Introduction
Riots containing an ethnic, racial, religious or xenophobic dimension are semi-regular occurrences in liberal democracies. The best known examples are the urban disturbances of 1960s America, yet even after 1980 – a time of declining legitimacy for public violence and of tough law-and-order politics – there have been serious riots in American cities such as Miami, New York, Los Angeles, and Cincinnati as well as in major European urban areas such as London, Bradford, Lyon, Paris, and Rostock. Such parallel occurrences in similar political systems suggest that there may be common causes, dynamics, and consequences surrounding these events. But so far there has been strikingly little systematic comparative work on such ‘ethnic riots’ across liberal democracies.Footnote 1
Our article addresses this lacuna by examining state responses to ethnic riots in Western Europe. We focus on state responses for three central reasons. First, while understanding the causes and dynamics of riots is critical for those seeking to prevent them or to limit their severity, it is difficult to envision a future without ethnic riots. How liberal democracies respond to them will therefore be an enduring object of inquiry, and one that has the potential to affect the likelihood of ongoing tensions within communities. Second, while observers of US disturbances of the 1960s have taken seriously the challenge of understanding state responses to riots, there has been little effort to explore analogous outcomes in European cases. This opens avenues for fruitful comparison. Third, closely examining responses to ethnic riots may reveal significant insights about the modern state, such as its contemporary role, the tools it possesses to manage violent events, and how it interacts with citizens who express anger and frustration, a desire for social change, or concerns about law and order.
We build on three bodies of research in this study. A recent wave of scholarship in the developing world has rejuvenated the comparative examination of ethnic riots (Horowitz, Reference Horowitz2001; Varshney, Reference Varshney2002; Wilkinson, Reference Wilkinson2004). These authors establish important propositions about the causes and dynamics of such events, yet the literature as a whole does not focus systematically on the consequences of such riots on politics and policymaking, nor does it explore whether its insights apply in stable liberal democracies. Studies of protest policing in advanced democratic states have developed detailed observations about the relations between social movements and the police (della Porta and Reiter, Reference della Porta and Reiter1998; della Porta, Reference della Porta2006; Waddington, Reference Waddington2007), but they have not examined the effect of less organized forms of rioting on multiple state institutions over the longer term. Scholarship on American responses to the black insurgency of the 1960s and 1970s parallels our topic most closely. It has evolved into a body of systematic and theoretically informed research on the consequences of riots (Piven and Cloward, Reference Piven and Cloward1971; Welch, Reference Welch1975; Button, Reference Button1978; Albritton, Reference Albritton1979; Isaac and Kelly, Reference Isaac and Kelly1981; Hicks and Swank, Reference Hicks and Swank1983; Schram and Turbett, Reference Schram and Turbett1983; Fording, Reference Fording2001). However, this research is limited by its focus on one type of riot in a single advanced liberal democracy in a particular time period, with one author concluding that ‘further work is necessary to determine whether these findings are applicable in other democratic systems’ (Fording, Reference Fording2001: 126). This article explicitly takes up this call by focusing on West European cases since the 1980s.
In order to unify concepts developed separately in research on the developing and the developed world, the following section first defines the broad category of ethnic riots. It then identifies the subset of ethnic riots investigated in this paper: post-1980 intense anti-authority ethnic riots in liberal democracies. We define the dependent variable of the study as the character of state responses, namely whether state actors respond to ethnic riots with repression or accommodation, as this variable helps to determine the ensuing dynamic of state-society relations and has important consequences for justice and interest representation. The subsequent section outlines six theoretical perspectives for analyzing ethnic riot effects based on propositions from the literature on riots in the US and the developing world and from scholarship on violent social movements in advanced democracies. We use these propositions to generate hypotheses about state behavior across our cases.
The next section discusses case selection and methodology before closely examining state responses to two riot episodes each in Britain and France. The penultimate section discusses the findings, highlighting the fact that a combination of the social control perspective advanced by Fording (Reference Fording2001) and electoral incentives perspective supported by Wilkinson (Reference Wilkinson2004) most compellingly accounts for outcomes in our cases. By contrast, the Weberian perspective, the neopluralist perspective, and national structures perspectives fail to explain comparative outcomes; and although the feedback perspective offers insights relevant to some cases, it is too indeterminate to generate a systematic account of state responses. Although we do not view our conclusions as definitive given the number of cases examined here, they strongly suggest that a combined social control/electoral incentives model offers an important foundation for further theory development and testing.
Ethnic riots: definition and framework of analysis
Ethnic riots are episodes of sustained collective violence with an ethnic, racial, religious or xenophobic character. Ethnic, racial, religious and xenophobic motivations share a family resemblance in that they represent core elements of individual and group identities, and are viewed by many as problematic for liberal democracies when they reflect entrenched social or political divisions. At times, there is a debate about the extent to which ethnicity (used in this study as a short-form for ethnicity, race, religion, and xenophobic national identity) is operative in any particular conflict. For our purposes, prototypical examples of ethnic riots are those where all participants and analysts agree on the ethnic dimension of the conflict, but our category also includes events that are perceived by a critical mass of actors as being affected by ethnic considerations.Footnote 2
The term riot has also been problematized, and even disregarded, in studies of collective violence (Roy, Reference Roy1994; Tilly, Reference Tilly2003) and its definitions have been adapted and contested over time by governments and researchers alike (see Wilkinson, Reference Wilkinson2009). However, we follow work by scholars of 1960s and 1970s America and by students of South Asia and developing countries in focusing on riots as measurable events of interest. We build our definition of an ethnic riot around Halle and Rafter’s (Reference Halle and Rafter2003: 347) discussion of a generic riot as ‘at least one group publicly, and with little or no attempt at concealment, illegally assaulting at least one other group or illegally attacking or invading property’. The intensity of the event also helps distinguish a riot from lesser forms of violence, and one key element of intensity is the duration of violence.Footnote 3 Our full definition of an ethnic riot is thus: at least one group publicly, and with little or no attempt at concealment, illegally assaulting at least one other group or illegally attacking or invading property over a period of hours or days, in a manner motivated or patterned by elements of ethnicity, race, religion or xenophobia.Footnote 4
Although this broad definition is useful for comparing events that may at first blush seem unconnected, we limit our study to a subset of episodes that have been largely overlooked in comparative work on state responses to ethnic riots. Our universe of cases is post-1980 intense anti-authority ethnic riots in liberal democracies. The post-1980 and liberal democratic elements are designed to complement studies of pre-1980 riots in the United States.Footnote 5 We selected intense episodes because they are empirically more significant and theoretically more relevant, as their high profile is likely to amplify state responses, thus making it easier to identify patterns across cases.Footnote 6 By anti-authority ethnic riots, we mean those perpetrated by minority groups (or occasionally majority groups) out of frustration at real or perceived disadvantages related at least in part to their ethnic status. Such attacks are not aimed at people from other ethnic groups, but may be carried out against representatives of the state such as the police or against public or private property.Footnote 7 These correspond to what the American riot literature terms ‘black insurgency’ (Fording, Reference Fording2001), ‘black violence’ (Button, Reference Button1978), or ‘urban riots’ (Welch, Reference Welch1975; Isaac and Kelly, Reference Isaac and Kelly1981) or what are popularly referred to as ‘race riots’.
