Radical Social Movements in Western Europe: A Configurational Analysis

There has been little comparative research on the differences across radical social movements in the context of consolidated democracies. This paper analyses the squatting movement, as an exemplary case of contemporary radical movement. This study aims to identify the causal contexts that explain the differences of strengths within these movements across 52 large cities in Western Europe. It examines three main hypotheses drawn from the literature on social movements concerning the characteristics of political systems, the availability of resources and the presence of economic grievances. We use fuzzy sets qualitative comparative analysis (fsQCA) to identify configurations of causal conditions. The findings show that diverse contexts (multi-causation) lead to strong movements. A first causal context combines grievances, resources and closed or unresponsive institutions, and is typically found in Southern European cities. A second context highlights the presence of robust far-right parties in combination with less severe grievances and relative scarcity of resources, and is typically found in Northern European cities. These findings demonstrate that resources and grievances are quasi-necessary conditions for strong radical movements, although polarization can lead to a similar outcome where these characteristics are not present.

social centres' movement. Despite its nearly 40-year presence in European society, there has been little comparative investigation into this movement. In particular, the paper examines the causal contexts related to their differing strengths across cities. Why are squatters stronger in some cities, and weaker in others? How can contexts explain these differences?
Literature on contentious politics has long discussed whether social movements are explained by grievances and economic hardship, by resources both internal and external to movements, or by the characteristics of the political system. I examine the influence of conditions relating to each of these dimensions (grievances, resources, political system) to identify the contextual characteristics that explain differences in the strength of squatters' movements across cities, i.e. the capacity to reproduce, grow and engage in a broad range of contentious and non-contentious actions.
Employing fuzzy sets qualitative comparative analysis (fsQCA), I analyse an original data set of 52 Western European cities. This method identifies configurations of causal conditions, i.e. patterned combinations of causes leading to a certain outcome. The findings confirm that grievances do not explain the movement's strength by themselves. Second, they demonstrate that political systems mediate the effect of socio-economic factors. Indeed, the squatting movement is stronger where hardship combines with political resources or polarization. Third, political polarization can lead to strong movements, even in the presence of lower levels of hardship. Two broad patterns are recognized. In the first one, grievances and resources combine with closed or unresponsive institutions. This pattern is typically found (albeit not exclusively) in Southern European cities. In the second one, less severe grievances or the scarcity of resources combine with the presence of robust far-right parties. This pattern is typically found in Northern European cities. These results call attention to diverse pathways leading to RSMs, emphasizing the role of resources, grievances or polarization as quasi-necessary conditions for strong radical movements.

Squatters in Europe
The squatters' movement is the counter-cultural, leftist movement that emerged in Europe during the second half of the 1970s, through the convergence of three separate movements: the housing movement, the youth movement and the counter-cultural trends of the period (Provos, Punks, etc.). Squatters occupy and restore empty buildings to establish self-managed social and cultural centres, and/or set up residence. In the 1970s, several cities encountered conflicts surrounding the opening of these centres in disused buildings. At the time, a housing shortage was accompanied by a lack of venues for the young to carry out their individual and collective projects. A large number of vacant places -housing, industrial or commercial spots -allowed the first illegal occupations. The local administration's initial disposition to negotiation made it viable for squatters to settle down. The main participants in these struggles were the radicalized youth. Leftist organizations -especially those at the left of the main party of the left -were at the foundation of the movement, and have subsequently found a space for engaging in radical politics within this movement (Katsiaficas, 2006;Wright, 2002). The New-left was supportive of their demands (e.g. the Greens in Germany). Thus, social centres have traditionally represented a platform of the extra parliamentary left, as well as rebellious, political movements with diverse ideological affiliations. Indeed, one can find Marxists, anarchists and autonomist social centres. Nonetheless, there are others who tend to be less committed to such ideologies, despite sharing a leftist, libertarian political background. Today, the squatters' movement is engaged in a wide variety of campaigns, for affordable housing and minorities' rights, and against war, neo-Nazi groups, unemployment and precariousness, urban speculation, regeneration projects, gentrification and displacement, among other topics. Diffusion has spread this trend, with the ensign that traditionally designates a squat -an inverted black lightning over a circle -seen across various countries around the world.
This paper focuses on the political movement of self-managed social centres, and specifically on those groups that share symbols (flag), codes and recognize themselves as part of the movement. There are squatters' spin-offs and other movements in which squatters participate (e.g. housing rights' movement, asylum seekers' movement, migrants' rights campaigns) that have configured new, different phenomena, but they are not analysed in this paper. Furthermore, individuals or groups might illegally occupy buildings, but providing that they do not do so as means of protest and rather as survival strategy, they cannot be assigned to this category.

