Do gender regimes matter? Gender differences in involvement in anti-austerity protests - a comparison of Spain, Sweden and the United Kingdom

ABSTRACT Responses to the Great Recession are varied across welfare states and gendered in their consequences. Combining gender, social policy and social movement scholarship, this paper investigates how the differential policy responses to the Financial Crisis in three European countries shaped gender-differences in anti-austerity demonstrations. We compare the involvement and characteristics of women and men in anti-austerity protests using data collected at street demonstrations (2010–2012). We conduct cross-national multi-level analysis of demonstrators from countries representing different gender regimes (Spain, Sweden and the United Kingdom). Our results show that gender regimes have a significant impact on women’s and men’s involvement in anti-austerity protests. We thus make an important contribution to research on gender differences in participation in anti-austerity demonstrations post-Great Recession. Our comparison of women’s and men’s participation in anti-austerity street demonstrations suggests that at the country or regime level resources matter more than grievances, but that grievances matter at the individual level. This innovative paper links scholarship on gender regimes with research on protest participation. Resources and experiences of grievances are shaped by gender regimes which provide access to decision-making and social support. We reveal novel insights into the connection between gender regimes and demonstration participation.


Introduction and overview
The financial crisis, which began in 2008 with the collapse of the US American Bank Lehman Brothers, was at first labelled a 'mansession', initially affecting men working in the private sector (Walby, 2015, p. 76). However, in the following years austerity measures began to affect women more adversely than men, because women are disproportionately employed in the public sector and are more likely to be public sector service clients (Rubery & Rafferty, 2013). Governments' responses, such as the adoption of austerity measures and the (continued) erosion of welfare states, resulted in a political crisis and caused a cycle of contention (Della Porta, 2015;Peterson, Wahlström, & Wennerhag, 2015). Anti-austerity activism includes multiple forms and claims and represents some continuity with global justice movements (Della Porta, 2017a). It encompasses industrial action, demonstrations of labour unions, the formation of radical left parties and the collective occupation of public space (Hayes, 2017). Such antiausterity mobilisations address material issues as well as 'the nature and meaning of democracy' and can therefore be understood as 'pro-democracy movements' (Flesher-Fominaya, 2017, p. 14). However, the rise of populism is another response to austerity. We only analyse anti-austerity movements from the left (cf. Della Porta, 2015) even though we acknowledge that right-wing populist movements also represent a significant response to neoliberalism and austerity (Roth, 2018). Although we lack data for comparison, we suspect that the gender dynamics of anti-austerity movements from the left differ from those of right-wing populist movements.
The ample research addressing protest participation in general and leftist antiausterity protests in particular (Della Porta, 2015;Flesher-Fominaya, 2017;Giugni & Grasso, 2015a;Hayes, 2017;Peterson et al., 2015) has so far neglected gender differences in these forms of collective action. Furthermore, the wealth of comparative social movement research either neglects gender dimensions of contentious action (Kriesi, Koopmans, Duyvendak, & Guigni, 1995;McAdam, McCarthy, & Zald, 1996;McAdam, Tarrow, & Tilly, 2001) or is solely concerned with the strategies and success of women's movements (Banazak et. al., 2003;Beckwith, 2013;Katzenstein & McClurg Mueller, 1987;Mazur, McBride, & Hoard, 2016). Women's participation in protests outside of women's movements is yet under-researched and therefore our paper contributes to cutting edge research on gender in social movements. Leftist anti-austerity mobilisation provides an important opportunity to study the involvement of women in demonstrations. It is established that there are marked differences in the proportion of women and men at demonstrations across countries (Dodson, 2015). However, we still know very little about what explains unequal demonstration participation of men and women in times of austerity. Our paper closes this gap.
We compare the participation of men and women in leftist anti-austerity protests in three countries, Spain, Sweden and the United Kingdom (UK), which have been differently affected by and responded differently to the great recession. Moreover, they represent different gender regimes (Walby, 2009), which should not be confused with more narrowly conceived gendered welfare regimes 1 . In our cross-national study of anti-austerity demonstrators in Spain, Sweden and the UK, we investigate whether unequal gender participation of men and women in anti-austerity protests is best explained by the gender regime (country) in which the demonstration took place, by individual characteristics of demonstrators, or by both. We control for demonstration effects to understand whether the non-random selection of demonstrations affects the interpretation of the country-level (and regime level) effects. We explore this using data from 16 anti-austerity demonstrations (see Table 1), extracted from the Caught in the Act of Protest project database (Klandermans, Giugni, & Peterson et al., 2009), which uniquely collected data at over ninety demonstrations in the period 2010-2012. We selected all of the demonstrations against austerity that had been surveyed by the project during that time period in the three countries we study. Previous analysis of May Day and change demonstrations in this data-set confirmed that there has been a 'homogenising' effect across so-called old and new movements (Eggert & Giugni, 2012), which we see affirmed in the presence of universalistic goals of the demonstrations we analyse (see Appendix 2).
We start out with an overview over the main explanations for women's involvement in anti-austerity protests. This section discusses gender regimes and our country cases, including the extent and form of austerity measures, the general climate of anti-austerity mobilisation and individual characteristics. We continue by describing our data and methods. Our results show that gender regimes have a significant impact on women's involvement in anti-austerity protests. Our research therefore makes an important contribution to the knowledge of gender regimes and protest participation and how they are connected.
Austerity and resistance in UK, Spain and Swedenmain explanations to women's involvement in anti-austerity protests Feminist scholars immediately pointed out that the financial crisis is gendered 'in terms of causes, impact and responses' (Annesley & Scheele, 2011, p. 335). Some argue that women's under-representation in decision-making positions and masculine norms in financial institutions contributed to irresponsible risk-taking (for a critical assessment see Prügl, 2012). Women's organizations have early on emphasized that the financial crisis significantly affects women (Pearson & Elson, 2015). In September 2009, the European Women's Lobby (EWL) the largest alliance of women's non-governmental organizations from twenty-seven EU member states and three candidate countries highlighted that a gender perspective on the financial crisis is crucial and warned that policy responses could result in an increased gender gap. In this section we give an overview over explanations for women's involvement in anti-austerity protests. Drawing on Walby (2009Walby ( , 2015 we first introduce the concept of gender regimes, then we give an overview of austerity measures in the three country cases and the protests against them, and lastly, we summarise individual level factors explaining women's involvement in anti-austerity protests.