Our dependent variable is the character of state responses to ethnic riots. We focus on two types of responses at the heart of most of the American scholarship on riots: repression and accommodation, with each categorized as high, medium, or low in keeping with the prevailing practice of categorizing state responses along a continuum (McPhail and McCarthy, Reference McPhail and McCarthy2005: 3–4). We define repression as statements, proposals, or policies that castigate, punish or are otherwise detrimental to the interests of rioters and their community.Footnote 8 Accommodation is defined as statements, proposals, or policies that express sympathy or understanding for rioters or that offer rioters and their community tangible economic, social, or political benefits.Footnote 9
In the context of post-1980 liberal democracies, high repression includes recourse to multiple examples of the following: the sustained deployment of extraordinary policing and judicial powers (such as the use of rubber bullets, army units, or accelerated deportation procedures); the passage of legislation suspending normal legal procedures; enforcing maximum penalties for rioters or instituting increased penalties for riot-related crimes; funding additional units of the police or the civil service designed to monitor and suppress potential rioters; and one-sided public condemnation of the rioters by the overwhelming majority of government leaders or spokespersons.Footnote 10 Low repression entails the modest use of one or two of these strategies, such as some temperate condemnation of the rioters by a minority of government officials and the temporary deployment of additional police units during the riot episode. Medium repression involves either the extensive use of one significant repressive measure coupled with the modest use of others, or the use of a wider variety of repressive strategies but at a level not viewed as extraordinary. Examples include massive saturation policing without brutality, or a combination of steps such as increased surveillance of troubled zones, longer term investments in beefing up police forces, quick arrests and more aggressive prosecutions, and rhetoric that apportions significant blame for events to local residents.
High accommodation consists of multiple examples of the following: government allocation of substantial additional financial resources to welfare programs such as urban regeneration schemes, youth employment initiatives, or public housing; enhancing anti-discrimination laws, instituting affirmative action policies, or easing immigration and naturalization laws in the case of minority riots; prosecuting the police or racists for their participation in the riots; governmental outreach to representatives of the community associated with the riots; organizing government ministries or standing committees around issues relevant to aiding disadvantaged areas; and one-sided public expressions of understanding for the rioters by the overwhelming majority of government leaders or by spokespersons who lay the blame for the riots on underlying social, economic, or other aggravating factors. Low accommodation is defined by modest use of one or two of these strategies, such as proposing but not enacting changes in welfare programs, or infrequent acknowledgements by government officials that underlying factors may have contributed to the lawlessness of the riots. Medium accommodation involves either the significant use of one accommodating measure with the modest use of others, or using a wider range of accommodating strategies in a manner not deemed extraordinary. Examples include quietly allocating tens or hundreds of millions of dollars to urban regeneration funds with few other public signals of support, or a combination of steps such as organizing discussions about how to improve community–police relations, launching small or medium-scale programs to improve local employment prospects, and a few prominent officials suggesting that the blame for the riots can be laid primarily at the door of broad social or economic disadvantages.
Theoretical perspectives and hypotheses
The literatures on riots in the United States, on riots in the developing world, and on protest policing in Europe point to six distinct theoretical perspectives on our dependent variable. In developing these perspectives, we focus on the core of a particular school of thought in order to derive testable, falsifiable, mutually exclusive hypotheses. These perspectives serve as a foundation for analyzing cases and for future theorizing and empirical work. Three of the six perspectives hypothesize consistent patterns of outcomes across all riot episodes. Two hypothesize that a specific variable determines the nature of outcomes. And the final perspective is the least brittle as it suggests that outcomes will depend on contextual events that occur on a case-by-case basis.
Consistent pattern perspectives
The Weberian perspective is based on the assumption that states respond negatively to actors that challenge their monopoly of the legitimate use of physical force within a given territory. This is particularly true, according to theorists of social movements and political violence, when actors use significant force (Gurr, Reference Gurr1970; Button, Reference Button1978; Tilly, Reference Tilly1978; Davenport, Reference Davenport2000).Footnote 11 Therefore, according to this perspective:
Hypothesis 1: State responses will be primarily repressive, with little accommodation.
By contrast, the neopluralist perspective summarized by Fording (Reference Fording2001) draws on a variety of sources from agenda-setting to social movement literatures to suggest that marginalized groups that have difficulty mobilizing support through conventional means may turn to violence as a way to raise their profile among policymakers. According to this school of thought, once the marginalized groups have turned to violence, the state has an incentive to take notice and to respond to those groups (and their allies) with relatively favorable policies in order to undermine the potential for further disorder. Thus, from this perspective, we derive:
Hypothesis 2: State responses will be primarily accommodating, with little repression.
The social control perspective views the state as comprised of an elite class that wishes to maintain control over the territory by all means available (Piven and Cloward, Reference Piven and Cloward1971; Body-Gendrot, Reference Body-Gendrot2000; Fording, Reference Fording2001). It is therefore likely to apply ‘coercive control’ (repression), but is also able to tap into state resources to exercise ‘beneficent control’ (accommodation). Piven and Cloward (Reference Piven and Cloward1971) interpret the expansion of US welfare rolls in the 1960s as a means of responding to urban civil disorder, and Fording (Reference Fording2001) has also found support for the social control theory through an examination of welfare and criminal justice policies that emerged in the 1960s and 1970s. If this perspective is correct:
Hypothesis 3: State responses will entail an even balance of repression and accommodation.