Defining Radical Social Movements
Major topics of research in the social sciences during the twentieth century included the radicalization of the working class (Calhoun, 1982;Gallie, 1983;Lipset, 1983), the radicalization of the middle class (Bagguley, 1995;Parkin, 1968) and the emergence of the 'new social movements' (Kriesi, Koopmans, Duyvendak, & Giugni, 1995). However, the study of radical movements lost its currency after the collapse of the socialist countries in Eastern Europe (1991). Subsequently, authors' interest has migrated to areas such as the emergence of far-right groups and parties (Klandermans & Mayer, 2006;Koopmans, 1997;Mudde, 2013), as well as the study of political violence and terrorism (Della Porta, 2006;Goodwin, 2012;Tilly, 2003).
Radical movements have been implicitly defined in opposition to mainstream or moderate groups. Pizzorno (1978) drew the most influential distinction so far, stating that instrumental movements establish a separation between means and ends, while expressive or counter-cultural movements see collective action as an end in itself. Della Porta and Rucht (1995) proposed the category of 'families of movements' -alike to the idea of 'movements industries' (McCarthy & Zald, 1977) -to distinguish among the variety of groups that emerged from the struggles of the cycle started in 1968. Fitzgerald and Rodgers (2000) acknowledged that RSMs are different to other movements in a range of features such as structure, ideology, tactics, methods of communication and measures of success. In general, contentious politics literature frequently considers RSMs as either byproducts of cycles of protest (Della Porta, 2006;Koopmans, 1995;Kriesi et al., 1995) or consequences of social dynamics that push groups into processes of radicalization (Portes, 1971;Stekelenburg & Klandermans, 2010).
Any definition of what radical movements are is essentially relational. Something (e.g. policies, groups) is 'radical' in respect to something else that is regarded as moderate. Thus, radicalness depends on contexts and periods. Therefore, to define the concept in a useful manner for empirical research, we should (1) establish historical and geographical parameters, against which similar instances of this type can be reasonably compared and (2) define the components that constitute the concept. I suggest that RSMs can be characterized according to three components, which together illustrate their differences relative to other groups: (1) agenda, (2) repertory of contention and (3) identity. Consequently, RSMs (1) pursue an agenda of drastic changes that concerns a broad range of issues, especially the political and economic organization of society, whose implementation would affect elite interests and social positions. In order to implement their agenda, they (2) perform a repertory of contention characterized by the employment of unconventional means, specifically civil disobedience. In addition, these groups adopt (3) counter-cultural identities that frame and justify unconventional objectives and methods, although this identity might not be present at early stages. On occasions, such an identity is the outcome of contentious interactions between these groups, authorities and opponents/counter-movements.
The definition of RSMs proposed here approximates Della Porta and Rucht's (1995) notion of 'left libertarian movements' . However, this concept is an ad hoc construct aiming to distinguish the radical groups that emerged after 1968 from their predecessors. Instead, by paying attention to agenda, repertory of contention and identity, we can include within the same 'family' -in this case, RSMs -groups that emerged from different cycles of protest, with diverse organizational forms (vertical, horizontal, etc.), and diverse ideological backgrounds. Our definition offers the advantage of permitting the allocation of different historical generations of groups to the same cluster. For example, traditional anarchist factions and new autonomist or alter-globalization groups can be sorted together despite their differences in organization and ideology, given that they share a similar agenda, repertory and identity. These similarities lead them to participate in similar campaigns of protest, and converge upon the same enterprises; for example the opening of self-managed social centres.

Explaining Squatters' Strength
This study focuses on the characteristics of contexts that might foster or prevent strong squatting movements. A 'non-intentionalist' approach is adopted, whereby contexts are assumed as patterned relationships beyond the manipulative control of any single group or individual (Skocpol, 2007, p. 200). By definition, contexts are multidimensional, i.e. they consist of diverse conditions acting together. In this section, I describe the specific conditions that are expected to explain the differences in squatters' strength.
The departure point originates in the observation that the squatters' movement has traditionally attempted to establish solid roots within cities. Squatters have framed their identity in terms of the trajectory of local struggles. As stated by Le Galès (2002) and Brenner (1999), among other authors, cities are local societies with their own conflicts and dynamics. Calhoun (1982) showed how the radicalism of the nineteenth century in Europe mainly consisted of populist movements based on locality and communitarian ties. Therefore, cities are proper contexts in which to understand differences between squatters' movements. Changes at the local level (e.g. policies, institutions) have the potential to mobilize categories created in and by the urban fabric (Nicholls, 2008). However, as these changes do not occur coherently or simultaneously across locations or scales, their effects on the local conflict vary significantly (Brenner, Peck, & Theodore, 2010;Künkel & Mayer, 2011).
The social movement literature has paid attention to three major dimensions in explaining differences in the trajectory and dynamic of contentious groups. Grievances are the motives of collective action, sentiments of injustices stemming from wrongdoings attributed to authorities or third parties. Resources are those social or communitarian assets that facilitate the coordination of people in campaigns aimed at changing the state of affairs. Political institutions, in turn, define those allowed to decide on public issues and shape the options of challenging groups to voice their claims.

