Fighting for Their Neighborhood: Urban Policy and Anti-State Riots in France

Abstract:Why does anti-state, violent rioting take place in advanced democracies? The paper investigates the role of the urban environment in shaping grievances expressed and mobilization/counter-mobilization processes observed during a riotous episode. In particular, I look at large social-housing estates as a propitious urban setting for the eruption and sustenance of anti-state violence. I identify three mechanisms (stigma amplification and inversion, failure of state intervention in the form of everyday administration and emergency policing, and advantages for network activation and resource mobilization among potential rioters) that complement standard explanations of rioting based on socioeconomic and ethno-cultural grievances. I test the theoretical model using a controlled case study of two neighboring suburbs in the North of Paris, with similar socioeconomic, demographic, and political characteristics but different violent outcomes in the 2005 nation-wide wave of French riots. The paper traces the source of local variation to the exogenous presence of large, concentrated social-housing estates in one, but not the other. The analysis here treats anti-state rioting as a form of urban protest and looks at state-society divisions rooted in urban geography and policy that have been overlooked in conventional scholarship on minority mobilization in Europe.

Fighting for Their Neighborhood 1719

Fighting for Their Neighborhood: Urban Policy and Anti-State Riots in France
Michalis Moutselos, Max Planck Institute for the Study of Religious and Ethnic Diversity W hy does anti-state, violent rioting take place in advanced democracies?The paper investigates the role of the urban environment in shaping grievances expressed and mobilization/counter-mobilization processes observed during a riotous episode.In particular, I look at large social-housing estates as a propitious urban setting for the eruption and sustenance of anti-state violence.I identify three mechanisms (stigma amplification and inversion, failure of state intervention in the form of everyday administration and emergency policing, and advantages for network activation and resource mobilization among potential rioters) that complement standard explanations of rioting based on socioeconomic and ethno-cultural grievances.I test the theoretical model using a controlled case study of two neighboring suburbs in the North of Paris, with similar socioeconomic, demographic, and political characteristics but different violent outcomes in the 2005 nation-wide wave of French riots.The paper traces the source of local variation to the exogenous presence of large, concentrated social-housing estates in one, but not the other.The analysis here treats anti-state rioting as a form of urban protest and looks at state-society divisions rooted in urban geography and policy that have been overlooked in conventional scholarship on minority mobilization in Europe.
The occurrence of mass-scale, violent, anti-state protests and riots is a puzzling phenomenon in consolidated democracies.Potential costs are high as participants face almost certain punishment if caught by well-organized police forces.Gains from violent action are uncertain, given its extra-political nature and lack of wider legitimacy.Yet in the last 50 years, there have been many instances of anti-state riots in the United States (from the MLK-assassination riots to the more recent Baltimore and Ferguson riots), the UK, France, Sweden, and elsewhere (Dancygier 2010; Moran and Waddington 2016).What explains the incidence and intensity of anti-state violent protest?Furthermore, such riots I would like to thank Mark Beissinger, Rafaela Dancygier, Fabien Jobard, Rebecca Thomas and anonymous reviewers for their input when writing this article.I would also like to thank Nicolas Oran for kindly allowing the reproduction of his photos from Saint-Ouen and Aubervilliers for the Appendix.Direct correspondence to Max Planck Institute for the Study of Religious and Ethnic Diversity, Hermann-Föge- Weg 11, 37073, Göttingen, Germany; e-mail: moutselos@mmg.mpg.de.feature puzzling regularities.Though typically leaderless and lacking specific demands, they resemble organized collective action: diffusion across several locations happens in a seemingly coordinated way and protesters selectively target state agents and property (police cars, government buildings, local schools, etc.).How can we explain such regularities?
Existing literature has primarily focused on rioter motivations, distinguishing race, ethnic, immigrant, or poor people's riots (Lieberson and Silverman 1965,  Piven and Cloward 1977, Olzak 1994, Kawalerowicz and Biggs 2015; see Wilkinson 2009 for an overview).Others highlight the state's dilemma between mediation and repression (Bleich, 2010) or the consequences of policing tactics (Schneider 2014).Much less prominence has been given to the "urban" aspect of anti-state riots; that is, the apparently banal observation that rioting unfolds in particular urban environments and not others.
Yet this seems to be an important omission.Living in circumscribed urban spaces is associated with simmering, long-term conflicts over the value of housing, the use of public space, and the provision of public goods, such as order and cleanliness.Neighborhoods nurture thick networks activated during mobilization and diffusion of violence (Myers 2010).Urbanistic characteristics facilitate or hinder formal and informal policing.Rioters present themselves as the defenders of a well-circumscribed territory and its inhabitants whom the state should accommodate rather than repress (Bleich et al.  2010).
In many West European democracies, such conflicts over urban resources, thick peer-group networks, and trade-offs between state accommodation and repression are associated with social-housing estates.Indeed, nearly all major rioting incidents in France since the 1980s have taken place in such estates, known as grands ensembles Jobard (2009b).Similar neighborhoods have experienced anti-state riots in European countries with varying integration policies, immigrant populations and policing traditions, such as Sweden, England, and Belgium.France, the country from which the cases for this paper are drawn, has an exceptionally large and geographically dispersed stock of social-housing estates. 1 It has also featured more sustained and widespread anti-state rioting than other countries.
I draw empirical evidence from a controlled case-study comparison of two neighboring municipalities in the Northern suburbs of Paris, which experienced different levels of violence during a nation-wide wave of anti-state riots in October-November 2005.The presence of concentrated social-housing estates in one (Aubervilliers) but not the other (Saint-Ouen) generated local conflict before the riots, nurtured peer-group networks activated and tactics mobilized during rioting, and determined repression success/failure.In particular, rioters identified with and defended their cité (social-housing estate) from intruding security forces; they relied on resources acquired during the everyday, low-level conflicts there; formal and informal policing was hampered by architectural features and tactical disadvantages related to land use and ownership status.The research design controls for other explanatory factors highlighted in the rioting literature-poverty, post-colonial immigrant presence, political underrepresentation of minorities, and police tactics.
Highlighting the "urban" in urban riots shifts the focus towards the unintended consequences of urban policy.Decisions over land use, construction type, and ownership status have important consequences for mobilization, as scholars of urban social movements have long argued (Castells 1983; Gould 1995;  Pickvance 2003).Zhao (1998) has shown that urban ecology can determine the spread of dissident ideas and networks, and the success of spatial strategies for collective action.Collins (2009) has similarly argued that the study of violence, including riots, should focus less on structural or cultural factors and more on temporally and spatially situated interactions between perpetrators and victims/targets.In the case study presented here, the choice of cheap, mass-produced and geographically concentrated social-housing units made by French politicians and central planners more than five decades ago reverberated throughout the 2005 riots.The argument also nuances the relationship between West European states and populations of immigrant origin, because it focuses on state policy not traditionally associated with conflicts around immigration. 2