Specific variable perspectives
The electoral incentives perspective stresses that partisan political calculations about likely voting patterns of rioters, their supporters, and their opponents will determine how leaders utilize state resources to repress or accommodate rioters.Footnote 12 Wilkinson (Reference Wilkinson2004: 6) has deployed this perspective in analyzing state mobilization to prevent group-on-group violence. Focusing on anti-civilian riots in India, he argues that ‘democratic states protect minorities when it is in their governments’ electoral interest to do so’. Studies of advanced democracies have also found a significant partisan effect on state responses, both in quantitative studies (Fording, Reference Fording2001) and in qualitative ones (della Porta, Reference della Porta2006: 74), although both emphasize this as one variable within a complex web of factors.Footnote 13 Accordingly, we examine one general and one specific proposition:
Hypothesis 4: State responses will be primarily accommodating with little repression when rioters or their sympathizers are aligned with the party in power, and primarily repressive with little accommodation when they are not.
Hypothesis 4a: State responses to ethnic minority rioters will be primarily accommodating with little repression when the Left is in power, and primarily repressive with little accommodation when the Right is in power.Footnote 14
National structures perspectives draw on insights akin to those from the literatures on policy styles, new institutionalism, path dependency, and political culture, to suggest that the national context (of policymaking, institutions, culture, etc.) is likely to be the most significant factor determining state responses to riots. This theoretical perspective is substantially weaker than the preceding ones in terms of generating a unified set of falsifiable hypotheses, as there are multiple dimensions of national differences, which produce contradictory hypotheses. Because, we wish to examine testable theories, we utilize the insights of prominent scholars of West European social movements who highlight the fundamental difference between states where the ‘informal procedures and prevailing strategies with respect to challengers are either exclusive (repressive, confrontational, and polarizing) or integrative (facilitative, cooperative, and assimilative)’ (Kriesi et al., Reference Kriesi, Koopmans, Duyvendak and Giugni1995: 33–7). In the context of their study on Western Europe, the authors identify France, Germany, and Italy as exclusive states, and Britain, the Scandinavian countries, the Netherlands, and Switzerland as integrative. Consequently, we test the following:
Hypothesis 5: State responses will be primarily repressive with little accommodation in exclusive states (such as France) and primarily accommodating with little repression in integrative states (such as Britain).
A contextual perspective
Finally, the feedback perspective relies on the assumption that no two riots will produce exactly the same response. Utilizing observations from social movement theory, many authors have focused on the on-going interactions between protesters and the state in influencing the patterns of specific reactions in any given instance (Tarrow, Reference Tarrow1998; Oliver and Myers, Reference Oliver and Myers2003; McPhail and McCarthy, Reference McPhail and McCarthy2005; della Porta, Reference della Porta2006). Unfortunately, this perspective yields no concrete testable hypotheses about state reactions, other than that they are likely to fluctuate from case to case and that their trajectory will be logically discernable from the context of each particular riot. Although we cannot provide a conclusive test of this perspective, it gains credence in the absence of consistent patterns across incidents and loses plausibility if there are common outcomes over significant numbers of cases.
Methodology, case selection, and evidence from the cases
We employ a methodology of controlled comparison of four carefully selected cases that allow us to assess the theoretical perspectives and their hypotheses. In doing so, we contribute a middle-range study, which Abu-Lughod (Reference Abu-Lughod2007: 9) has praised as ‘benefitting from both the hypotheses suggested in the large-scale comparisons and the details available in the individual case studies’.Footnote 15 Our information sources include local, national and international newspapers, official documents and reports, participants’ journals and memoires, and scholarly sources. We use these in combination to ensure balance and comprehensiveness of coverage, privileging verifiable facts over interpretation and analysis in order to assemble a systematic record of state responses to riot episodes.Footnote 16
As examples of post-1980 intense anti-authority ethnic riots in liberal democracies we selected riots that occurred in or around London (1981), Bradford (2001), Lyon (1990) and Paris (2005).Footnote 17 Drawing on cases from Britain and France holds relatively constant certain important structural features, such as population size, presence of a significant colonial history and post-colonial immigration, socio-economic development, and membership in the European Union. The four individual episodes are each among the most intense anti-authority ethnic riots in each country,Footnote 18 thus generating clear state responses and allowing us to test hypotheses 1 to 3, which posit consistent outcomes across cases. In addition, our episodes provide variation across the dimensions required to test hypotheses 4 and 5, namely party in power and type of national structures. Rightist governments were in national office in the London and Paris cases and Leftist governments were in power during the Bradford and Lyon riots,Footnote 19 and Britain represents an integrative state, whereas France is an exclusionary one. Finally, selecting episodes across 15 or 20 year time periods within each country and examining state responses to each riot episode over multiple months allows us to investigate whether feedback processes have a systematic impact on state responses to ethnic riots in these cases.
In what follows, we briefly summarize the riot context before highlighting the most significant repressive and accommodating state responses.
London (Brixton) 1981
At the time of the 1981 riots, the London neighborhood of Brixton was known for its widespread social and economic problems as well as for its large West Indian community. Unemployment in the area was severe, and among black youth it was estimated at 55% (Scarman, Reference Scarman1986: 27). Police-community relations in Brixton were quite strained; under the so-called ‘sus’ law, the police were able to question and even jail a ‘suspected person…with intent to commit an arrestable offence’.Footnote 20 Moreover, separate ‘stop and search’ provisions allowed police to detain anyone suspected of possessing stolen goods.Footnote 21 This rather vague law lent itself to abuse, and reports show that more blacks were arrested under this law than whites (Bridges and Bunyan, Reference Bridges and Bunyan1983: 87).Footnote 22
It was in this context that the Brixton riots started. During the week preceding the disorders, the Metropolitan Police Service launched a plainclothes operation called Swamp ’81, which boosted their presence in an effort to curb robbery and burglary (Scarman, Reference Scarman1986: 94–6). Almost 1000 people were stopped by the police, but fewer than 150 cases ended in an arrest, leaving hundreds of community members ‘angry and humiliated’.Footnote 23 On 10 April 1981, as policemen were trying to help a black youth who had been stabbed in the back, a crowd started to build up around the officers. Soon community members and police were facing off with the tensions erupting into riot proportions. The disorders lasted for 3 days and primarily involved youths of West Indian origin. The rioters, armed with stones and petrol bombs, overturned and set ablaze cars, looted shops, and damaged or destroyed buildings in the area. Approximately 400 police officers and at least 45 civilians were injured over the duration of the riots (Scarman, Reference Scarman1986: 38–72). The Brixton riots were preceded by similar encounters in Bristol (April 1980), and followed in July 1981 by a wave of outbreaks in London/Southall, Liverpool/Toxteth, Manchester/Moss Side, Birmingham/Handsworth, and in a number of other cities, towns, and boroughs across the country, including Brixton for the second time that year.