Gender regimes
Gender regimes are highly complex and comprise multiple levels which include the overall social system, the degree of gender inequality and a series of social practices. Indicators of gender regimes comprise gendered inequality in the workplace, education, political participation, equality legislation, gendered civil liberties and childcare public expenditure (Walby, 2009, p. 303). Different policy domains may vary with respect to gender equality (Walby, 2009). 'Domestic' and 'public' gender regimes represent ideal types, though in reality gender regimes of countries vary between these poles constituting a continuum (Walby, 2009). In domestic gender regimes, women do the majority of unpaid care work, are underrepresented in the paid labour force and politics, and have limited access to abortion. Although not equal to men, women are significantly more involved in the public sphere (the labour market and politics) in public gender regimes, where they also have easier access to abortion than women in domestic gender regimes. Among public gender regimes, more social democratic and more neo-liberal gender regimes can be identified (Walby, 2015). Social democratic public gender regimes are characterised by the provision of public services including childcare and employment regulations, whereas in neo-liberal public gender regimes the market plays a central role in the provision of services including childcare and the regulation of employment through market mechanisms. However, over the last few decades, state support for child-care has increased in different gender regimes including economies undergoing recession (Walby, 2009). 'The most important differences between the neoliberal and social democratic forms of the public gender regime lie in the gendered depth of democracy and the degree of gender inequality' (Walby, 2015, p. 148). In contrast to the concept of gendered welfare regimes (see endnote 1), the concept of gender regimes thus addresses public services, women's employment, and women's political involvement which is central for our analysis. Our analysis includes one social democratic public gender regime and two neoliberal public gender regimes. The UK represents a neo-liberal public gender regime, whereas Sweden represents a social democratic gender regime due to the high integration of women in politics and paid employment and the dual earner family model. Spain had experienced a shift from a domestic to a social democratic public gender regime in the early 2000s, however, austerity politics following the financial crisis transformed Spain into a neoliberal public gender regime (Lombardo, 2017). Among the three countries, Spain was hardest hit by the great recession, whereas Sweden was only moderately affected and the UK less than Spain but more than Sweden. In summary, we select these three countries for analysis because they represent different ideal-types of gender regime: Sweden as social democratic, the UK and Spain both neo-liberal though varying in their experience of the great recession and the policy responses to it.