An Urban Theory of Anti-State Riots
Anti-state protests in consolidated Western democracies are often triggered by perceived police brutalities.The 2005 French riots were no exception: three teenagers fleeing a police operation in the Parisian suburb of Clichy-sous-Bois hid in an electric substation and were electrocuted on the evening of October 27.Two were pronounced dead soon afterwards.Participants in the violent clashes that raged for 3 weeks came predominantly from families of North-African and Sub-Saharan African origin (Delon and Mucchielli 2006;  Mazars 2007).Scholars argued that the riots reflected ethno-cultural grievances over discrimination in French society and within French state institutions, in particular the police (Schneider 2014; Murray 2006).Others offered different versions of an ethno-culturalist explanation, such as the socialization experiences of young people of (mostly Sub-Saharan) African origin (Lagrange 2008) or minority frustration with French ethnicity-blind policymaking (Suleiman 2005).Yet more sophisticated accounts took note of subnational variation among localities with heavy post-colonial immigrant presence and argued violence mirrored uneven political incorporation of minorities (Jobard 2009a; Garrett  2013).
Taking ethno-cultural grievances as the sole basis of rioter motivation does not account for the diversity of backgrounds and fluidity of experiences among immigrant descendants, especially in a country like France that lacks consolidated ethnic and racial identities (Wacquant 2006; Roy 2005); it sidesteps the "positive" mobilizational aspects of the phenomenon, focusing exclusively on grievances, unlike accounts of rioting in developing countries (Wilkinson 2006,  Varshney 2003); it ignores relevant state institutions and policies beyond the police.

Grievances
Why would a progressive policy like accession to social housing become an important source of grievances, leading to anti-state riots?First, towering over the edges of cities, social-housing estates act as concentrated and visible epicenters of social marginalization (Dikec 2011).They are thus vulnerable to stigmatization, which amplifies pre-existing structural inequalities.Furthermore, municipalities with such estates become polarized over their impact on local safety and economy.Such processes of stigmatization and polarization resemble concentration effects, familiar to urban sociologists studying neighborhood effects (Massey, 1991; Sampson and Morenoff 1997; Bowles, 2006).The stigma of living in a housing estate and the need to reverse it are instrumental in taking up protest.
Second, grievances spring from everyday failures to administer these estates.The lack of private ownership, the large scale of maintenance operations and outsourcing to housing firms, and building superintendents creates principalagent and collective action problems, familiar to scholars studying diseconomies of scale (Williamson 1976; Riordan and Williamson 1985).The subsequent under-maintenance of buildings (repairing broken elevators and windows) and public spaces and poor information about the problems of inhabitants causes negative spillovers for life in these neighborhoods.High costs of abandoning the existing social-housing infrastructure mean that social housing estates exert strong lock-in effects in conducting policy (Pierson 2000, 256).A host of actors, including building managers, housing companies, and town hall employees, are heavily invested in their upkeep.It is costlier to tear down the estates and start anew, than manage their negative social consequences, including occasional riots.

Repression
Successful policing is instrumental in preventing a violent protest and curtailing an ongoing riot.Police interventions in the French suburbs are tinged with accusations of institutional racism (Fassin 2013; Schneider 2014; Body-Gendrot  and de Wenden 2003).Without denying discriminatory police practices (racial slurs and overindulgent stop-and-frisk operations), the argument here focuses on failures to conduct successful patrolling operations in social-housing estates.A mismatch of police tactics, goals, and organization encourages low-intensity, pre-rioting conflict and hinders police interventions during actual rioting.
Officers deployed in these neighborhoods on a regular basis often consist of young and inexperienced recruits with poor knowledge of local terrains.The more experienced, nonuniformed officers who intervene in exceptional situations using unmarked cars (the so-called anti-criminality brigades, or BAC) tend to perform blanket controls/operations, including random identity-checks that fuel suspicions of ethno-racial profiling and lead to consternation among local youth (Roché 2010; Fassin 2013).In case of a violent flare-up, heavily armed units (in France, the Republican Security Companies, or CRS, and the mobile gendarmeries) use large-scale, indiscriminate repression, encircling the entire estates and occupying the main entrances, open-spaces, and parking lots.The combined tactics of the BAC and the CRS, while ill-suited for the guerillastyle tactics of rioters, fuel the rallying cry of the "neighborhood under siege".They inadvertently create an "audience" (Collins 2009: 198-200) staged within the confines of the estate for rioters to escalate insults into actual violent acts and to recruit bystanders.
Why do police forces not adapt their tactics to the exigencies of the urban environment?Large social-housing estates pose tactical obstacles for an institution keen on protecting its agents from exposure to danger.Heavy concentration of youth groups, rapid diffusion of rumors, availability of escape routes, and vantage points for confrontation, and the possibility of being "trapped" increase the costs of mobile, targeted policing.As Dhattiwala (2016) has shown, the combination of alternative escape points for attackers and high concentration of targets emboldens rioters.Meanwhile, police agents cannot rely on informal information and prevention, as the estates lack the mixed residential and commercial uses that provide "eyes on the street" (Jacobs 1961).In response to these problems, blanket operations focus on remote territorial control and nontargeted patrolling of public spaces, as opposed to targeted interventions and the collection of relevant information (Dufresne 2007: 56-9).Rioters understand that the costs of violent escalation within the cordoned-off areas are low, and rioting resumes for several hours, even days, eventually dying out without major police casualties.