The initial response to the April 1981 riots was almost purely repressive. During the events, the police responded forcefully, confronting the rioters rather than containing them. The second day of disturbances, for instance, started after the police decided to continue with operation SWAMP ’81, which reignited the tensions between police and locals. By the end of the third day, 2500 police had been sent to the area to quell the disturbances (Scarman, Reference Scarman1986: 38–72). In the immediate aftermath of the riots, the government responded severely as well. Although community leaders argued that economic depression and hostility toward the police were the underlying causes of the riots,Footnote 24 Prime Minister Thatcher and Home Secretary Whitelaw insisted that the unrest was purely a law-and-order issue. Mrs Thatcher deflected calls for increased social spending,Footnote 25 declaring that unemployment was not the ‘primary cause’ of the riots and that ‘money cannot buy trust and racial harmony…. They [the riots] were criminal’.Footnote 26
In a potentially accommodating step, the government immediately launched an official inquiry into the disturbances. On April 14, Home Secretary Whitelaw appointed the liberal Lord Scarman to ‘inquire urgently into the serious disorder in Brixton on 10 to 12 April 1981 and to report, with the power to make recommendations’. Although in some respects this was an exercise in deflection, its impetus and focus were directly linked to concerns about racist policing raised by ethnic minorities and their allies (Scarman, Reference Scarman1986; Thatcher, Reference Thatcher1993: 144), and the choice of the well-respected Lord Scarman indicated that the government was not simply looking for a sympathetic report.Footnote 27
With the surge of new disturbances in July, the government’s medium and long-term response to the Brixton riots became intimately intertwined with these events. Prime Minister Thatcher defended the performance of the police during the July riots and considered extending their powers through a Riot Act, making it a crime to be in the vicinity of a disturbance after a police warning is issued.Footnote 28 The Home Secretary signed an order on July 10th – the same day rioting broke out in Brixton – banning marches and demonstrations throughout London for a month.Footnote 29 The government also emphasized two general policies in response to the riots: first, they insisted on the need for better equipment for the police and second, instead of developing programs to address the needs of the inner-city, they urged more parental restraint in the areas where the disturbances had taken place.Footnote 30 Eventually, the former proposal materialized: the government allowed the traditionally unarmed British police to use plastic bullets, armored vehicles and water cannons to quell disorders in what was seen as a severe counteroffensive against the rioters.Footnote 31
In a much more accommodating manner, government officials admitted for the first time in early July that complex social and economic forces lay behind the riots. Home Secretary Whitelaw acknowledged in Parliament that there was an ‘underlying strand of racial difficulties in some of the riots’, and that many of the young rioters were from inner-city areas, which suffered from a range of disadvantages, including serious unemployment.Footnote 32 According to one anonymous Cabinet member whose 1981 diary was published in The Economist, even Prime Minister Thatcher was ‘rattled’, fearing that ‘her policies are causing bloodshed on the streets’.Footnote 33 In a more concrete symbolic gesture, the Secretary of State for the Environment Michael Heseltine was dispatched to Liverpool, where he spent over 2 weeks meeting with local residents and getting a better sense of the challenges of depressed inner-city areas (Heseltine, Reference Heseltine2000: 216–24).
The Conservative government took one major and one minor accommodating measure following the riots of 1981. Most significantly in light of its firm commitment to monetarist austerity, it made funds available for programs aimed at inner-city neighborhoods. The Prime Minister announced a raft of measures in late July that included hundreds of millions of British pounds for launching or expanding programs such as youth training schemes, early retirement provisions (to open up jobs for young people), and subsidies to firms hiring young employees, among others.Footnote 34 In addition to this concrete support, Environment Secretary Heseltine boasted that urban funding had more than doubled between 1980 and 1983, all in an era of decreased budgets (Heseltine, Reference Heseltine1983: 12–13), a finding consistent with those of independent researchers (Glazer and Young, Reference Glazer and Young1983: 19–20; John, Reference John2006). Brixton benefited immediately from the attention, seeing its central government funding rise from £9.75 million to £12 million in the fiscal year following the riots.Footnote 35
A second set of accommodating measures revolving around police powers and procedures unfolded over the longer term. In late 1982 the government published its Police and Criminal Evidence Bill, which contained a number of elements suggested in the Scarman report on the Brixton disturbances, such as clarifying the powers of the ‘stop and search’ provisions of the law, a new and improved mechanism for filing and processing complaints against the police, and statutory obligations for police-community consultations (Scarman, Reference Scarman1986; Joly, Reference Joly2007: 111–112).Footnote 36 It is important not to overstate the accommodation embedded in these steps, however, as the 1982 Police and Criminal Evidence Bill was criticized as broadly restrictive (Bridges and Bunyan, Reference Bridges and Bunyan1983) and as many observers were disappointed by the government’s tepid application of most of the Scarman recommendations (Benyon, Reference Benyon1984).
In sum, the April riots in Brixton and the follow-up riots in July generated high repression in the form of continued saturation policing, the deployment of military equipment, a blanket month-long ban on demonstrations, and weeks of strong rhetoric condemning rioters and casting the problem as uniquely one of law and order. The riots also resulted in medium accommodation as measured by the eventual release of funds for inner-city areas, a few new checks on police powers, and the symbolic 2-week presence of a Minister in a troubled area. The accommodating response was more significant than frequently acknowledged (especially after July), but it did not rise to the level of high accommodation given the government’s failure to implement most of the reforms proposed by the inquiry it commissioned.
Bradford 2001
The city of Bradford has one of the lowest shares of white population of Britain’s three dozen metropolitan districts, with over 25% of its residents coming from visible minorities (Great Britain: Office for National Statistics, 2004: 91). Bradford has also been nationally prominent as a region that embodies tensions involving the Muslim community, including the Honeyford affair of the mid-1980s (in which a white school headmaster was forced to resign for publishing articles critical of Pakistani people), the 1989 burning of Salman Rushdie’s book The Satanic Verses, and widespread protests against Britain’s participation in the first Gulf War.
The 2001 Bradford riots began on July 7th during a demonstration held by the Anti-Nazi league to protest the far right National Front’s plan to march in the city (Hussain and Bagguley, Reference Hussain and Bagguley2009). Confrontations between far-right militants and local Asian men quickly evolved into an anti-authority riot as the police drove protesters into the largely Asian Manningham section of the city.Footnote 37 That night, between 400 and 500 people battled almost 1000 police officers and destroyed property using bricks, stones, and Molotov cocktails (Bradford University, 2002: 9).Footnote 38 By the end of the riots 3 days later, at least two people and one police horse had been stabbed,Footnote 39 approximately, a dozen other civilians had been hurt, 326 police officers were reported injured and between £7.5 and £27 million of damage had been done (Denham, Reference Denham2001: 7; Hussain and Bagguley, Reference Hussain and Bagguley2009: 71).