Gendered government responses to the great recession in comparative perspective
Following the financial crisis, the EU and its member states adopted a 'disciplinary neoliberalism' which combined deregulation and liberalization of the market, austerity measures, fiscal and monetary policies, and a marginalisation of gender and other aspects of social equality (Kantola & Lombardo, 2017). Even though the EU adopted gender mainstreaming as a principle in 1997 and demanded from candidate states to adopt gender equalities in order to be accepted as member states (Roth, 2008), gender was not mainstreamed into crises measures (Gonzalez Gago, 2017). In fact, EU gender equality policies were increasingly dismantled and gender equality-specific budgets were significantly reduced (Jacquot, 2017). These shifts started prior to the financial crisis, but have been intensified by the economic crisis (Jacquot, 2017;Kantola & Lombardo, 2017). The increased marginalisation of actors and policies promoting gender equality in the EU during the financial and political crises undermined efforts to address the gendered impact of the responses to the crises (Weiner & MacRae, 2017).
Since the 1980s, neo-liberalism and associated welfare state retrenchment have resulted in growing inequality. Workfare policies introduced by conservative as well as by social democratic or centre left parties around the world led to stagnating and falling wages and increasingly precarious working conditions. Not even Sweden, which has traditionally been characterised by a generous welfare state has been spared from neoliberalist policies accompanied by worsening working conditions, a rise in youth unemployment, and urban segregation (Schierup and Alund, 2011;Schierup, Ålund, & Neergaard, 2018;Therborn, 2018). However, despite the ubiquity of neoliberal policies, governments responded differently to the great recession (Kiess, Norman, Temple, & Uba, 2017).
Policy responses included support for the financial sector, stimulus packages and public sector cuts. Farnsworth and Irving (2012) distinguish three responses: first, economies most severely affected by debt-related crises in which public sector cuts were inevitable (including Spain); second, economies which introduced austerity measures even though the economic conditions did not require public sector cuts (including Sweden and the UK); and finally a group with fewer cuts (none of our study countries). Responding to the great recession, in the UK public sector cuts considerably exceed other ways to balance the budget (Farnsworth & Irving, 2012, p. 138). In contrast Sweden reversed the introduction of neoliberal policies and returned to earlier, generous policies (Kiess et al., 2017). Thus, rather than the depth of crisis the interaction between the generosity of the welfare state and the government composition explain the response to the economic crisis. More generous welfare states (such as Sweden) are more likely to respond with counter-cyclical expansion of public spending, whereas the degree of retrenchment or expansion varies in the less generous welfare states due to the government (Starke, Kaasch, & Van Hooren, 2014). Retrenchment and expansion are only two possibilities of welfare state transformation which also include flexicurity, welfare adjustment and welfare protectionism (Haeusermann, 2012). New social policies address 'new social risks' which tend to be concentrated among women, the young and the low skilled (Bonoli, 2005). All these measures have gendered consequences, which we now discuss.
As noted above, Sweden, the UK and Spain responded differently to the crisis with the most severe cuts in Spain and some extensions of social benefits in Sweden. Forced by the EU, Spain made significant public sector cuts and deregulated the labour market which increased women's unemployment, worsened working conditions, and increased the gender pay gap (Lombardo, 2017). The measures included downgrading gender equality policies by eliminating the Equality Ministry, ceasing the implementation of the Dependency Law in 2010 which supported women's labour force participations by providing financial support for day care work, cuts to children's care facilities, and postponing the extension of paternity leave (Gonzalez Gago, 2017). The Labor Market Reform of 2012 which made it easier to fire permanent workers contributed to briefly decreasing the gender employment gap by worsening men's working conditions and contributing to a temporarily slightly increased women's employment rate as women compensated for male un-and underemployment (Gonzalez Gago, 2017). By raising the retirement age and the required years to access minimum as well as full pensions, the pensions reform, which was passed in 2011, also had significant negative effects for women who usually work fewer years due to care obligations (Gonzalez Gago, 2017).
The UK restricted poverty protection to the 'poorest of the poor' through cutting all social benefits including those directed at low-income families. In June 2010, the newly elected Conservative-Liberal Democrat coalition government announced its plan to close the deficit within the following five years through a combination of 79% spending cuts and 21% tax raises starting from 2009/2010 (Reed, 2017). The tax rises contributed to less than 11% deficit reduction and changes to taxation even included a real-terms increase of tax-free personal allowance, whereas child tax credits and working tax credits were cut. The biggest losers of the public sector cuts were lone parents (90% women), working age couples as well as single pensioners (75% women). Public sector cuts affected health care, community funded social care, housing cuts, transport cuts, and cuts to higher and further education. Reed (2017) estimates that women's income cuts were twice as high as men's.
As noted above, although Sweden also experienced a gradual demise of the welfare state (Schierup and Alund, 2011;Therborn, 2018), this was to some extent reversed during the great recession (Kiess et al., 2017). In contrast to Spain and the UK, Sweden engaged in a 'stabilizing strategy' and extended the amount and coverage of its social benefit and increased the in-work tax credit to preserve a high level of consumption and to prevent 'traumatization' and declining birth rates due to family and other social benefits (Bothfeld & Rouault, 2015, p. 76). The impact of activation policies, which seek to encourage paid employment by lone mothers, varies across welfare states. Whether employment prevents poverty depends very much on the type of work that is available. Short-hours part-time work typical for the UK may even increase poverty risks especially if it is accompanied by a cut in benefits (Warren, 2015). In contrast, in Sweden, the poverty risk of single mothers is much lower due to higher earnings and higher levels of benefit entitlements in case of unemployment. However, even in Sweden, minimum income benefits fell compared to median earnings (Jaehrling, Kalina, & Mesaros, 2015). The commodification, corporatisation and marketisation of children and the elderly has considerable consequences for care workers (predominantly women) as well as those in need of care. For-profit care work tends to be 'more codified, less personalized, poorly paid, and less sensitive to the changing needs of the people to whom it is oriented' (Farris & Marchetti, 2017, p. 126). Although a corporatisation of care work can be observed across all Western countries, the extent varies significantly in Sweden and the UK, where it is far more extensive and over 80% of elderly care is provided by for-profits. Sweden introduced forms of care marketisation in the early 1990s, but at the end of the 2000s only 17% of elderly care and only 8% of child-care were provided by for-profits (Farris & Marchetti, 2017, pp 119-120). Furthermore, the gender pay gap is much lower in Sweden compared to the EU average. Nevertheless, in Sweden a gender pay gap of 8% can be noted, Swedish women represent only 28% of all high-income earners, face striking occupational segregation and have been strongly affected by public sector cuts (Tsarouhas, 2011, p. 434). Thus, even if women have been hardest hit in Spain, to a lesser extent in Britain and least in Sweden, in all three countries women have been affected by public sector cuts.
How is the fact that the economic and political crises are gendered reflected in protests in times of austerity? In the next section, we turn to anti-austerity protests in Europe.