Identities, Networks, and Resources mobilized
What is the real or imagined collectivity with which rioters identify?Why do they join and what is their intended audience?In other words, what are their selfidentity (in the sense of self-perceived groupness) and its composites (Abdelal  2009; Brubaker and Cooper 2000)?The particular urban environment of the social-housing estate provides a basis of territorial solidarity, understood both as group identification and network activation (Lepoutre 1997; Kokoreff 2008;  Jobard 2009a).This is where potential rioters spend most of their time outside school, make friends, and encounter state and town-hall agents.This cité identity and status-based incentives within rioter peer groups for participation and mimetic diffusion of violence are the "positive" dimensions of rioting, beyond grievances.I follow Scacco (2010) in arguing that informal estate networks apply peer pressure, provide basic information (where the fighting is taking place and who participates), and induce social costs upon those shirking.In other words, they act as relational networks of mobilizing for urban protest.
Furthermore, rioters use urbanistic aspects and knowledge of the terrain to their advantage.Vantage points for attacking and hiding, easy coordination across hallways, and access to information about location of police agents make these urban spaces propitious terrains for guerrilla-style tactics.If the crucial resources for social movement leaders are visibility, size of participation, and financial assets, the important resources for rioters are clandestinity, high mobility, and precise information.Large estates of social housing provide such resources in abundance, to the extent that rioters are able to evolve into specialists of hit-and-run violence and deploy this specialized knowledge during riots, facing a police force unwilling to intervene directly (Collins 2009, 71).
The following section provides empirical evidence for the hypothesized mechanisms (grievances, counter-productive repression tactics, peer-group networks of mobilization, and opportunity structures particular to social-housing estates) using a research design that controls for main alternative explanations.

Case Studies
Aubervilliers and Saint-Ouen are neighboring municipalities (communes) located just north of the city of Paris and separated by a narrow 2-km strip of the municipality of Saint-Denis.They belong to the western part of the Department of Seine-Saint-Denis, which has a reputation for poverty, high post-colonial immigrant presence, and concentrated social handicaps, encapsulated in the iconic postal code 93' (Kepel 2012).In 2006 (the closest year to the 2005 riots for which we have complete census data), Aubervilliers had a population of 72,300 compared to 42,950 of Saint-Ouen, and both had among the highest population densities in the region. 3Both are easily accessible from central Paris by metro, regional railway, and several bus lines.It would be misplaced to compare a relatively peaceful municipality like Saint-Ouen, adjacent to Paris, well-served by public transportation and easily penetrated by the forces of order, to a riotous municipality at the edge of the Paris region.Violence in Aubervilliers contradicts the argument that riots happen in isolated and geographically detached municipalities (Body-Gendrot 2005:10).Municipalities adjacent to both Aubervilliers (La Courneuve, Bobigny, Saint-Denis) and Saint-Ouen (Saint-Denis, Villeneuve-la-Garenne) experienced violence-both were vulnerable to diffusion from neighboring territories.
The cases provide a fertile testing ground for the effects of socioeconomic decline and anomie (Beaud and Pialoux 2003).The municipal archives of Saint-Ouen and Aubervilliers feature photographic and cartographic traces of a rich industrial past; both cities saw major factories close between 1965 and 1980 (Cohen 1990), followed by a steep population decline.In 2006, unemployment levels were much higher than the national average of 8.4 percent (see Table 1), but their local economies grew more than the national average prior to the riots.Still, average household revenue remained considerably low.About one-third of Saint-Ouen inhabitants and 40 percent of Aubervilliers inhabitants lived below the poverty line in 2005, the national average being 13.2 percent.
Aubervilliers and Saint-Ouen also share very similar post-war political histories, dominated by the local Communist Party (PCF).Prior to the riots, local municipal councils had experienced limited progress towards including persons of immigrant origin.In 2005, three Adjunct Mayors out of fourteen in Aubervilliers (Employment, Sport, and Child Policy) and one out of eleven in Saint-Ouen (Urbanism) were of post-colonial immigrant (North-African) origin. 4These numbers are quite low, given local demographics.The very slow pace with which the French Left, and especially the Communist Party, incorporated immigrant descendants has been noted in country-wide studies (Garbaye 2005; Geisser 1997).
The two municipalities also shared similar immigration patterns.In 2006, Aubervilliers was home to about 31 percent of non-EU citizens, with about 1 / 3 from North Africa (Algeria, Tunisia, and Morocco) and 1 / 4 from sub-Saharan Africa. 5Saint-Ouen population was made up of 29.1 percent foreign citizens with 2 / 3 originating from North and sub-Saharan Africa (both far higher than national and departmental averages) 6 .Local histories indicate similar patterns of incoming migration: heavy inter-war Southern European inflows (Spaniards in Saint-Ouen; Italians, Spaniards and Portuguese in Aubervilliers), followed by North Africans from French colonies after World War II, with recent additions of sub-Saharan Africans and Asians (Lillo 2009; Lounici 2014).
Finally, both municipalities suffered from high crime rates, among the highest nationally for municipalities of their size, suggesting a link between everyday delinquency and rioting.A careful breakdown of crime types recorded by the French police highlights some differences: Saint-Ouen featured many arrests for illegal-substance use, while Aubervilliers scored slightly higher in categories indicating small-scale conflict with the authorities (destruction of public property, verbal and physical attacks against forces of order).In 2004, a year without major riots, Aubervilliers witnessed 64 incidents of damages of public buildings and 131 incidents of aggression/violence towards police officers-Saint-Ouen featured 37 and 90, respectively (Direction Centrale de la Police Judiciaire 2004).
The proportion of apartments dedicated to low-rent social housing (HLM) in 2005 was roughly equal (40.64 percent in Aubervilliers and 45.5 percent in Saint-Ouen).Both stood higher than the department's average of 35.73 percent. 7 This reflects a long-standing political choice of the Communist-led councils to cater to poorer constituents and solidify electoral support (Subra  2004; Guglielmo and Moulin 1986).The crucial difference between the two municipalities, quickly noticeable to the casual observer, lies in the concentration and type of social housing.The biggest social-housing estates in Saint-Ouen are the cités Dhalenne, Soubise, Allende, Michelet, and La Motte, all built between 1960 and 1974, with 200-300 apartment units. 8The remaining socialhousing estates rarely surpass 100 units.The urban history of Saint-Ouen can be starkly contrasted with that of Aubervilliers: the latter experienced bouts of concentrated construction with 800 apartments in 1954 (Emile Dubois estate), 520 apartments in 1958 (cité Gabriel Peri), the very large Vallès-La Frette grand ensemble (more than 1,500 apartments over a dozen neighboring towers in Aubervilliers' North-East), and 1,000 apartments in 1974 (Maladrerie). 9hese differences reflect different circumstances during postwar reconstruction.The Aubervilliers municipality owned large tracts formerly used in agriculture, ideal for large-scale construction and not available in Saint-Ouen.The then Mayor, Charles Tillon, made municipality-driven housing construction his priority to combat dire lodging conditions and teamed up with famous modernist architects to build some of the earliest grands ensembles in the Paris region.In comparison, municipal authorities in Saint-Ouen built social-housing units faster and at a smaller scale, facing pragmatic constraints.Much of land in Saint-Ouen fell under the shared ownership of private and public institutions.For instance, the Docks, in the northwestern part of the municipality consisted of land co-owned by private companies Total and Alstom, the City of Paris, and the National Railways.Meanwhile, the local municipal council did not construct at a larger scale, because about one-third of residential buildings were destroyed during World War II.This meant that construction had to take place very fast, too fast to incorporate the architectural and industrial advances that facilitated the construction of the grands ensembles elsewhere.