Grievances
The first attempts to explain the rise of collective action focused on relative deprivation and hardship (Gurr, 1970). Feelings of injustice arise particularly during rapid social change (Inglehart & Welzel, 2005), or when people compare their own situation with that of the upper classes or with the promises subscribed by their rulers (Walker & Smith, 2002). Previous studies on squatters suggest that this movement is related to the dwindling supply of affordable housing and the worsening prospects of the young in the labour market, which began with the neoliberal reforms in the 1970s and 1980s. However, these studies have not examined how differences in housing policies or unemployment levels across cities might affect the strength of these movements.

Resources
The resource mobilization approach maintains that ordinary people need resources to sustain collective action over the long run (Dalton & van Sickle, 2005;McCarthy & Zald, 2001). Resources include the entrepreneurial core leading protest campaigns (Jenkins 1983), as well as social capital (Minkoff, 1997), community infrastructures (Almeida, 2012), churches (Brown & Brown, 2003), unions (Gallie, 1983), lifeworlds (Edwards, 2008) and cultures of solidarity (Fantasia, 1989). These components configure a social basis by which people, through mechanisms of socialization, engage in contentious politics (Skocpol, 2007;Tilly, 1978). In this paper I focus on leftist communities, i.e. groups of persons identified with left-wing ideologies. These can be conceived as sources of supply for radical movements for a number of reasons. Large leftist communities increase the chances of leaders and activists to find like-minded persons willing to engage in squatting or other radical movements, which in turn increases the chances of further imitation. Besides, where the leftist communities are larger, the public opinion can be more tolerant of these modes of contestation. Therefore, leftist communities are conceived as crucial resources for squatters' movements.

Institutions
Comparative studies on social movements show that the political system's degree of openness to challenging groups affects the options of RSMs. Important studies have observed that strong centralized states, limited openness and a lack of directdemocracy institutions foster expressive or radical movements (Kriesi & Wisler, 1996;Kriesi et al., 1995;Wisler & Giugni, 1996). These studies suggest that closed political systems encourage RSMs. Moreover, RSMs are those most prepared to protest in closed settings, because they are not easily discouraged by repression, or the lack of response from authorities. Accordingly, one could expect that the more closed the political system, the stronger the squatters' movement will be. In a similar vein, early contributions of the political process theory showed that radical protest was related to authorities' responsiveness to social demands. Eisinger (1973) found that very low levels of responsiveness (i.e. institutional capacity to meet social demands) discouraged mobilization, creating a perception of protest's futility. Similarly, very high levels of responsiveness deactivated the motives for mobilization. In this study, and similar to the effects of closed institutions, I suggest that lower levels of responsiveness are related to radical responses.
The political process theory has shown that social movements interpret their environment in terms of opportunities or threats based on identities and values (e.g. Koopmans, 1995;Steinberg & Ewick, 2013). Activists do not decide in a cultural vacuum, and their decisions do not only proceed from rational calculations of costs and benefits. Tilly (2008) has shown that how people protest, the repertories of contention, are determined by culture and history. Since its origins, anti-fascism has been a major feature of squatters' identity. The radical left is often very sensitive to developments in the extreme right, and resistance to these groups is a central dimension of their agenda, as demonstrated by frequent clashes between far-right groups and radical left activists in several European cities (e.g. Athens, Berlin). Empirical research has established that threats have a mobilizing effect on social movements (Almeida, 2003;Fox & Squires, 2001;van Dyke & Soule, 2002). Moreover, relevant far-right parties are evidence of elite polarization, and might have radicalisation effects in society (Halmai, 2011). Therefore, I suggest that squatters are stronger where far-right parties are stronger.