Anti-austerity protests in comparative perspective
A wealth of scholarship has analysed anti-austerity protests responding to the financial, economic and democratic crisis associated with the Great recession (for example Della Porta, 2015; Giugni & Grasso, 2015a;Peterson et al., 2015). Like the Global Justice Movement, contemporary anti-austerity movements comprise networks of groups and organisations on the left including labour, alter-globalisation and women's movements. They include particularistic mobilisations organised by trade unions and addressing particular material issues; and universalistic demonstrations including the occupations of public space that take on broader grievances and bring together a diversity of actors (Klandermans and van Stekelenburg, 2016;Sabucedo, Gómez-Román, Alzate, van Stekelenburg, & Klandermans, 2017). However, as part of larger mobilisations against neoliberalism and austerity, trade unions go beyond particularistic mobilisation and other actors in making universalistic claims. In fact, Peterson, Wahlström, Wennerhag, Christancho, and Sabucedo (2012) argue that far from being a 'hollow ritual', May Day demonstrations in 2010 addressed grievances following rising unemployment thus representing protest against austerity (see also, Appendix 2). Thus, trade union demonstrations need to be included in analyses of anti-austerity protest in Europe (Peterson et al., 2015).
Participants in anti-austerity protests include groups facing 'new social risks' (Bonoli, 2005) and 'new losers' of globalisation (Della Porta, 2015): young people who consider their future as insecure; retired people facing pension reductions and cuts to much needed social services; and public sector workers. As della Porta (2015, p. 65) writes 'protesters do not belong only to a specific precarious class, but are rather moved by a plural alliance of citizens whose existence is made less and less secure in neoliberalism'. Della Porta (2015) does not consider gender, but as we explain above, women are disproportionately insecure compared to men.
Research on recent anti-austerity protests has been carried out in many countries. Such studies either lack cross-national comparison (Gerbaudo, 2017;Giugni & Grasso, 2015b;Ketelaars, Walgrave, & Wouters, 2017) or hardly address gender Flesher Fominaya, 2017) except as control variable. Thus, we make a valuable contribution to the literature by conducting a cross-national comparison, which also considers women as a significant sub-group of 'new social risk' groups and 'new losers'.
Only a few studies of contemporary protest focus on gender and feminist perspectives, and gender differences in protest in particular (Roth & Saunders, 2019). Fuentes (2015) notes that despite the conscious adoption of a non-hierarchical organisational structure and distance from traditional political organisations such as political parties or trade unions, initially feminist proposals did not make it on the agenda of the 15M movement. Conflicts between the feminist committee and the rest of the movement lead to a reactivation of feminism and eventually to a transformation of the 15M movement (Fuentes, 2015, p. 363). In France and the UK, minority women felt ignored by the 'raceless discussions of the white Left' (Bassel & Emejulu, 2017, p. 99) and created DIY activist spaces. While these studies provide important insights in women's mobilisation in times of austerity, what is lacking so far are systematic analyses of women's involvement in anti-austerity protests in comparative perspective.
Until recently, studies have found a male-biased gender gap in protest participation. However, longitudinal evidence shows it is closing (Gallego, 2008). In 1974 60% fewer women than men protested, but this had reduced to 13% in 2002 (Stolle & Hooghe, 2011). By 2012, in Europe the most frequent protesters are actually more likely to be women than men (Saunders & Shlomo, 2016). However, there is variation across demonstrations, varying by issue (Gómez-Román & Sabucedo, 2014), and across countries: In May Day demonstrations over 60% of the Swedish participants were women compared to less than 40% in Spain and the UK (Peterson et al., 2012). We consider that it is particularly important to compare men's and women's involvement in anti-austerity demonstrations for three reasons. First, men and women have been differently affected by public sector cuts (Pearson & Elson, 2015). Second, austerity has led to a resurgence of feminism (Evans, 2015). Third, feminism may play a factor in the revitalisation of the left (Maiguashca, Dean, & Keith, 2016). Before we introduce data and methods, we briefly summarise individual level explanations for protest participation.

Individual level explanations
Social and political contexts such as the financial crisis and public sector cuts shape participation in demonstrations and reflect participation motivations (Peterson et al., 2012). However, as we indicate above, with reference to the 'new losers' (Della Porta, 2015) who face 'new social risks' (Bonoli, 2005), broader social and political contexts impact individuals. It is already well established that grievances, among other factors, can explain preparedness for protest (Klandermans, van Stekelenburg, Damen, van Troost, & van Leeuwen, 2014). Framed in this way, sufferance from new social risks can provide people with the motivation to demonstrate in anti-austerity protests to protect their own interests. It is certainly very clear that scholarship on anti-austerity protests has helped to bring about a resurgence of grievance theories (Della Porta, 2015;Kern, Marien, & Hooghe, 2015). Peterson, Wahlström, and Wennerhag (2018) additionally found that women disproportionately self-identify as part of a discriminated category which might also motivate them to participate in protests with marginalised and discriminated groups.
But there are also other factors that are known to support protest participation at the individual level, which might compliment or counteract grievance theory. Biographical, structural and psychological resources are known predictors of participation in demonstrations (e.g. Dalton,1996;Schussman and Soule, 2005;Klandermans et al., 2014;Saunders, 2014). Biographical aspects concern age, employment and education which are related to available time, monetary and knowledge resources (McAdam, 1986;Peterson et al., 2018). Structural resources refer to being in the right place at the right time to hear about protest and are therefore related to inclusion in physical or online networks. Psychological resources refer to interest in politics, which is known to be a significant predictor for a variety of forms of political participation (Dalton, 2005).