The Location and Intensity of the 2005 Riots in the Two Municipalities
Saint-Ouen and Aubervilliers were touched by the 2005 riots, but differences in intensity of violence were stark.According to the official statement of the Aubervilliers fire brigade, recorded incidents included 104 destroyed vehicles, 30 instances of waste containers set on fire, and 23 damaged buildings, including two important instances of fires, one at the gym of the Gabriel-Peri high school and a 500-square-meter storage facility.Molotov bombs were thrown at firemen, two police agents were injured, and a bypasser suffered a flash-ball shot. 10In addition, a Korean journalist covering the riots was injured by a group of local youth and transferred unconscious to a nearby hospital. 11Aubervilliers was also one of the earliest municipalities to experience riotous incidents beyond Clichysous-Bois (on November 1).The incidents, according to newspaper reports, lasted a whole week.
In contrast, Saint-Ouen was only touched by 3 nights of violence, between November 3 and 5. Overall, 17 vehicles were set on fire, according to the official fire-brigade report used by the Mayor in her address to a council meeting on November 14.Two small fires were recorded and one major incident of verbal confrontation between the police and a group of young men took place next to the local ice rink, but did not escalate.The entrances of three buildings were lightly damaged.Despite fewer incidents, the police were able to make two arrests for immediate trial.The Mayor's report noted that the municipality was not touched by the same scale of incidents as other municipalities in the Department and that the mobilization of municipal police and association leaders was unprecedented, a "new situation", since Saint-Ouen "has not been touched by these types of urban violence before". 12In the December 2005, the issue of the municipal newspaper, Éric Pereira-Silva, a local councilor of the non-Communist Left opposition, was surprisingly upbeat: "Despite some intolerable incidents, our city was until now relatively spared by the wave of urban violence that has shaken the rest of the country.The elements of local explanations: urbanism at a human scale, social housing of high quality, active social policy, potential for activities and employment, a developed network of public transport, active local associations". 13ccording to the explanation proposed here, recorded riots in Aubervilliers, but also the few incidents in Saint-Ouen, should be geographically circumscribed in dense agglomerations of social-housing estates.In Aubervilliers, the Gabriel Peri high school is located next to the Maladrerie estate; the storage facility set on fire is adjacent to the other big estate in northeastern Aubervilliers, the Vallès/La Frette complex.Journalists of Le Monde newspaper, who attempted to interview Aubervilliers rioters as they were committing violent acts, highlighted the "Cité du 112" as a focal point. 14Two failed attempts to instigate a fire took place at the Le Corbusier high school, across the street from the estate Émile Dubois and the local court of first instance.In Saint-Ouen, there is more detailed information on the location of incidents, drawn from a fire brigade report cited by the Mayor to the municipal council.Two major fires were recorded, one on 72 Avenue Michelet and the other on 41 rue Albert Dhalenne.Consistent with the empirical implications of the argument, both face the largest social-housing buildings in the municipality, the towers in Michelet and the Dhalenne/Soubise estates in Vieux Saint-Ouen.
Did affected estates represent the poorest residential areas of Aubervilliers?Did they host the highest percentages of foreigners?More fine-grained census data, corresponding to the level of IRIS 15 , the smallest geographic unit for census data in France, allow us to test these alternative hypotheses.The evidence for the link between unemployment and incidence of violence is at best mixed.While the IRIS areas corresponding to the Maladrerie and Robespierre estates scored high (3rd and 6th, respectively, out of 28 sectors), Vallès/La Frette and La Villette were not the neighborhoods most affected by unemployment (11th and 17th).Regarding the presence of foreigners, Maladrerie, Robespierre, and Vallès/La Frette hosted well below the municipal average of foreign citizens, while La Villette was the only neighborhood where more than 40 percent of residents had foreign nationality.Most striking was the percentage of social-housing units: an astonishing 89.7 percent of all residences in Vallès/La Frette correspond to this type of housing; the percentages are 73.7 percent in Maladrerie, 63.7 percent in Robespierre, and 43.7 percent in La Villette, all above municipal average and denoting the presence of large grands ensembles.
Table 2 presents the results of a simple logistic regression analysis for the 28 Aubervilliers IRIS units.The binary outcome variable is the occurrence of violence in a particular IRIS, as recorded by police sources and the local press.Seven ( 7) out of 28 IRIS were recorded as having experienced violence.The analysis demonstrates that the high percentage of social-housing units in an IRIS, indicating the presence of large estates, is positively and significantly associated with the occurrence of violence, controlling for the share of foreigners and unemployment rate.Indeed, the average percentage of social housing units was 66.1 percent of total housing stock for affected areas, 25.1 percent for neighborhoods that remained quiet.The violent and nonviolent IRIS had, on average, similar levels of unemployment, while nonviolent IRIS in Aubervilliers hosted higher percentages of non-EU foreigners than neighborhoods affected by violence.
Aubervilliers experienced a serious anti-state riot a few months before the November incidents in the large social-housing estate of La Villette, and confrontations were relatively contained both geographically and temporally.A local youth, who had moved to another suburb south of Paris some months before, was killed in a motorcycle chase initiated by the anti-criminality squad (BAC).Escalating violence, triggered by rumors that policemen had caused the fatal fall, involved several dozens of estate youth, numerous car burnings, and the destruction of a local storage facility. 16Interestingly, the riot did not erupt close to where the young boy was killed, nor where he lived at the time; instead mobilization was initiated by networks of youth at the cité where he grew up.In November, anti-state rioting would erupt in La Villette once more, but it would also encompass estates on the northern, opposite side of Aubervilliers.