Data and Methods
Originally proposed by Ragin (1989), QCA is a method aimed at comparing macro-level data, such as qualitative, historical and institutional characteristics (Schneider & Wagemann, 2006). Recent studies on protest movements have used this method (Cebotari & Vink, 2013;Hussain & Howard, 2013;Ragin & Alexandrovna-Sedziaka, 2013). QCA offers some advantages over statistics in terms of dealing with the research problem. While statistics rely on a probabilistic and correlational logic, QCA employs set theoretical thinking to relate the explanandum to the explanant. This method uncovers the extent to which causing conditions are a subset or superset of the dependent variable. QCA is configurational in that it shows how the effect of one variable is contingent on values of other relevant variables. Thus, QCA visualizes combinations of conditions leading to certain outcomes, rather than independent effects of single variables (Ragin, 2008a). QCA is further suitable for this research in that it allows the existence of multiple causes for a given dependent variable.
QCA causal logic is based on the identification of sufficient and necessary conditions related to a given outcome. Thus, 'given some plausible theoretical arguments, a condition can be considered sufficient if, whenever it is present across cases, the outcome is also present in these cases'. A condition will be necessary if, 'whenever the outcome Y is present, the condition is also present' (Schneider & Wagemann, 2012, pp. 57, 69).
To identify necessary and sufficient conditions, a truth table is created and subsequently minimized. Unlike the traditional data matrix, the truth table shows all the logically possible combinations of conditions, indicating whether they are represented by the cases of the sample. There will be 2 X logically possible combinations, where X is the number of conditions (or independent variables). Thus, rows are not individual cases but rather combinations of conditions.
In QCA, independent variables have to be calibrated, i.e. transformed to depict the membership of cases in sets. For example, a country with an unemployment rate above a certain threshold determined by the researcher will belong to the set 'countries with high unemployment'. I employ fsQCA, which allows for the existence of variables depicting degrees of membership (Rihoux & Ragin, 2009). This attribute renders fsQCA a suitable alternative when the explanatory variables are ordinal, ratio and/or continuous indicators. Accordingly, variables have to be calibrated within a scale from 0 to 1 of intensity of their membership in all the conditions of the model. To calibrate a variable, three points of membership have to be indicated: (1) the threshold of full membership in the set, which is represented in the 0-1 scale by the score 0.95; (2) the threshold of full exclusion of the set, represented by the score 0.05; and (3) the point of maximum ambiguity, when cases are not clearly in or out of the set, represented by the score 0.50.

Dependent Variable
I define a squatting movement's strength as the capacity to reproduce itself and/or increase its constituency by engaging in an array of actions: squat or occupy venues, run social centres, organize and plan activities (artistic, political, educational, etc.), protest against something or someone, draw attention to their ideas, engage in campaigns launched by other actors, publish outlets and websites, collaborate in parties, political meetings, happenings, etc. Interestingly, this definition is compatible with the capabilities approach (e.g. Nussbaum & Sen, 1993).
In this study, I adopt a mixed, quali-quantitative approach due to the nature of the movement at hand and the limitations inherent to a comparative study with a significant number of cases. The approach adopted here triangulates data from different sources to avoid their limitations in the estimation of the dependent variable.
Considering potential selection biases of newspapers (they tend to overlook countercultural movements, and only pay attention to them when they are involved in negative situations), as well as the limitations of other indicators, such as the number of illegal social centres or squatter activists, I decided to estimate the dependent variable with the Delphi method (Adler & Ziglio, 1996). This method is used when there is a lack of reliable information from secondary sources (data sets, statistical offices, etc.). It has been applied in sociology (Sheptycki, 2003), policy analysis (Prante & Bohara, 2008) and management studies (Davenport, 2000). Like other qualitative methods, Delphi does not require a large number of respondents, because its estimations are not averages but consensual statements based on arguments. Yet, it relies on informed respondents, selected because of their outstanding knowledge on the topic. The process of repeated consultation allows to crossvalidate the individual knowledge of each experts.
I contacted a group of 12 key informants, based on their academic background and experience as (former or current) activists or collaborators. Informants hold university positions and have published academic papers on urban protests and/or squatters. The objective was to systematize the common knowledge of researchers and activists into a formal and comparable indicator that visualizes the differences of strength of each movement relative to other locations in a given period, thus generating a categorical, relational indicator. Since I did not hold the objective of generating an absolute, ordinal or scale indicator, I could dispense with the exact count of protest events, squatted venues or activists. Key informants were submitted a brief questionnaire and asked to rate the strength of the squatters' movement in the 82 largest cities of the EU-15 plus Norway andSwitzerland (2001 -2007), according to the following categories: strong, moderately strong, weak and non-existent squatters' movement.
After three rounds of consultation, where they could argue, comment or contend the answers of other participants (under anonymity), a consensus on the categorization of 60 cities was reached. The final list of 52 cities resulted from the exclusion of cases with missing data in the independent variables (see Table 1). In summary, 23% of the cities (12 cases) were regarded as having a non-existent movement (category 1), 50% (26 cases) weak movements (category 2), 21.2% (11 cases) moderately strong movements (category 3) and 5.8% strong movements (category 4).
We conducted a reliability test to corroborate the key informants' opinions. This test allowed to check for potential biases and assessment errors. Drawing on Silver (2003), I compared data from diverse locations from only two quality newspapers. Using the Lexis-Nexis database, I conducted a search for the terms 'squatter ' and'social centre' in The Guardian, and'okupa' in El País (2000 -2008), two quality newspapers with significant European coverage. These terms allowed us to count contentious events related to the squatting movement, including protests, occupations/evictions of squatted venues and activities usually performed by social centres, ranging from art exhibitions to public debates. Overall, the test showed that the probability of a city considered by the group of key informants as having strong and moderately strong movements reporting contentious events is substantially higher (3.29) than the same probability in the group of cities regarded as having a weak and non-existent movement (0.06). Therefore, the reliability test offers reasonable grounds to validate the Delphi classification.