Hypotheses
Based on our review of the literature on gender regimes, differential effects of austerity across countries and individual-level predictors of anti-austerity protest, we propose two sets of hypothesesat the country and individual levelto determine which combination of factors lead women to participate in anti-austerity protests. Since gender regimes are ideal types, we operationalise different gender regimes by comparing countries.

Regime (country)-level
We purport that women in countries with policies and practices that support domestic, political and social equalitysuch as a social democratic public gender regime like Swedenwill be more likely to participate in anti-austerity demonstrations. This is because they will feel more empowered as women are included in politics to a greater extent and receive more state support compared to neoliberal public gender regimes such as Spain and the UK. This leads to our first hypothesis: H1: In a social democratic public gender regime (Sweden) we find a higher proportion women in anti-austerity demonstrations compared to neoliberal gender regimes (Spain and the UK).
We also posit a contrasting hypothesis, which anticipates that it is not so much empowerment that determines whether women engage in anti-austerity demonstrations, but rather it is the extent to which public sector cuts have been introduced in their countries. Thus, we hypothesise: H2: In countries with significant public sector cuts (UK and Spain) we find a higher proportion women in anti-austerity demonstrations compared to a country which has reversed public sector cuts (Sweden).

Individual-level
At the individual-level, we hypothesise that biographical, structural and psychological factors predict women's participation in demonstrations regardless of the gender regime. This leads to hypothesis three: H3: Female anti-austerity demonstrators have more biographical, structural and psychological resources for protest than male anti-austerity protesters.
However, anti-austerity demonstrators are likely to be different from protesters in general. They are known to consist of the 'new social risk' groups (Bonoli, 2005) and 'new losers' (Della Porta, 2015). Given that public sector cuts affect women more adversely than men, they are more likely to be motivated to participate in protest to protect their own interests. We include this as a separate hypothesis because social movement scholars have, since the late 1970s, had a tendency to favour theories of agency (H3) to grievance theory (H4) until a recent resurgence of the latter in relation to anti-austerity protest. Thus, we expect to find, at the individual-level that: H4. Female demonstrators are more likely than male demonstrators to be motivated to participate in anti-austerity demonstrations to protect their own interests.
Finally, merging country-and individual-level expectations, we anticipate that self-interested motivations are more strongly held by women demonstrators most adversely affected by cuts (Spain and the UK) compared to those in Sweden, who are less adversely affected.
H5. Women compared to men in neoliberal gender regimes such (Spain and the UK) will have stronger motivations to participate in demonstrations for self-interest than women compared to men in a social-democratic gender regime (Sweden).

Data and methods
We analyse data from 2,765 demonstrators at sixteen anti-austerity demonstrations that took place in the UK, Spain and Sweden, 2010-2012 (see Table 1 for a list of demonstrations). Our selection of these demonstrations was determined by the available data. However, there are also good theoretical reasons to choose protest events in these years. As discussed above, during this time-period, unemployment rose significantly, the UK and Spain announced and started to implement austerity cuts, whereas Sweden reversed some of its neoliberal policies. Of course, some austerity measures were implemented later and it would be very welcome to replicate our study with more recent protest data.
We use a mixed effect logistic regression model, in STATA, to discern which factorscountry (regime) or individual level variablespredict whether a demonstrator is a woman. We nest the individual level data within the sixteen demonstrations, to allow us to make comments about the conditioning effects of the specific demonstrations surveyed. The multi-level model is important because a) we do not have a representative sample of anti-austerity demonstrations in each country and b) there are, indeed, significant demonstration effects. The Swedish demonstrations, for example, are all organised by trade unions and parties. Due to space constraints, we focus our analysis on the country (regime) and individual levels and pay less attention to demonstration effects, which, as we show in our analysis, are cancelled out by country (regime) level effects.
The selection of demonstrations by national protest surveying teams was beyond our control. Given the impossibility to retrospectively select a representative sample of demonstrations in each country, we additionally repeated our analysis using the European Social Survey (ESS, Waves 5 and 7, 2010 and 2012, n = 1178), with design weights applied. It is important to note, however, that the ESS sample does not disaggregate demonstrators by protest issue and it therefore includes data on demonstrations in times of austerity rather than those that are specifically anti-austerity. However, we do know, based on protest event analysis data from the POLCON (Political Conflict in Europe in the Shadow of the Great Recession) project (2019) that between 2008-15, 18.5% of Swedish demonstrations reported in the media had an economic claim to public institutions, compared to 25.3% in the UK and 27.1% in Spain. We bear these figures in mind in our interpretation of the ESS data and our comparisons with protest survey data.