The Social-Housing Argument Described by Local Actors
The comparison draws from interviews with local actors conducted in two waves, between March and December 2013 and in February 2015.In total, 20 local actors were interviewed (9 in Saint-Ouen and 11 in Aubervilliers).
The interviewees held crucial positions related to prevention of urban violence (local police commissariat, municipal office responsible for security, youth or urbanism, and local association leaders) or had relevant professional experience (high-school teacher and social-housing caretaker).Most of them already held their positions in 2005, and the few that did not were otherwise involved in municipal life in other official capacities. 17erhaps the most direct finding from the interviews is the belief that the dispersion of social housing and other urbanistic features mattered for the exceptional tranquility in Saint-Ouen.Nearly all Saint-Ouen respondents mentioned this when directly confronted with the puzzle motivating the case selection.
"The distance from the city center is important.Then there is an urbanistic factor.In Saint-Ouen, we have no large agglomerations of housing apartments.In Saint-Ouen, the most dense agglomeration is 500-600 apartments in some hectares."(G, representative of local OPHLM office-social housing).
The same interviewee hinted that violent mobilization in Saint-Ouen did not profit from concentration effects present in other municipalities; when violence erupted, it never reached the critical mass of a large riot: "[In Saint-Ouen] There is no enclavement [ . . .] You cannot really go and burn the car of your neighbor, because you know well his brother, his sister.You cannot gather 300 or 400 youth, for instance to destroy, I don't know, the school, the town hall, that's practically not possible." A municipal councilor noted that it is "up to the state" to intervene in large estates, revealing the strong lock-in effects of urban policy: "There are grands ensembles in Saint-Ouen, but they are indeed smaller than elsewhere in Seine-Saint Denis.You see these neighborhoods close to the town hall, they have nothing to do with La Courneuve or Aulnay-sous-Bois [ . . .]There are a couple of places like that in Saint-Ouen, but they are rather small.If you have a neighborhood like that that is not 3,000-4,000 people but 30,000-40,000 people, I think it is up to the state to change that."(H., municipal councilor in Saint-Ouen).
Another Saint-Ouen councilor reflected on the difficulties of administering these spaces: "People are piled up one over the other, housing conditions are not good and public services do not exist, the housing caretakers (bailleurs) have difficulty doing their job.And people have the sentiment of living in a ghetto.I am not saying that these are ghettos, but people have the perception of living in ghettoes.There is no voluntarist policy to create ghettoes in France, socialhousing providers have a duty to assign apartments to everyone." To be sure, the municipality of Saint-Ouen was not a paragon of municipal management.Interviewees revealed anxieties about armed violence but noted that this was related to the booming narcotics trade benefitting from Saint-Ouen's proximity to Paris. 18In 1998, a Saint-Ouen high-school, Marcel Cachin, located close to the cité Allende, experienced everyday brawls, racist, and sexist insults and threats towards school personnel.The teachers started a strike to protest the violence and lack of financial means and acquired the status of "Educational Priority Zone". 19In another 2003 police report, again inside the cité Allende, two policemen were injured after twenty local youth showered them with rocks. 20Although these incidents demonstrate similar problems to Aubervilliers (and, indeed, these were the places touched by low-intensity violence within Saint-Ouen in 2005), the scale of everyday violence was smaller and less geographically concentrated.