As with the Brixton events, the Bradford riots were embedded in a series of proximate disturbances, including a smaller-scale confrontation in April that left eight injured and several shops and cars damaged.Footnote 40 There were also similar outbreaks of violence in the northern cities of Oldham, Leeds, and Burnley in May and June (Thomas, Reference Thomas2009).Footnote 41 These recurring disorders represented a challenge for the Labor government. Prime Minister Tony Blair was reelected on a pledge to control crime and had appointed David Blunkett as his Home Secretary following the June 2001 elections.Footnote 42 The quest to fight criminality and Mr Blunkett’s firm views shaped the government reaction to the subsequent outbreaks.
The initial response to the riots was a combination of repressive rhetoric and accommodating action. Blair condemned the ‘thuggery’ and violence, which had led to protesters ‘destroying their own community’.Footnote 43 Home Secretary Blunkett proclaimed that he would not tolerate ‘the wanton destruction and violence’ of the riots, and there were debates about whether law enforcers should use water cannons, tear gas, and plastic bullets to quell the disturbances.Footnote 44 Yet, in spite of the police calling in reinforcements from neighboring jurisdictions and admitting that they had nearly lost control of the situation, they elected not to use more extreme force because they did not want to provoke further disorder (Bujra and Pearce, Reference Bujra and Pearce2009: 67–68).Footnote 45
Beyond showing restraint, British officials immediately initiated several forms of accommodation. Home Secretary Blunkett publicly condemned rightwing extremists and called on intelligence services to ‘step up their surveillance of email and internet chat-rooms used by far-right extremists to foment trouble’.Footnote 46 In the immediate aftermath of the riots, the government established an inter-departmental Ministerial Group on Public Order and Community Cohesion to make suggestions for reducing the risk of further disorder (Denham, Reference Denham2001: 1). In July, the group agreed to £7 million of support for summer activities for young people, which funded 207 projects involving over 200,000 youths (Denham, Reference Denham2001: 19). It also established a network of community facilitators who could be dispatched to hotspots in danger of descending into violence (Denham, Reference Denham2001: 19). In addition, the government appointed a second task force, the Community Cohesion Review Team, led by Ted Cantle, designed to canvass ideas from local communities about how to bring about social cohesion in ethnically divided neighborhoods (Cantle, Reference Cantle2001).
The December 2001 publication of these two groups’ reports largely shaped the long term response to the riots, which focused on the common catchphrase of building ‘cohesive communities’ as a solution to segregation and social unrest (Cantle, Reference Cantle2001; Denham, Reference Denham2001; Thomas, Reference Thomas2009). Through a combination of accommodating policies – such as social programs aimed at improving housing and tackling unemployment in the inner-city – and repressive policies – namely programs promoting British ways of life that implicitly criticized ethnic minorities for a perceived lack of integration – the community cohesion agenda encouraged individuals to commit to being Bradfordian or British citizens.Footnote 47 Echoing the tenor of these reports, Home Secretary Blunkett declared that ‘if we are going to have social cohesion, we have got to develop a sense of identity and a sense of belonging’ and he called for minorities to adopt British ‘norms of acceptability’.Footnote 48
In response to the riots and to the relevant reports, the Bradford Council developed its September 2001 action plan, its May 2002 outline Community Cohesion Plan and finally its June 2003 District Community Cohesion Delivery Plan.Footnote 49 These initiatives included accommodating measures such as investing in programs to aid business start-ups and place youths in jobs, and more repressive elements such as stepped-up funding for law enforcement officers and for closed-circuit TV cameras to monitor city streets, and an increased curricular focus on citizenship in schools. Although a number of scholars have been skeptical of the community cohesion agenda (Bradford University, 2002; McGhee, Reference McGhee2003; Bagguley and Hussain, Reference Bagguley and Hussain2006), the aspect of overriding importance for our purposes has been its attempt to achieve what the 2005 Bradford review of its policies referred to as ‘an appropriate balance between “law and order” and a “softer” more holistic approach to cohesion’.Footnote 50
Over the medium and longer term, other repressive elements emerged that were directly or substantially linked to the 2001 riots. Over 100 participants in the Bradford riots were eventually convicted for the most serious possible charge and sentenced to long terms in prison.Footnote 51 In spite of complaints by Bradford’s Asian community that these were unfair jail sentences, judges denied their appeals and Home Secretary Blunkett decried that the ‘maniacs who were engaging in this are now whining about sentences they have been given’.Footnote 52 In addition, a February 2002 governmental White Paper outlined a stricter approach to citizenship and naturalization in light of the northern riot reports’ ‘vivid picture of fractured and divided communities, lacking a sense of common values or shared civic identity to unite around’ (Great Britain: Home Office, 2002: 10). This document proposed the introduction of an oath to be sworn at naturalization ceremonies and the toughening of the English language test requirement for citizenship. Both policies eventually materialized: in 2004 a pledge of loyalty became mandatory for those acquiring British citizenshipFootnote 53 and in November 2005 the government launched the new UK citizenship test entitled ‘Life in the UK’.Footnote 54
On the whole, therefore, British authorities responded to the Bradford riots with medium repression as evidenced by its restraint in police tactics offset by stiff sentences and increased surveillance, by modestly intrusive measures aimed at immigrants and minorities such as reforms in schooling and citizenship policy, and by occasional rhetoric denouncing minorities for lack of integration. This was coupled with medium accommodation, demonstrated by policies that delivered modest funding for social services projects in the affected communities, by the enhanced monitoring and condemnation of right-wing extremists, and by a rhetoric that balanced chiding minorities with an injunction to majorities to help build cohesive communities.