The protest survey data
The protest survey data were collected using Caught in the Act of Protest project protest survey methodology. This involves state-of-the-art techniques for random selection of demonstrators and for assessing response rate bias. The surveying technique requires around 1,000 demonstrators to be present at a demonstration. The team aimed to distribute up to 1,000 mail-back questionnaires at each demonstration using a 'pointer' or team leader to randomly select respondents (Walgrave & Verhulst, 2011). One-fifth of these questionnaires was accompanied by a short matched numbered face-to-face interview. Since refusal rates for the interview are usually low (c.10%), it is possible to gain an impression of the types of people that do and do not respond to the mail-back questionnaire by comparing the two sub-samples.
The response rates of the protest surveys we analysed range from 15.5% (Occupy London) through to 49.5% (For Employment Not Capital Reforms, Vigo) (Table 1). This makes it essential to test the representativeness of the data. We used statistical tests to compare face-to-face with mail-back-data to understand the extent of non-response bias. We tested for significant differences in the aggregated responses to questions that are repeated in both versions of the questionnaire. We used Fisher's Exact to compare the mailback and face-to-face samples for the binary variable (e.g. man or woman), Kendall's tau-b for ordinal variables (when a firm decision was made, education, previous participation in demonstrations [ever]), and t-tests for interval data (age, satisfaction with democracy, political interest and membership in organisations that staged the demonstration).
There are some small but statistically significant differences between the two subsamples (mail-back compared to face-to-face). Women are slightly more likely than men to complete the mail-back survey (42.7% of face-to-face respondents were female, compared to 46.4% of mail-back respondents). 2 The mean age for those interviewed was 41.0 years, compared to 44.7 for those returning the mail-back survey. 3 Satisfaction with democracy is 0.54 higher, on average, for mail-back respondents (mean = 4.42) compared to face-to-face ones (mean = 3.87). 4 The average number of organisational memberships in both samples is almost identical (0.52 for face-to-face, 0.53 for interviewed), as is the average degree of political interest (3.35 for interviews, 3.34 for postal surveys). Those returning questionnaires in the post had committed themselves to attending the demonstration earlier: 55.5% of mail-back respondents decided to attend over a month ago, compared to 49.7% of those interviewed. 5 Mail-back survey respondents are more highly educated (64.6% claim to have a university education, compared 54.5% of interviewees) 6 and more likely to have previously participated in protests. Only 2.6% of mail-back respondents were complete novices, compared to 5.5% of interviewees. 7 These differences are consistent with other protest surveys (Walgrave, Wouters, & Ketelaars, 2016) and express the need for slight caution in the interpretation of the results. Women are slightly over-represented, even though they remain in fewer numbers to men at most demonstrations that were surveyed.

Choice of variables
Our dependent variable is whether the demonstrator is male or female. Individual level variables measure biographical, structural and psychological resources required for participation in protest and include age, employment, education, political interest, political participation, individual efficacy and instrumental motivation. Finally, we add left-right self-placement (LRSP) as a control variable, given that the demonstrations we survey are left-wing.