Stigmatization, Polarized Public Opinion, and Maladministration in Social Housing Estates
According to the urban-ecology model of rioting presented here, local grievances over maladministration of the estates fuel rioter complaints that they reside in "abandoned" neighborhoods.The 2004 Aubervilliers Local Urbanism Plan (PLU) contains results of an opinion survey among inhabitants.They give the overall impression of an ugly, unsafe, and badly managed municipality but also of one profiting from proximity to Paris and municipal facilities (theater, library, and sports venues).Tellingly, respondents evoke social-housing estates as one of Aubervilliers' least attractive aspects, citing "too many social-housing units", "social-housing buildings that are almost unhealthy", and "the big HLM towers". 21More than half (53 percent) single out housing as an area for improvement, above employment (42.8 percent).One finds specific comments on "diversification of housing offers", the need to "allow access to property developers for the construction and renovation of small buildings", to "obtain government credits for the rehabilitation of social or private housing", to "improve the relations between OPHLM and customers", and to "demolish degraded houses to make new ones". 22any respondents are bothered by the heavy immigrant presence ("too many different cultures" and "too many foreigners")-Aubervilliers was no interethnic Eden in 2005-but the issue does not stand out, unlike unemployment and improvement in the condition of the estates.Reflecting a local discourse polarized around the estates, reports of dilapidated housing and everyday vandalism, and lack of commercial activity are restricted to neighborhoods with grands ensembles.In Vallès/La Frette, inhabitants denounce "the acts of vandalism", "the dead commercial center", and "too many apartments that should be renovated"; the Maladrerie/Emile Dubois complex is castigated for its urban decay, reigning insecurity and lack of civility.The neighborhood is described as "obscure", in reference to its highly complex urban form, and several respondents note that "everything is permitted" there. 23n January 2005, two national agencies specializing in matters of security conducted a victimhood survey in Aubervilliers and discovered low-level conflict inside the estates.A third of the 994 respondents resided in the municipality's grands ensembles and reported having been affected by "urban disorders", such as vandalism in public spaces and destruction of private vehiclesa significant 15 percent more than the rest of the municipality (Nevanen  et al. 2006: 102-105).Further evidence of simmering tensions within the grands ensembles is drawn from minutes of neighborhood consultative councils (conseils des quartiers).Ironically, a few weeks before the riots, the Mayor visited the neighborhood councils, collected the concerns for the municipal magazine, and revealed a bright boundary between the inhabitants of the estates and other neighborhoods.At La Villette, many "consider [ed] the situation to be very fragile after the spring riots", bemoaning "intimidations and repeated aggressions"; at the Maladrerie-Emile Dubois, a local pharmacist noted that "after a period of lack of security things are better", while at Vallès-La Frette and Cochennec-Péri, the inhabitants emphasized the slow deployment of policemen for everyday nuisances. 24hronically bad administration was another source of estate-specific grievances.Aubervilliers building superintendents who were interviewed tied problems to scale and ownership type.Three superintendents from La Villette emphasized the responsibility of inhabitants-who do not clean up their waste and neglect mailboxes, elevators and public spaces.The superintendents also highlighted the negative externalities inherent in having many families living in the towers in cramped conditions.If there was one apartment with loud music, the entire block would suffer; when the entryways were occupied by youth, many residents would feel the repercussions.The same would apply to leaving one facade dirty or one window broken: the entire block would get a bad reputation. 25Urban planners devising regeneration plans for Aubervilliers and Saint-Ouen a few years after the riots would find differences in social-housing administration.In Aubervilliers, they focused predominantly on neighborhoods that had experienced violent rioting, the grands ensembles La Villette, Vallès-La Frette, La Maladrerie, and Emile Dubois.The planners underlined a lack of cooperation between estate administrators (identified as social-housing agencies and technical staff) and users of public spaces and a lack of "situational prevention". 26Neighborhoods in Saint-Ouen, in contrast, featured "advanced poverty and conditions of poor hygiene in a collection of ancient buildings, coexisting in an environment where social housing, small town houses and zones of detached housing participate in common in the construction of a dynamic social life, marked by demographic, cultural, social and economic diversity". 27n short, enacting urban policy in Aubervilliers was dominated by problems and actors associated with the gigantic estates, while Saint-Ouen was spared this veritable institutional lock-in.

Anti-Riot and Everyday Policing in the Two Municipalities
The most proximate cause of the disparity in violent incidents between Aubervilliers and Saint-Ouen was the ease and rapidity of police and fire-brigade intervention.In the post-riot report by the Saint-Ouen Mayor, the efficacy of intervention is explicitly contrasted to failures in other municipalities in Seine-Saint-Denis. 28as the Aubervilliers commissariat discriminatory towards the youth of postcolonial immigrant origin?Recruiting, training, personnel transfers, and turnover for the Seine-Saint Denis commissariats took place at the national and departmental level, irrespective of local conditions.In fact, evidence of racist practices makes Saint-Ouen a more likely candidate for conflict: according to a 1997 survey among 1,000 Saint-Ouen high-school students, only 5 percent felt comfortable with policing in their municipality; 45 percent reported having experienced or heard about racist incidents. 29Nor was the Saint-Ouen commissariat perceived by locals as more competent.A local politician and community organizer noted that national police forces stationed in Saint-Ouen took decisions in a very bureaucratic fashion, often did not know neighborhoods well, and lacked proper training for situational prevention. 30ndeed, a main difference lies in the involvement of unarmed municipal forces in neighborhood patrolling-significant in Saint-Ouen, but minimal in Aubervilliers.A second, related, difference was the mode of deployment of national police agents before and during the riots: static, heavy-handed, and isolated from local actors in Aubervilliers versus mobile and engaging with municipal forces, building caretakers and inhabitants in Saint-Ouen.Firemen in Saint-Ouen relied on reports by inhabitants and association leaders to intervene in spots where fires had broken out.The head of the then fire brigade highlighted the support of inhabitants, who "were concerned about our morale" 31 .There was also a notable mobilization of specially trained municipal employeesthey would patrol the streets in groups of two until the early morning hours and report to the local commissariat or intervene themselves to resolve any issues. 32In particular, these patrolling teams focused on five neighborhoods adding "knowledge of the terrain, the streets, the people and the buildings", signaling Molotov bombs, and ascertaining the safe passage of firetrucks. 33argeted interventions based on high-quality information prevented the escalation of isolated incidents into sustained rioting.Bystanders (the "audience") provided informal policing rather than a source of recruitment.The intervention of heavier police forces was minimal-70 CRS agents over three nights (between November 3 and 5).Besides inhabitants, local firemen could rely on help from the building caretakers and shopkeepers.Both groups signaled fire locations and removed combustible garbage after police interventions.Mayor Rouillon was quick to capitalize politically on the relative calmness of the city stating before the municipal council that Saint-Ouen did not project an image of a "war zone [and lacked] the atmosphere of anxiety or psychosis, largely nurtured by the media and played, from the point of view of fire-setting, a particularly negative role." 34 A municipal councilor at the time, when interviewed some years later, had no memories of large fires in the municipality, and of no substantial findings of projectiles in the corridors and balconies of suspected buildings. 35n contrast, even though there was mobilization by municipal employees, prevention and suppression in Aubervilliers necessitated heavy intervention by several units of departmental CRS, and even regional police forces. 36The pattern in La Villette, Emile Dubois, and La Maladrerie was similar to other big estates at Seine-Saint-Denis and around the country: deployment of police cars and trucks at the entries and on the perimeter of the grands ensembles and static interventions against mobile rioters or frequent ID checks. 37During the riots, the police also prioritized defending important public buildings outside the estates, such as schools, libraries, and old-people's homes, instead of mobile policing and information-gathering within the cités. 38In one occasion where police forces and firefighters intervened successfully and stopped an attempted fire at the local court of first instance at Stalingrad Square 39 , the building stood outside large estates and next to small and large commercial units.
Differing police reactions in 2005 reflected long-standing patterns in the two municipalities.Heavy-handed interventions were common in Aubervilliers, but infrequent in Saint-Ouen.For instance, when probed about the specifics of policing in Saint-Ouen, a local HLM officer noted: "When the cops intervene in the large grands ensembles (Stains, Aulnay etc.), they arrive in greater numbers, and it's not the same Brigade.The Brigade that intervenes over there . . .they are armed and move like the CRS.Here we sometimes have them, but when they intervene . . .it's more the BAC in civil clothing.They don't have the protection, the helmets, etc.At Clichy it's the 'Robocops'.And they have to, when you expect a fridge to fall on your head.They advance like they are against a demonstration.They advance by outbidding the violent demonstrators."[G, representative of local HLM office].
When rioting broke out in Aubervilliers in April 2005, the immediate reaction was to send a hundred stationary, heavily-armed CRS agents to guard the entrances of La Villette-promptly receiving projectiles and Molotov cocktails from high vantage points in the towers. 40Asked about the violence in his municipality, and whether there was any information about the organization of the perpetrators, Bernand Vincent, special councilor to the Mayor of Aubervilliers on matters of security, provided an honest account of the micro-dynamics of violent escalation: "These are not structured groups.They are informal groups that are created as the events develop . . .We have our operations of observation, but we do not have informants among these groups and we cannot be sure how the events might develop." Such divergent reactions to the eruption of riots reflect different experiences with local security operations.Aubervilliers relied almost entirely on national police forces.In 2004, the Mayor of Aubervilliers expressed the opinion that "public security is and will rest the competence of the central state.While public acts of delinquency continue to rise, it is the job of the national police to concentrate on this front". 41In 2005, a few months before the riots, security concerns of residents in La Villette were met with a tepid "we will petition the competent authorities".At the Maladrerie-Emile Dubois complex, the Mayor promised a regeneration program "which will start, if we receive funds from the state in the context of urban renovation".At Vallès-La Frette, he noted: "A minority of individuals cannot poison the life of others.The municipal police will surely give solutions to certain problems, such as parking and waste collection.Nevertheless, the national police have to take up their responsibilities." 42he Saint-Ouen town hall, on the other hand, had long experience with patrolling operations by municipal employees that did not involve national police forces.In 1982, a "Local Security Commission" instituted informal patrolmen, called "îlotiers". 43In the early 2000s, the municipality assigned "city correspondents" a group of young municipal employees to "talk to people, sense the climate in different neighborhoods, find cases of malfunctions, report areas where waste was deposited". 44In January 2003, a full municipal force was inaugurated, reporting to the Saint-Ouen commissariat the complaints of social-housing caretakers and shopkeepers. 45Nonarmed, mobile, and with good knowledge of local hot-spots, they would be instrumental in reporting problems to the fire brigade and the national police during the 2005 riots.
Why did the two municipalities follow different approaches to local security and their relationship to national police?After all, Saint-Ouen faced serious security concerns, throughout the 1990s and early 2000s.The difference lies in territorial aspects of policing the grands ensembles.Just as renovating buildings and public spaces inside the estates requires costly interventions by national authorities, policing vast areas, multiple entryways, and intermingling corridors/balconies necessitated a scale of operations that only the commissariat and the departmental CRS forces could muster.Local policemen and administrators pointed at the fear of being encircled by large groups or attacked from above with heavy objects.Indeed, everyday violent attacks against policemen in Aubervilliers had been a regular occurrence leading up to the riots indicating the existence of a pool of "violent specialists" that were able to sustain rioting in 2005. 46A former member of the Aubervilliers commissariat confirmed that intervening in the municipal estates meant carrying "obligatory bullet vests, sometimes heavy ones, tonfa batons . . .tasers for those who are trained, flashball guns, laser flashball guns, teargas bombs of the size of a fire extinguisher, heavy masks, protective shields for the legs and the body".The equipment was, according to the young officer, "a strict minimum in order to penetrate these zones" where the most basic identity check has policemen "encircled by dozens of delinquents whose principal desire is to hurt." 47 Meanwhile, patrolling troupes could not rely on help from local shopkeepers inside the estates, to prevent and resolve tensions before they escalate, since ownership was largely restricted to residential uses.