Independent Variables
The six factors discussed in the theoretical framework are incorporated as conditions of the model: youth unemployment (jobless); housing stress (owners); leftist communities (leftcommunity); far-right relevance (fright); openness of the political system (openness); and responsiveness of local institutions (responsive). Table 2 offers detailed information on these indicators and their calibration, including the dependent variable. Table A1 shows the data set with original values and their corresponding fuzzy scores.

Results
The software fs/QCA 2.5 was employed. The analysis of necessary conditions for the outcome 'strong squatters' movements' shows that none of the conditions alone are necessary for strong movements, given that their consistency scores are below 0.90 (detailed information in Table A2). The analysis of sufficiency first presents the intermediate solution.
In order to minimize complexity (i.e. reduce the rows in the truth table to their minimum logical components), I incorporate the logical remainders (combinations of conditions with no empirical cases) that are coherent with the theoretical framework (the so-called easy counter-factuals), 1 while the necessary conditions are not removed. Thus, the intermediate solution is considered the most cautious approach (Rihoux & Ragin, 2009). Consistency and coverage measures are included. Consistency indicates the extent to which cases with higher scores in the dependent variable are include in the solution term (or set of terms), with a consistency score of 1 indicating a perfect intersection between the sets formed by the conditions and the outcome. Coverage measures the proportion of memberships in the outcome that is explained by the combinations (Ragin, 2008b). Table A3 shows the membership scores of each city for all combinations resulting from the minimization process. Table 3 summarizes the results, showing that five combinations are conducive to strong movements. The overall consistency of this solution is high (0.77), while the solution coverage is satisfying (0.67). The rather low coverage of each combination suggests that there are different contexts related to strong squatters (no combination explains the majority of the cases). Twenty-four cities are included in the combinations, including 12 cases with high scores in the outcome and 12 cases with moderately low scores in the outcome (i.e. cities with rather weak movements).
The intermediate solution shows that no single variable is sufficient for the outcome. Instead, there are five combinations of characteristics leading to strong squatters'  (3); weak (2); and non-existent (1). To enable a clear-cut distinction between cities with strong and moderately strong movements, and those with weak and non-existent movements, the Delphi scores 3, 2.1 and 1 were assigned to the crossover scale points 0.95, 0.50 and 0.05, respectively. Thus, cities regarded as having strong or moderately strong movement have fuzzy scores of 1 and 0.95, respectively. Cities regarded as having weak and non-existent movements have fuzzy scores of 0.43 and 0.05, respectively. Youth unemployment (JOBLESS) Source: Urban Audit, Eurostat. An empirical criterion is utilized to calibrate this variable. The OECD assesses countries according to their position in respect to the youth unemployment rate average (OECD employment outlook, various years). Thus, before the financial crisis, countries typically regarded as having a high youth unemployment rate included Greece, Spain and Italy, whose rate climbed above 20%. Conversely, countries with low youth unemployment included countries such as Denmark and the Netherlands, whose rates were below 10% in the same period. Therefore, the threshold of full membership (fuzzy score 0.95) to the group of cities with high youth unemployment was set at 20%. The threshold of full exclusion (fuzzy score 0.05) was set at 10%. The maximum ambiguity point was set at 15%. Housing stress (OWNERS) Source: Urban Audit, Eurostat. This condition is based on the percentage of households living in owned dwellings (Eurostat). Given the lack of comparable indicators on other sources of housing stress, this indicator is a proxy assuming that higher levels of home ownership are related to higher levels of housing stress. Accordingly, cities with a share above 60% of owner occupation (fuzzy score 0.95) are considered as being fully in the group of cities with high home ownership. Cities with shares below 40% are considered fully out of that group (fuzzy score 0.05). The threshold of maximum ambiguity is set at 50%. The size of leftist communities was measured by utilizing survey data. We calculated the ratio of individuals who, at the city or regional level (NUTS 2), positioned themselves on the values 9 and 10 in the left-right axis with respect to those located between the values 3 -8. Having observed the normal distribution of these scores through a Kolmogorov-Smirnov test, I set the average (ratio 0.10) as the maximum ambiguity point. By adding one standard deviation to the average, I set the threshold of full membership to the group of cities with large leftist communities (0.15). The threshold of full exclusion from this set is obtained by subtracting one standard deviation from the average (0.05). Far-right relevance (FRIGHT) Sources: Municipality (websites); Wikipedia. An empirical criterion is utilized to calibrate this variable. This indicator measures the average electoral share of far-right parties during the last three city elections held up to 2008. Political parties' classification in the left-right axis draws on Armingeon, Gerber, Leimgruber, and Beyeler (2008) and Ignazi (2006). As numerous political systems have established a 5% threshold of the vote to obtain a parliamentary seat, this share was set as the maximum ambiguity threshold. By observing cases of far-right parties' success before the Euro-crisis, I established 7% as the point of full membership (fuzzy score 0.95) in the group of cities with strong far-right parties. A share of 3% was established as the threshold of full exclusion (fuzzy score 0.05) in the set of cities with strong far-right parties.