Results
Our descriptive analysis shows the proportion of men and women at each of the 16 demonstrations, the aggregated totals across the three countries, and the ESS results ( Table 2). The proportion of anti-austerity women demonstrators is highest in Sweden (53.2%) and lowest in Spain (43.1%). The UK is similar to Spain, at 46.4%. It is important to note, also, that there is some variation in the proportion of women demonstrators within countries. In the UK it varies from 32.5% for Occupy through to 51% for the TUC's March for the Alternative. In Spain, the proportion of women at demonstrations surveyed ranges from 31.5% for Against the Europe of Capital, Crisis and War through to 52.3% for Real Democracy Now! Although the proportion of women at demonstrations in Sweden is markedly higher, it also varies. 44.2% of survey respondents were women at the May Day (SAP/LO) demonstration in Malmo, compared to 60.4% at the May Day Left Party event in Stockholm. Overall, the aggregated percentages of women demonstrators per country are broadly comparable to the ESS datahighest in Sweden (54.3%) and similar for the UK (49.3%) and Spain (48.6%). However, it is important to remember that the ESS data aggregates data across demonstrations on a range of issues.
In Table 3, we present our mixed effects regression model results. Model 1 shows a null model for demonstration effects. It finds that there are significant differences in the proportion of women at demonstrations across the 16 demonstrations in the sample. However, this effect is cancelled out when, in Model 2, we add the fixed effect of country. This, along with our descriptive statistics presented above, indicates to us that we can be fairly certain to have found country-level effects, despite the non-random sample of demonstrations analysed within each country and some intra-country variation. The positive and significant co-efficient in Model 2 for 'Sweden' indicates that anti-austerity demonstrations in Sweden have a higher proportion of female demonstrators compared to the UK and Spain. In Model 3, we add the individual level variables. We find that 'Sweden' continues to be a significant predictor even when we add a host of individual level variables. Although not significant, the co-efficient for Sweden is positive also in the ESS, which aggregates demonstrating on any issue (ESS model in Table 2). Thus, women are more likely to demonstrate in Sweden regardless of the demonstration issue, which provides further support for H1 (gender regime effects) rather than H2 (extent of cuts). This is especially noteworthy given that Sweden has the lowest proportion of antiausterity demonstrations in each of our study countries (18.5%, compared to 25.3% in the UK and 27.1% in Spain, POLCON 2019).
Our results also indicate that women anti-austerity demonstrators are more precarious: they tend to be younger and to not work full-time. They are more highly educated than their male counterparts, more likely to vote in a general election, to be asked by someone else to attend, to have a sense that their political participation can make a difference and to be left-wing. However, they state that they are less interested in politics than men, except for in the ESS data-set, where they are more interested. The contrary ESS data is likely an artefact of the manner in which the ESS aggregates demonstrations on different issues. We posit that demonstrations on issues that do not involve protecting one's self-interest are more likely to require political interest. Women demonstrators' relative lack of political interest might be compensated for by the fact that they are more likely to participate in anti-austerity demonstrations to protect their own interests compared to male demonstrators.
In the final model, Model 4, we include an interaction term between protesting for self-interest and country to test H5. Although the interaction term is not significant, it is important to note that it cancels out the positive significant coefficients for both country (Sweden) and self-interest as a motivation. This suggests that motivation to protest to protect self-interest is differentially distributed across the three countries. Indeed, in Spain, 59% of women demonstrators claim that they are very much motivated by self-interest, compared to 53% of men. In the UK, the proportion of men and women demonstrators who participate in an anti-austerity demonstration to protect their own interests is almost the same (32% for women, 30% for men). In Sweden, slightly more men than women Table 3. Mixed effects regression model results, predicting whether a demonstrator is a woman (n = 2529).  demonstrate to protect their own interests (23% women, 25% for men). The mean score for 'motivated to participate to protect own interests' for women antiausterity protesters in Spain is 4.5 (on a five-point scale, where 5 = very much), compared to 3.9 in the UK and 3.5 in Sweden. The mean scores for male demonstrators are 4.3 for Spain, 3.8 for UK and 3.4 for Sweden.

