temporary contracts, depending on the sector. Moreover, employment protection of permanent jobs also leads to a higher part-time gap, while both the protection of temporary workers and minimum wage legislation have the opposite effects.
44Finally, we find that the product market regulation index (PMR) is associated with higher participation and employment gaps. Because less regulation of the product market is thought to be favorable to employment, we conclude that men benefit less than women do from product competition gains in terms of employment and activity rates. This might indicate a buffer or substitution effect in industries where competition is high, with firms favoring a more flexible or relatively cheaper labor force. On the contrary, the primary sector, where regulation is higher and jobs are more secure, would tend to be male dominated.
45The situation of women in the labor market has largely improved in the 14 European countries in recent decades: participation and unemployment gender gaps have narrowed. We raise the question of how the crisis, which broke out in 2007-08, affected this process and whether it changed this trend.
46The results of our paper should be read as a robust association between gender gaps and certain variables and not necessarily as causal relationships. Our results show that gender gaps and inequalities declined during the crisis, particularly concerning participation and employment. The large number of women working in the service and public sectors had overall protective effects for women’s employment. When it comes to the employment gap, growth rates have significant positive effects—the gap widens. Good times are more beneficial to men than to women in terms of employment.
47In addition, we highlight several results with respect to labor and product institutions. Higher unionization and minimum wage levels reduce gender gaps. Greater unionization is associated with lower participation, employment and unemployment gaps. Higher minimum wages are associated with a reduction in the part-time gap. Product market regulation, however, seems to have relative effects more beneficial to male participation and employment, while labor market regulation has mixed gendered outcomes. Importantly, we also find that public and fiscal policies (social spending, unemployment replacement rates, and participation tax rates) have contributed to narrowing gender gaps. In any case, labor and product market institutions are not gender neutral.
48Women have massively entered the labor market over the last few decades and this fundamental shift appears permanent. However, they remain more vulnerable than male workers in the labor market: they are more present in segregated, lower-paying sectors, and are subject to the buffer effect. Women remain dependent on childcare and are more affected by attempts to lengthen the workweek. Hence the importance of better infrastructure, a stronger institutional environment, and public policies that encourage their participation in the labor market, their employment, and gender equality.
49From a broader perspective, the present research seeks to help build bridges between gender studies and regulation theory approaches (Amable et al., 2005, 2011). In particular, our results highlight that social protection, professional relationships, the education system, and labor market institutions are not neutral from a gender perspective. In the near future, we hope to be able to draw a more complete picture of how the evolution of gender gaps is related to different “varieties of capitalism”.