Lyon (Vaulx-en-Velin) 1990
In 1990, the Lyon suburb of Vaulx-en-Velin was beset by a mix of demographic and economic challenges. Compared to national averages, it had a surplus of foreigners, of youth, and of the unemployed,Footnote 55 and it was the poorest commune in the department in terms of its per capita income. Yet it was also hailed as a model of urban rehabilitation in the wake of policy initiatives enacted following the 1981 riots in the nearby neighborhood of Les Minguettes (Jobard, Reference Jobard2009). As a result of these policies, brand new facilities had recently been opened in Vaulx-en-Velin, including a climbing wall inaugurated in the week preceding some of the most intense riots of the era.Footnote 56
On October 6th, violence broke out in the neighborhood of Vaulx-en-Velin known as Mas du Taureau. When two young men were knocked off a motorbike by a police car and one was killed, local youth, believing that the police had acted intentionally, burned cars, smashed windows, looted stores, and threw rocks and Molotov cocktails at the police, who in turn used tear gas and truncheon charges to subdue the rioters.Footnote 57 The riots continued for 2 days before calm was restored to the area (Dikeç, Reference Dikeç2007: 71–4). Residents directed much of the blame for the events at the police, accusing them of harassing minorities.Footnote 58
In the immediate aftermath of the riots, leaders on the extreme right claimed that money spent on helping immigrants was wasted. ‘The solution for the immigrants is not integration; it’s sending them back to their country’, said Thierry Derocles, head of the far right Front National party in Vaulx-en-Velin.Footnote 59 Yet very little repressive rhetoric or action came from the Socialist government or from the mainstream right opposition. It took subsequent riots in the Paris suburb of Sartrouville over 5 months later for Prime Minister Rocard to forcefully declare such violence ‘not acceptable’ and announce a long-planned government bill that targeted urban delinquency in public transportation.Footnote 60 The Prime Minister also foreshadowed the establishment of 10 maisons de justice in the vicinity of the troubled neighborhoods in order to help mediate petty conflicts and provide information about law and justice to community residents.Footnote 61 In addition, in response to the events of Vaulx-en-Velin, in late 1990 the French Intelligence Service (Renseignements généraux) created a monitoring system to track urban violence in hotspot neighborhoods (Dikeç, Reference Dikeç2007: 80–7; Bonelli, Reference Bonelli2008: 391–401). Finally, in the wake of the May 1991 disturbances in Mantes-la-Jolie (another Paris suburb), the Minister of the Interior allocated additional police units to the Paris region.Footnote 62
Overall, however, President Mitterrand’s government responded to the Vaulx-en-Velin riots with a great deal of accommodation both by word and by deed (Dikeç, Reference Dikeç2007: 75–80). Most mainstream leaders agreed to redouble national efforts to prevent France’s 4 million immigrants, half of them Arab, from living on the social and economic margins. There was widespread agreement that more should be done to integrate alienated youths living in ‘sensitive neighborhoods’ (Tissot, Reference Tissot2007). In December 1990, the President opened a 2-day conference outside Lyon at which he focused on fighting exclusion to reinvigorate the troubled neighborhoods, announced forthcoming measures to beef up public services, and called upon businesses to hire local youth as a ‘duty of national solidarity’.Footnote 63 The government subsequently appointed a high-profile cabinet-level Minister for Urban Affairs to oversee the diverse strands of the preexisting public bureaucracy (Delarue, Reference Delarue1993: 8; Delebarre, Reference Delebarre1993).
Also in December 1990, the government began planning for two landmark urban affairs laws.Footnote 64 The urban solidarity law (loi de dotation de solidarité urbaine), enacted in May 1991, was designed to transfer upwards of $180 million per year from rich towns to poor areas to help revitalize them (Body-Gendrot and Wihtol de Wenden, Reference Body-Gendrot and Wihtol de Wenden2007: 57).Footnote 65 In addition, Minister Delebarre shepherded the ‘Anti-Ghetto’ law (loi d’orientation pour la ville) through the Parliament to its eventual implementation July 1991.Footnote 66 It aimed to disperse the concentration of low-income, high-density immigrant housing projects that had fed unrest. It encouraged cities of more than 200,000 inhabitants to obtain a social mix by pressuring each commune to have public housing total at least 20% of its stock (Body-Gendrot and Wihtol de Wenden, Reference Body-Gendrot and Wihtol de Wenden2007: 57).
These laws were interspersed with various small-scale youth integration projects, many of which were proposed in the months immediately following the riots. The French government allocated 35 million francsFootnote 67 for suburban youth programs in the summer of 1991 (the summer following not only the October 1990 riots in Vaulx-en-Velin but also the March 1991 events in Sartrouville and the May disturbances in Mantes-la-Jolie). These programs included the development of 500 mini-sports fields in housing projects, trips for 4000 suburban youths to stay in rural families, 300 internships through the Ministry of Defense, 100 artistic workshops and a similar number of concerts funded by the Ministry of Culture, and a tutoring program for youth who were interested in professional internships for the summer. They also included 5 million francs in credits for youth-sponsored local projects. Overall, general funding for summer events was increased by 10 million francs from 1990 levels in the hopes of expanding the number of youth involved from 200,000 to 300,000.Footnote 68
In sum, the October 1990 riots in Vaulx-en-Velin, Lyon provoked low repression as reflected in the minimal levels of police mobilization, the minor initiatives to track and mediate conflict in the most dangerous neighborhoods, and the reluctance of government authorities to condemn the violence. The riots also generated high accommodation that included the launching of a new cabinet post, the passage of two major urban renewal laws, the allocation of millions of dollars for programs affecting tens of thousands of youths, and sympathetic rhetoric from mainstream politicians.
Paris (Clichy-sous-Bois) 2005
The widespread French unrest of late 2005 began in the Paris suburb of Clichy-sous-Bois on October 27th, when two French teenagers of immigrant-origin died while fleeing a police identity check.Footnote 69 Young men of predominantly Arab and African descent reacted by setting fire to over 9000 cars and damaging dozens of schools, day care centers, public buildings, and private businesses over the ensuing days and weeks, causing approximately €200 million of damage and resulting in one death and over 200 injuries (Joly, Reference Joly2007: 262–3).Footnote 70 Thousands of people were arrested and hundreds were sentenced in France’s worst civil unrest since the events of May 1968 (Joly, Reference Joly2007: 263–4; Waddington et al., Reference Waddington, King and Jobard2009b: 5). The incidents spread out from Clichy to the broader Paris region by early November and then quickly to the country as a whole, peaking between November 4th and 8th, and winding down by the third week of the month (Roché, Reference Roché2006). They were focused almost exclusively in the banlieues where most people live in public housing projects, where unemployment rates are among the highest in France, and where there are large concentrations of second and third generation immigrants from North Africa and recent immigrants from sub-Saharan Africa, many of whom see the police as the enemy.