Discussion
Our results at the country-level appear to support the gender regime/empowerment hypothesis (H1), rather than the differential effects of public sector cuts hypothesis (H2). The gender regime/empowerment hypothesis (H1) suggested that:In a social democratic gender regime (Sweden) we find a higher proportion women in antiausterity demonstrations compared to neoliberal gender regimes (Spain and the UK).
The highest proportion of women in anti-austerity protests was found in Sweden, a social democratic public gender regime, which had the highest proportion of women in employment of our three country cases (75%), the highest proportion of female MPs of our three countries. H2, the differential effects of public sector cuts hypothesis led us to anticipate higher numbers of women protesters on anti-austerity protests in the neoliberal public gender regimes UK and Spain compared to elsewhere. However, to the contrary, we found that Sweden, which was less affected by public sector cuts compared to Spain and the UK, still had the highest number of women demonstrators. Spain had the lowest proportion of women demonstrators. In 2010, of our three countries the employment rate of women was lowest in Spain (at only 56.3%), but the unemployment rates for men (19.6%) and women (20.2%) in Spain hardly differed (Eurostat, 2010). Although public sector cuts were, arguably, most damaging to women in Spain, the proportion of women turning out to anti-austerity demonstrations was the lowest there. At the individual level, biographic availability seems important in determining the gender of anti-austerity protesters (H3). Women demonstrators were younger than their male counterparts and less likely to be in full time employment. This suggests that they may have had more time resources, but that they may also have been in part-time or precarious employment and thus were adversely affected by public sector cuts. They were also better educated than men and therefore we might say that they had greater knowledge resources. They claimed to be less politically interested than men, while at the same time they were more politically active in national elections and rate themselves as more left-wing on the left-right-self-placement scale. It might be that they perceived themselves to be less politically interested than they really are. Indeed, they hadcompared to mena relatively high sense that their individual actions would make a difference to public policy. Their relative lack of full-time work may suggest that they constitute a new wave of women's activism, demonstrating mobilisation of those affected by 'new social risks' (Bonoli, 2005), supporting the notion that women are a substantial part of the 'new losers' of globalisation (Della Porta, 2015).
The notion that women protested because they belong to the 'new social risk' groups and 'new losers' is further supported by our finding that demonstrating to protect selfinterests (H4) is a significant predictor of women's participation in anti-austerity demonstrations. This is particularly the case for Spain, where nearly all women on anti-austerity protests claimed to 'very strongly' agree with the statement that 'I participated in the demonstration in order to . . . defend my interests'. We find no support for H5 that country and self-interested motivations interact to predict participation of women. Selfinterest is similar for men and women in each of the countries, even though it is differentially distributed across countries when men and women protesters are aggregated.
Our analysis shows that there are important differences in the extent of women's participation in anti-austerity demonstrations within countries as much as across them. However, the demonstration level effects disappear as soon as we add 'country' (regime) to the models. This indicates that the type of public gender regimesocial democratic or neo-liberaland the extent of cuts matter. Nonetheless, it is worth mentioning that the calls for action of demonstrations may attract or repel women. The May 1 March, Left Party march in Stockholm, for example, explicitly included challenging patriarchy in its call for action and managed to attract demonstrators comprised approximately (given response rate bias) 59% of women. The Demonstration Against the Europe of Capital, Crisis and War in Barcelona which attracted proportionately fewer women was, in contrast, predominantly about work in a male dominated employment sector. Spain has historically had one of the lowest labour force participation rates for women in Europe (Lombardo, 2017). This suggests that country-level variables, gendered cultures of protest and demonstration contexts interact to determine whether women turn up to demonstrations.
A good turnout of women used to be rare in many trade union/political party organised demonstrations generally (Walgrave, Wouters, Van Laer, Verhulst, & Ketelaars, 2012) and in our sample, except for in Sweden. Indeed, female demonstrators were more likely to lack organisational embeddedness (Klandermans et al., 2014) and were more likely to protest alone (Wahlström and Wennerhag, 2014). Our results contrast with previous findings. The London-based Trade Union Congress organised demonstration returned the highest rates of participation from women of the three UK demonstrations analysed. This reflects the fact that women represent a growing proportion of the unionised workforce in the UK due to structural transformation and the unionisation of the public sector. At the same time, some Occupy camps have been characterised by 'gendered hierarchies, hostility to feminism and sexual violence' (Eschle, 2018, p. 534) which might explain the lower involvement of women in Occupy compared to the TUC demonstration.
Moreover, these are not the only forms of anti-austerity protests. Public sector cuts have coincided with a resurgence of feminism (Evans, 2015). In the UK, the Women's Budget Group analysed the gendered impact of the government budget and spending review in various reports. In November 2011, the Fawcett Society organised a demonstration against austerity measures, warning that they might reverse gender equality and 'turn back time' in the UK. Furthermore, austerity measures have undermined as well as inspired minority women's activism and protest in the UK (Bassel & Emejulu, 2017), including the formation of new organisations such as Sisters Uncut (Ishkanian and Saavedra, 2019). In Spain, women have been involved in anti-austerity protests in a range of contexts and organisations including political parties, trade unions, feminist organisations as well as the 15M or Indignados and the 22M or March for dignity (Lombardo, 2017;Palomo, 2016). Whereas women represent less than half of the participants in anti-austerity protests in the UK and Spain, their participation remains significant (over 40%). In Sweden, women represent more than half of the participants in anti-austerity protests. These differences might be related to variations in European's women's movements (Roth, 2017). Whereas in Sweden women are more fully integrated in traditional political institutions, historically the women's movements in Spain and the UK are more independent from the state. Thus, in addition to participating in mixed-sex anti-austerity protests, women in Spain and the UK mobilise separately (Evans, 2015;Fuentes, 2015). On 8 March 2018, the Spanish feminist movement organised a 24-hour labour, education, care and consumption women's strike (Campillo, 2019).

Conclusion
Our analysis makes an important contribution to research on gender differences in antiausterity demonstration participation post-Great Recession specifically and to the understanding of women in protests other than the women's movement. Furthermore, our paper is innovative in that it links scholarship on gender regimes and social policy with research on protest participation. Resources and experiences of grievances are shaped by gender regimes which provide access to decision-making and social support. We found that the proportion of female demonstrators was highest in Sweden, the country with most equal gender regime. It was lowest in Spain, the country hardest hit by the Great recession and austerity measures. Access to resources and entitlementsas experienced by women in Swedenencourages women's participation in anti-austerity demonstrations and defending their gains. In contrast, Spanish women are faced with the double burden of employment in precarious working conditions and the responsibility for reproductive labour (Lombardo, 2017). In addition, they participate in feminist protests (Fuentes, 2015). Perhaps it is unsurprising, therefore, that Spanish women are most likely to demonstrate to protect their own interests. In the UK, the Great recession has been instrumentalised to justify public sector cuts, but unemployment is lower than in Spain. In the UK, men's unemployment rate was higher than women's and women's political participation lower than in Sweden and Spain which appears to be linked to the lower participation of women in demonstrations. Our comparison of women's and men's participation in anti-austerity street demonstrations suggests that at the country level resources and thus gender regimes positively predict, and the extent of cuts negatively predicts, women's participation in anti-austerity demonstrations. At the individual level, grievances positively predict women's participation. Based on our analysis, we conclude that gender regimes matter with respect to demonstration participation of women and men. In addition to mixed-sex organizations, feminist mobilisation also needs to be taken into consideration when studying anti-austerity protests. Future analyses of cross-national protest survey data should consider the impact of the implementation of austerity measures from 2012 onwards. In addition, we call for cross-national comparisons of gender differences in involvement in right-wing populist anti-austerity movements.
Notes 1. We distinguish gender regimes from gendered welfare regimes. We are aware of Esping-Andersen's (1990) typology of three welfare regimes (conservative, liberal, social-democratic welfare states) which, due to its male bias, has been criticised by feminist scholars (e.g. Leitner,