Openness of political systems (OPENNESS)
Sources: Type of electoral system according to Van der Kolk 2007; rate of local government representatives per 1000 inhabitants (Urban Audit, Eurostat). The indicator is a factor created through principal component analysis. The factor created explains 61% of the variance (Eigenvalue ¼ 1.109). High scores in the factor indicate a higher degree of openness (i.e. proportional systems, higher number of local elected representatives). Lower scores indicate lower degree of openness (i.e. majoritarian systems, lower number of representatives). A Kolmogorov -Smirnov test demonstrates that these scores do not exhibit a normal distribution. Therefore, scores were scrutinized and two breaking points were observed. The upper score (0.4) was established as the threshold of full membership to the set of cities with open political systems. The median of the lower split between scores (0.04) was set as the threshold of full exclusion of this set (below this score, we can speak of cities having closed political systems). The maximum ambiguity threshold was established at the lower end of the cluster with the larger number of cases (0.375).

Responsiveness (RESPONSIVE)
Source: spending per capita of the local administration (Urban Audit, Eurostat); World Bank's indicator of effectiveness of the national government (Kaufmann, Kraay, & Mastruzzi, 2009). This condition is measured by means of an indicator created through principal component analysis. The factor created explains 69% of the variance (Eigenvalue ¼ 1.381). Higher scores point at greater effectiveness of national institutions and higher levels of local governments' spending. Normal distribution was observed (Kolmogorov-Smirnov test). The threshold of maximum ambiguity was set at the average, while the thresholds of full membership and full exclusion were established by adding and subtracting one standard deviation from the average, respectively. movements: low levels of unemployment rates and relevant far-right parties (f1); relevant far-right parties, low levels of housing stress,and little leftist communities (f2); high levels of youth unemployment, significant leftist communities and closed local political systems (f3); relevant far-right parties, significant leftist communities and high levels of housing stress (f4); and high levels of youth unemployment and housing stress, low levels of responsiveness of local institutions and significant leftist communities (f5). While the first two combination explains cities predominantly located in Northern Europe (except for Milan, Bologna and Lyon), the latter three combinations account mainly for cities of Mediterranean countries (except for Berlin and London). Some cities appear in more than one combination. Interestingly, Milan and Lyon are Mediterranean cities with strong movements included in combinations (f1) that do not include other cities from Mediterranean countries. In combinations f3 and f5, the conjoint effect of significant leftist communities and high youth unemployment appears as a necessary sequence for the outcome. However, this pair also interacts with other characteristics: political systems either closed or unresponsive, or high rates of home ownership, or relevant far-right parties. Overall, combinations f3 to f5 show that significant leftist communities explain strong movements in the context of severe grievances (youth unemployment, housing stress) and local institutions being unresponsive or closed political systems. It must be noticed that the cities explained by combinations f1 and f2 are not the same as those explained by combinations f3, f4 and f5, except for Milan. Likewise, London and Berlin are included in patterns that explain mostly Southern European cities.
The presence of relevant far-right parties appears as an INUS condition in three combinations, i.e. as a condition that is insufficient in itself but nonetheless a necessary part of an unnecessary yet sufficient combination of conditions. Similarly, the presence of significant leftist communities is also an INUS condition in three combinations, while the presence of grievances (high youth unemployment, high levels of housing stress) is an INUS condition in the same three combinations.
The consideration of the parsimonious solution offers further insight into the data set. Here, all logical remainders are used as if they produce the outcome, to minimize the truth table. This solution shows four combinations, as indicated in Table 4. The coverage (0.65) and consistency levels (0.77) are satisfying. The cities explained in this solution are the same as those included in the intermediate solution. The first three combinations are also identical, while the last combination slightly differs from the last two combinations of the intermediate solution.
Overall, the parsimonious solution allows a further simplification of its results.
Combinations f6 and f7 can be merged into a new pattern, FRIGHT*[jobless þ (owners*leftcommunity)], and interpreted as follows: relevant far-right parties combined with low youth unemployment or the absence of leftist communities and low levels of housing stress explains strong movements. Cities included in this combination belong mainly to Northern European countries. This simplified combination indicates that, within this pattern, the presence of relevant far-right parties is a necessary condition for strong squatters' movements.
Combinations f8 and f9, in turn, can be merged into the pattern LEFTCOMMU-NITY*JOBLESS*(openness þ OWNERS), and interpreted as follows: significant leftist communities and high youth unemployment leads to strong movements when combined with either closed political systems or high shares of home ownership. This new combination explains strong movements in Southern European cities, plus London and Berlin. Likewise, it indicates that the combined presence of significant leftist communities and high rates of youth unemployment are necessary conditions for strong squatters' movements. The simplification of the four combinations of the parsimonious solution into two simplified explanatory patterns lends ground to the identification of two major contexts leading to strong squatters' movements. The main disadvantage of both the intermediate and parsimonious solution is that 10 cities with rather weak movements were nonetheless included. These cases have the conditions to develop strong movements, but nevertheless they do not do it. Case studies and further research should account for the factors that explain this. Nevertheless, no flagrant contradiction emerged from the analysis performed (i.e. no city with non-existent movements was included in the aforementioned combinations), and the coverage and consistency scores validate the whole procedure.
A final examination of the data set consists of the analysis for the absence of the outcome, i.e. the combinations of conditions that explain weak or non-existent movements. The intermediate solution results in two combinations, with high general consistency (0.79) and a satisfying coverage (0.68), as shown in Table 5. Two combinations lead to weak or non-existent movements: lack of relevant extreme right parties and low levels of youth unemployment (f10); and high levels of responsiveness, lack of significant leftist  communities, and high youth unemployment (f11). They explain 23 cities with weak and non-existent movements, and no contradiction is included.