The center right government’s response to the events entailed a mix of repressive and accommodating rhetoric and action. As the riots unfolded, Interior Minister Nicolas Sarkozy declared ‘zero tolerance’ for rioting and threatened to immediately deport over 100 foreigners who had taken part.Footnote 71 His law-and-order sentiments dominated the news and were echoed in the ensuing days by President Chirac and Prime Minister de Villepin, who declared for example that the recent violence was ‘unacceptable’ and ‘inexcusable’.Footnote 72 However, the Minister of the Interior and especially the President and Prime Minister also struck some conciliatory notes by meeting with the victims’ families, pledging an investigation into their deaths,Footnote 73 and calling for respect, dialogue and equal opportunity.Footnote 74
On November 8th, as the violence continued to spread, the government invoked a 1955 law to impose a state of emergency in mainland France, thereby stepping up its repressive policies. Parliament approved a 3-month extension of the law on November 17th, ensuring that local authorities retained the power to impose curfews and ban public meetings wherever necessary (Joly, Reference Joly2007: 264–5).Footnote 75 Paris, Lyon, Marseille, Toulouse, Nice and other cities and suburbs throughout the country made use of one or more of the state of emergency provisions.Footnote 76 In addition, these measures were supplemented by the deployment of thousands of extra police that saturated many of the hottest neighborhoods for the duration of the riots, resulting in thousands of arrests and hundreds of convictions (Joly, Reference Joly2007: 263–4).
Over the medium term, the government continued in a restrictive vein by spotlighting the problem of youth crime. It created an inter-ministerial committee on the prevention of delinquency in January 2006 and introduced a June 2006 bill on the same subject. These steps fit into a trajectory of policymaking that had been underway prior to the riots and that continued through the eventual passage of the law in March 2007 (Dikeç, Reference Dikeç2007: 162–6).Footnote 77 Yet the events of October and November 2005 pushed the process forward, both spurring the administrative and legal developments, and generating pressure on bureaucrats to release resources to cities as early as late November 2005, when Interior Minister Sarkozy announced his plans in a speech to his regional representatives that referred explicitly to the urgency created by recent developments.Footnote 78
As a counterpoint to its most repressive measures, the government also quickly announced steps to accommodate riot-torn communities. In his November 7th speech, the Prime Minister outlined a social program for the suburbs that involved restoring recently cut subsidies to local non-profit associations, lowering the age for apprenticeships from 16 to 14, tripling the number of merit-based scholarships, and speeding up employment consultations and urban renovation in hard-hit areas.Footnote 79 A week later, President Chirac used a nationally televised speech to urge employers and media leaders to fight discrimination and promote minorities, and to announce his plan for a ‘voluntary civil service’ program that aimed to provide training for 50,000 youths in 2007.Footnote 80 On November 21st, the government’s new urban and housing plan – summarily withdrawn on November 9th to be updated – paved the way for fifteen new tax-free urban zones to be added to the 85 existing ones, and increased the overall budget by €181 million to push the total to over €1 billion.Footnote 81
Many of these provisions were inscribed in the January 2006 bill that eventually became the Equal Opportunity Law enacted at the end of March.Footnote 82 Although that legislation contained modest repressive elements (such as a section on the ‘Contract of Parental Responsibility’ and one on the ‘Fight against Incivility’), it was overwhelmingly accommodating. It embodied the President’s proposals for a voluntary civil service, strengthened antidiscrimination institutions, and codified provisions for apprenticeships and tax-free urban zones. Most monumentally, however, it proposed creating a ‘First Employment Contract’ to encourage businesses to take on young workers by relaxing the restrictive rules under which they can be fired during their first 2 years on the job. The Prime Minister claimed that this would make it easier for employers to take a chance on young people who had experienced difficulty in gaining work experience.Footnote 83 However, the prospect of reducing job security and the idea of being treated as a ‘ “throwaway” generation, deprived of the employment protections enjoyed by their parents’,Footnote 84 infuriated many youths, mostly middle class students, and huge demonstrations against the First Employment Contract led to its repeal less than a month after it was enacted.Footnote 85
Once elected President in 2007, Sarkozy appointed several ethnic minority members to his cabinet and recruited the dynamic Fadela Amara as a junior minister for urban policies. He also followed up on a campaign pledge designed to soften his image vis-à-vis disadvantaged groups by announcing a ‘Marshall Plan’ for the suburbs in February 2008. According to most reports, however, the initiative – labeled ‘Espoir Banlieue’ – did not live up to its billing.Footnote 86 Over the ensuing year, it helped promote private sector job recruitment, it supported public transportation projects serving the suburbs, it provided targeted help for young people seeking employment, and it offered a few new initiatives in the field of education.Footnote 87 Yet its scale was not as ambitious as that of a number of previous initiatives, and its goals have been curtailed or delayed at several turns.Footnote 88 As with the First Employment Contract, the state’s delivery of accommodating measures was significant, but it also fell short of its promises.
In sum, the state response to the 2005 Paris riots embodied high repression as measured by the institution of the state of emergency, by the saturation policing of hot-spot neighborhoods, by the creation of committees and the speeding up of resource delivery to fight youth delinquency, and by some particularly sharp rhetoric condemning the riots. It also generated medium accommodation in the form of allocating additional resources to troubled communities, offering young people additional help in getting jobs or internships, and through public statements and policy proposals signaling some sympathy for the plight of people in down-and-out suburban zones. These steps fell short of reaching high accommodation because two key components – the First Employment Contract and the Marshall Plan for the suburbs – were either retracted or failed to materialize.
Findings
Based on the evidence, are there discernible patterns in state responses to ethnic riots? Do some theoretical schools of thought provide more insight than others, or are the outcomes random or context-dependent? Table 1 offers an overview of the character of state responses to facilitate assessment of the theories.
Aggregate levels of repression and accommodation in four riot cases
R, right government in power; L, left government in power.
These findings lend little support to three of the six perspectives outlined above. No episode conformed to the Weberian perspective’s prediction of primarily repressive responses with low accommodation. The neopluralist perspective’s predictions of primarily accommodating responses with low repression held true in one case, but not in the other three cases. National structures perspectives that focus on systematic differences across countries also fail to shed light on outcomes here. According to hypothesis 5, state responses will be repressive with low accommodation in exclusive states (such as France) and accommodating with low repression in integrative states (such as Britain). This was not the case in any episode. Moreover, France employed slightly higher accommodation and slightly lower repression than Britain on average, and the only episode with low repression was in France, not Britain. There are no other systematic differences in state responses across the two countries to support a parsimonious national structures argument, as outcomes differed across time within each country.