Neighborhood Networks and Cité Identity
The interviews I conducted in Saint-Ouen and Aubervilliers included an openended question on perceived identity of the rioters 48 to overcome an initial difficulty in establishing the pertinent group category.The interviewees almost unanimously mentioned the salience of the neighborhood-whether they were policemen, politicians, or social workers (also noted in Simiti 2012: 149-151).An Aubervilliers councilor gave a typical answer, after being prompted that riots in the UK and the US are often identified with specific racial groups: "But no! It's their neighborhood [quartier].When they riot, it is always with their friends from the cité." 49wo building superintendents from La Villette estate in Aubervilliers with direct experience in 2005 (one of them of North-African origin) highlighted that the rioters were, typically, the young men loitering the entryways of tower buildings every day.They underlined that riots were the acts of ever-younger participants seeking recognition among their peers. 50A French sociologist who served as a high school teacher in Aubervilliers in 2005 confirms these insights.Based on interviews with former rioters, he concludes that "their words systematically err towards a 'we' (the 'cité youth'), opposed to a 'them', the context of which is less well-defined, but which takes most often the form of the state and its 'right hand', incarnated by 'the cops'" (Truong 2015: 98).This neighborhoodbased framing also underlined nonviolent forms of engagement in Aubervilliers.A researcher studying neighborhood associations in Aubervilliers was struck by the absence of ethnicity-based participation.Young men of immigrant (in this case, North African) origin exhibited no transnational connections with the countries of their parents.Instead, they would form associations and clubs to help their friends from the neighborhood "get out", acquire a space for afterschool socializing and organize activities to counter the negative image of the estate (Kiwan 2005).
The estate is also the springboard for diffusion of rioting.One Aubervilliers rioter would later explain that the decision to join other rioters was the curious byproduct of solidarity and competition among estates, reinforced by the conviviality of participation and the sense that the whole neighborhood was affected.His account matches statements of Aubervilliers superintendents that peer groups from the cité/estate are the relevant audience providing recognition for anyone turning from a bystander into a rioter: "[the riots in] 2005, that had nothing to do with organized crime!It was more a movement of groups.I was a follower.I did not wake up one day thinking I would go burn cars!I saw other people do it and told myself 'that's what I should do!'You are there inside your neighborhood, you talk.There are fires almost everywhere . . .And then you tell yourself 'Let's go!We will have some fun too'.I know that there is a group from that cité that has left to do some stupid things [faire des conneries]; everyone knows, it's on the street, it's hearsay.The teams start burning things, you join yours.You have grown up in a cité, therefore you hate cops by definition!" (Truong 2015: 104).
Attachment to the estate appears in online posts and comments from the banlieues blogosphere.Blogs were a privileged means of communication during the riots and 13 blogs were shut down by the French police during the riots.On a website documenting the "hottest [the most violent, the most tense] neighborhoods in France", featuring thousands of entries and shut down briefly during the 2005 riots, the Vallès-La Frette estate features prominently among other famous grands ensembles.It contains panoramic tower views, while dozens of comments clamor for the estate to "represent" Aubervilliers. 51There is a manifest obsession with the corresponding postal code, which becomes a symbol of estate reputation.
Another blog includes a typical post, written in the banlieues lingo, which reads: "Aye, La Frette, Pont Blanc, Hemet, Vallès screw you all . . .Auber [villiers] beats you all, Mala [drerie], Pont Blanc, Vallès". 52The comment merges reputations of the Aubervilliers estates with the reputation of the entire municipality.This process was noted by Aubervilliers rioters interviewed for the French Newspaper Le Monde in the heat of the riots: "there is not a competition among the cités, it is pure solidarity".They thus recognized that the mimetic process happened through solidarity with the inhabitants of other estates.Competition took place at the inter-municipal level: estates "represent" the municipality in order not to be outdone by other municipalities and, at least during a large riot, they turn regular competition into emulation to unite against a common enemy.As a rioter would later recall: "everyone was talking about Clichy-sous-Bois.We said 'we can top that'.They are burning cars over there in Pierrefitte.Let's get the buses here in Aubervilliers" (Leiken 2011, 40).
The particular importance of estate pride in Aubervilliers was also evident in April 2005, following the accidental death of the young inhabitant of La Villette.400 people, three quarters of whom were youth from the estate, gathered in a silent march.One young woman stated that she wanted "to associate herself with the distress of the youth at the cité République [another, somewhat smaller, estate in Aubervilliers] where she lived and where Karim would often come."She added: "His death should mark the end of what goes on in our neighborhoods, the delinquency, the vandalisms, the pain." 53There was no equivalent evidence of estate-based solidarity and networks of mobilization in Saint-Ouen, because there are no large estates.As the quotes provided above suggest, the absence of large youth groups and informal policing saved the municipality from the concentration of a critical mass of rioters in 2005.Before these events, there were no inter-estate wars mentioned in local newspapers nor blogs dedicated to particular neighborhoods.Thus, Saint-Ouen enjoyed the "ideal" structural conditions for rioting, the poverty, high post-colonial immigrant presence, and criminality, but lacked the important networks of mobilization and solidarity that led other places to experience rioting.