Discussion
The analysis indicates that two causal contexts are particularly favourable for the development of strong squatters' movements. In the first context, grievances are less prominent, as the levels of youth unemployment and housing stress are low. Resources are also limited in this causal context, as left communities are small. Instead, characteristics of the local political system such as the polarization effect generated by relevant far-right parties appear as a crucial cause for strong squatters' movements. Relevant far-right parties are an incentive for political action, especially among those more sensitive to the developments of these groups. This causal pattern accounts mainly for Northern European cities, as well as Lyon and Milan, although this latter city can also be explained by other causal patterns.
In the second context, I observe the simultaneous presence of grievances, resources and closed or unresponsive political systems. While far-right parties are not a necessary condition in this context, as we have seen in the results, they might be present in certain cases (see combination f4). In this causal pattern, significant left communities provide the resources required for strong movements; high levels of youth unemployment or high levels of housing stress generate a context of grievances that particularly affect the youth; and closed local political systems and unresponsive institutions enhance the options for radical groups mobilization (or at least do not inhibit the mobilization of radical movements). This context is typically found in Southern Europe, as well as main European capitals such as London, Berlin and Paris. The presence of large communist parties, various left-wing movements and radical unions during most of the twentieth century are the historical basis of the pervasiveness of significant leftist communities in this causal context.
Similarly, I have observed that low levels of grievances and polarization prevent strong movements to appear. Moreover, lack of resources and high levels of responsiveness prevent strong movements even if unemployment is high.
Overall, the findings contribute to debates on radical movements by emphasizing the relevance of the interplay between resources, institutions and grievances. Although a strong base of leftist communities or political cultures appears as an important asset for radical movements, they can prosper in contexts where this social base is more limited. As radical movements often consist of relatively small groups, they do not require large numbers of activists to exist. The ability of these groups to persist and reproduce requires abilities at the organizational and micro-level (Corrigall- Brown, 2011;Fine & Harrington, 2004), including recruiting like-minded persons (McAdam, 1986, Schussman andSoule, 2005), creating commitment and communities (Kanter, 1972;Nepstad, 2004), and shaping and updating identities (Polletta & Jasper, 2001;Valocchi, 2009). Given the scope and purpose of this study, it cannot dwell in detail on these aspects. However, they cannot be neglected. I assume that where resources are not particularly generous, groups must place special emphasis on commitment, identities or ideology to sustain themselves over the long run. The perception of threat can stress engagement and participation even if resources are not generous. The parties of the extreme right often polarize local societies, providing incentives for collective action, especially among persons who perceive a threat in extreme right discourses, as observed in Dutch cities (Mollenkopf, 2009). Indeed, such persons are more likely to engage in radical activism. However, the relationship between far-right and radical left could be more than reactive. The recent shift towards a social movement perspective to study the far-right suggests that they could share some causes (Blee & Creasap, 2010;Klandermans & Mayer, 2006). After all, far-right groups also need favourable contexts. The strong presence of nationalists or neo-Nazi activists in the subcultural scene, including music bands, alternative media, football fans, circuits of bars and alternative shops, demonstrates that these are effective mechanisms for recruiting membership and updating old identities. These far-right subcultures can be easily observed in countries that experienced Nazi-fascist regimes in the first half of the twentieth century. Indeed, fascist social centres have appeared in Italy in recent years (Di Tullio, 2006).
While less open and less responsive institutions are more common in Southern European cities with strong movements, considerable degrees of openness and responsiveness are crucial characteristics of Northern European cities that prevented strong movements during the period studied. However, previous research has shown that local governments of relatively wealthy cities ruled by local authorities more prone to negotiate with squatters actually faced intense squatters' protest (Bieri, 2002;Mikkelsen & Karpantschof, 2001;Pruijt, 2003;Uitermark, 2004). This confirms that squatting is not only a matter of unsatisfied needs but (or rather) political (counter-)cultures seeking a way of expression. Other authors have noticed that the political openings and the resources provided by a vibrant infrastructure of progressive alternative projects (Koopmans, 1995;Mayer, 1993) were crucial aspects that favoured the emergence of this movement in cities such as Milan, Berlin or Amsterdam in the 1970s. In turn, this facilitated movements' diffusion across several locations since the 1970s and throughout the 1980s (Owens, 2009;Squatting Europe Kollective, 2012). Finally, the findings suggest that vigorous radical movements are more likely in large, or capital cities, even in contexts of moderate national political cultures. Indeed, London, Berlin and to a lesser extent Copenhagen have significant bases of resources provided by large leftist communities (as indicated by their high fuzzy scores in the condition 'leftcommunity'). Thus, larger cities are more likely to detach themselves from national contexts.
This study has limited its scope to Western Europe and advanced economies, focusing on the period prior to the financial crisis. While the implementation of measures of fiscal austerity and the rapid deterioration of social indicators in several countries since 2008 have affected social movements, this study cannot assess such effects. However, we have observed squatters playing important roles in the demonstrations against austerity measures in countries such as Greece and Spain, and in the 'occupy' movement during 2011 (Psimitis, 2011;Romanos, 2013). Radical movements usually benefit from, and try to take advantage of, waves of social conflict (Della Porta, 1996;Tarrow, 1989), as popular mobilization offers visibility and opportunities for activism or recruiting new membership.