Although no single perspective is fully confirmed by these findings, a combination of the social control and electoral incentives perspectives does provide significant leverage for understanding outcomes in these cases. In keeping with the social control hypothesis, states have tended to combine both repression and accommodation in response to large-scale riot events rather than relying primarily on one strategy and excluding the other. In one of the four cases (Bradford), there was an equal balance between the two strategies, and only one case had a significant imbalance (Lyon). As befits the electoral incentives perspective, where the character of state responses was not identical, the Right was more repressive (London and Paris) and the Left was more accommodating (Lyon). As should be apparent, neither perspective alone is sufficient to account for the pattern of state responses. There was not an even balance of repression and accommodation across cases. And the Right proved capable of deploying considerable amounts of accommodation toward ethnic minority rioters and their sympathizers, whereas at least in one case the Left showed itself willing to exercise substantial repression.
Delving more deeply into the political context in these four cases provides additional support for the social control/electoral incentives model. The incidents where the Left was relatively more accommodating (Lyon 1990) and where the Right was relatively more repressive (Paris 2005, London 1981) were also the cases where key party leaders were least concerned with appealing to voters in the middle of the political spectrum. The Mitterrand/Rocard government of 1990 was 3 years away from the next legislative elections, which opened up space for it to pursue a Left-leaning policy. In 2002, the Front National had surprised everyone by defeating the center-Left candidate in the first round fight for the French presidency, leaving the center-Right covering its right flank in preparation for the spring 2007 elections. In Britain, Margaret Thatcher was just 2 years into her first term in 1981 as an avowed ‘conviction politician’ of the Right. Where the responses were most balanced (Bradford 2001), the Labor Party had just won an election on an ideologically centrist, law-and-order platform, which limited its margin to enact traditional Left policies.Footnote 89 In these cases, therefore, electoral incentives and party positions were important in shaping the balance between repression and accommodation even beyond the straightforward calculations of Left vs. Right in power.
As indicated above, it is not possible here to assess fully the feedback perspective on state responses to ethnic riots. There is some indication it mattered in the London case, as the short-term response in the late spring of 1981 was almost purely repressive, and it was the intensity and recurrence of riots in other venues that prompted the Thatcher government to moderate its position. It is also plausible that feedback patterns mattered over the long-term in France, where leaders may have decided between 1990 and 2005 that primarily accommodating responses were not effective. However, the transition to a Right government over that time period and the changing context of the far-right challenge are almost certainly more influential variables. There is little indication that feedback mechanisms were significant in shaping outcomes in the other riots, where the character of state responses was relatively stable over time.
Conclusions
This article makes several contributions to the study of state responses to ethnic riots. It conceptualizes the term ‘ethnic riot’ in order to define a class of common events that can be studied comparatively across liberal democracies. It draws together insights from disconnected literatures on riots in the United States, riots in the developing world, and violent social movements in Europe to develop six perspectives on state responses to ethnic riots. It then uses these theoretical frameworks to distill five testable, falsifiable and mutually exclusive hypotheses.
We find that a combination of the social control perspective and the electoral incentives perspective offers a convincing explanation of outcomes in our cases, whereas the other perspectives fare poorly as systematic explanations of state responses to ethnic riots. Our social control/electoral incentives model builds on findings such as Fording’s (Reference Fording2001) – which emphasizes the social control perspective – and Wilkinson’s (Reference Wilkinson2004) – which stresses the importance of electoral incentives. We demonstrate the significance of these perspectives in a greater number of cases and in new empirical terrain, as until now neither has been tested and proven useful in contemporary Western Europe. We also advance scholarship by arguing that each individual perspective is incomplete when considered in isolation from the other. We suggest that a combined social control/electoral incentives model offers a much more powerful explanation of state responses to ethnic riots than either alone, or than has previously been considered.
Our four cases were carefully selected to provide controlled variation among post-1980 intense anti-authority ethnic riots in liberal democracies. To increase confidence in our findings, the next step is to see if they can be replicated in a broader sample from within our universe of cases. Beyond increasing the number of cases, conducting even more fine-grained process tracing will increase confidence that the causal mechanisms that we have identified operate in the manner suggested by the theories. If the model developed here proves powerful in explaining outcomes in a wider variety of instances and in identifying precise causal mechanisms, it is worth probing its applicability to episodes with a family resemblance to those in our universe of cases, such as pre-1980 episodes, lower intensity events, anti-civilian riots, or those that lack an ethnic dimension or that take place in non-liberal democracies. The model might also prove useful for explaining state responses to other forms of extreme violence perpetrated by organized crime networks, terrorist cells, political extremists, or militias. We view this article as a step in the long-term research agenda of scholars seeking to understand how democracies respond to ethnic riots, to ethnic violence, and to violence in general.
Finally, as suggested in the introduction, this study also sheds light on the workings of the contemporary state. Our results clearly show that the state is much more than a Weberian entity that claims the monopoly of the legitimate use of physical force within a given territory. It is an institution that has multiple tools at its disposal for responding to violent actors. The stick of law enforcement was prominent during the riots examined here, but so was the carrot of the welfare state, as was the mixed political rhetoric of condemning or sympathizing with rioters. Contrary to the vision of the state as a bureaucracy animated by path dependent standard operating procedures, our findings emphasize the importance of political leaders driven by partisan electoral incentives in choosing the relative balance between repression and accommodation. In advanced liberal democracies, this dynamic means that states have frequently accommodated rioters and their communities in the aftermath of the events. This raises questions for social scientists about whether more organized forms of violence would elicit similar responses by states, and whether, how, and under what circumstances groups can mobilize violence strategically to attain their goals.Footnote 90 For citizens, it may also spark debate about whether it is most effective for elected leaders to respond to riots with accommodation, repression, or a mix of both, and about what impact different political strategies have on the likelihood of further violence down the road.
Acknowledgements
For helpful comments on this article or on earlier versions of this project, the authors thank Christophe Bertossi, Sophie Body-Gendrot, Rafaela Dancygier, Donatella della Porta, Mustafa Dikeç, Richard Fording, Romain Garbaye, Donald Horowitz, Roger Karapin, Ruud Koopmans, Daniel Myers, Robert Pekkanen, Sebastian Roché, Sanford Schram, John Skrentny, Sidney Tarrow, Charles Tilly, Sylvie Tissot, Loïc Wacquant, Steven Wilkinson, and the editors and anonymous reviewers of the European Political Science Review. The lead author also thanks Liz Campbell, Ryan Hart, and Ioana Literat for research assistance, and Middlebury College for grants enabling the writing of this article.