Conclusion: Generalizability and Theoretical Contributions
This paper has argued that the incidence and intensity of anti-state rioting are often rooted in particular aspects of urbanism at the neighborhood level.I have used the example of the 2005 French riots to show that urban ecology can act as a catalyst for violent collective action by amplifying local grievances, obstructing formal and informal policing and nurturing territorialized identities and networks.The argument can be extended beyond the two municipalities in Seine-Saint-Denis.In 2005, other poor and immigrant-heavy French towns, such as Dreux and Reims, remained largely quiet.Dreux, notorious for interethnic tensions in the 1980s escaped violence in 2005 because the long-time Mayor, Gérard Hamel, instigated a large-scale renovation of the local estates, via his privileged ties with the National Urban Renovation Agency (ANRU). 54At the same time, wealthier and less diverse municipalities, such as Chelles and Montereau in the Seine-et-Marne Department, experienced heavy violence in few, isolated social-housing estates.Since 2005, various municipalities across France, such as Romans-sur-Isère and Vitry-le-François-smaller, in fact, than Aubervilliers and outside the Paris region-experienced anti-state rioting, also geographically circumscribed to local housing estates.This French urban "exceptionalism" is also noted by Lagrange (2008) among possible explanatory factors for the recurring riots, and the present paper delves deeper into the pertinent mechanisms.
The explanatory framework can also be generalized to urban rioting in other West European countries.Similar patterns of anti-state violence (or the absence thereof) were observed during the 2011 London riots, with post-war council estates often providing focal points of protest. 55In Sweden, estates built in the 1960s and 1970s under the Million Program, in particular the estates of Husby, Rinkeby, and Tensta, were the epicenters of the 2013 riots.Brussels experienced a similar case of anti-state rioting in 2006, most importantly in the Marollen district.Local news reports identified rioters as members of a youth gang from the local Querelle social housing estate. 56The absence of similar anti-state rioting in Germany and the Netherlands may reflect different strategies for social-housing attribution.Germany features large estates, but with low occupancy and high private-ownership rates, while Dutch authorities have largely avoided building large estates altogether, opting for a geographic dispersal of social-housing units (Elsinga and Wassenberg 2007:133).
The findings also show that urban ecology and the territorial aspects of mobilization add nuance to the study of violent protest.They account for puzzling variation in eruption of anti-state riots, which has been overlooked by conventional explanations emphasizing racism, policing, and political representation of minorities.More attention needs to be devoted to territorial divisions, created by state policy and affecting political and social behavior.Some scholars have noted the effect of residing in a specific neighborhood on electoral participation (Maxwell 2010), political beliefs (Brouard and Tiberj 2011), and the development of nonviolent social movements (Hamidi  2009).Indeed, as West-European and other democracies become superdiverse (Vertovec 2007) and noncitizens come to enjoy basic social and political rights (Joppke 2001), salient dividing lines might be determined by longstanding state policies, like urban policy, and their, often inadvertent, consequences.

Notes
1.According to 2007 official data, building complexes with more than 100 units made up 31% of the national social-housing stock (56.5% around  Paris).Very large estates of more than 500 dwellings represented 6% of the social-housing stock nationally, but up to 12% around Paris-corresponding to somewhere between 2-4 million inhabitants.See Levy-Vroelant and Tutin 2007: 70-71.2. Citizenship and asylum policy, accommodation of religious rights, discrimination in public schools, courts or the army, representation in public institutions.