Conclusions
This paper has analysed differences in the strength of the squatters' movement across Western Europe, examining major assumptions of contentious politics literatures. The paper has offered an innovative manner to assess squatters' strength, triangulating data from different sources. This indicator, which visualizes the differences of the strength of the movement relative to other locations, allows a systematic comparison of the contexts related to these strengths. The paper has contributed to shed light on the dynamics of RSMs, a topic often sidelined by literature on contentious politics. This paper suggests that the interplay of grievances, resources and political institutions explains the variation of strength of squatters' movements, identifying two major contexts. In the first context, relevant far-right parties appear as a necessary condition for strong movements. Polarized local political systems lead to strong movements despite rather low level of grievances or lack of resources. This context explains strong movements of Northern European cities. In the second context, availability of resources and the presence of grievances appear as a necessary combination of conditions leading to strong movements. However, this couple interact with closed or unresponsive local institutions, or relevant farrights parties. The second context explains strong movements located in Southern European cities and other major European capitals. Overall, the analysis has shown that the socioeconomic characteristics of the local society must be joined by vigorous leftist communities and/or relevant far-right parties to generate strong squatters' movements. This confirms that the effect of grievances on RSMs is mediated by institutions and political resources.
To the best of my knowledge, this work represents the first extensive comparative research on this movement, and is the first study to systematically test competing hypotheses. It has grounded its observations in relevant theories of political change, offering fresh insights into the development and opportunities of RSMs. Further comparative research on the topic should examine variables that were not tested here (e.g. repression, city size), and combine contextual dimensions with longitudinal (remote and proximate) and internal characteristics of these movements. Appendix.