‘Inappropriate but not crime’? Policing racial hatred in Sweden

ABSTRACT Racial bias afflicts police practices across the globe. Police discrimination against and mistreatment of racial and ethnic minorities is indeed difficult to underestimate. While much attention has been thus paid to racially biased policing, fewer studies examine the question from the reverse angle, namely how the police themselves combat racist offences. This article offers empirical insights into the policing of racial hatred in Sweden, a relevant yet relatively understudied case. Drawing on interviews with police officers and crime investigators, I discuss law enforcement perspectives, e.g. perceptions and reasoning in relation to the investigation of racist offences. Findings evince a rather narrow approach as regards the constructions of racist motive that involves a relatively restricted use of bias labelling in identifying hate incidents, especially when the boundaries of racial hostility are perceived as blurred. I argue that while such an approach may reflect a legitimate effort to demonstrate the existence of a motive behind an offence, it may also lead to an underestimation of more mundane forms of racism and their harms inflicted upon racialized individuals and communities. The results have implications for ‘recognition’ and ‘belonging’ as benchmarks of democratic policing, and ‘the promise of inclusion’ associated with combatting hate crimes.


Introduction
Much research on police relations with racial and ethnic minorities has noted the pervasive nature of racial bias and exclusion (e.g. Rice & White, 2010). Policing is informed by negative bias against non-white communities in most western democracies, which may be due to both institutional predispositions or implicit, 'cognitive machinations' (Trinkner & Goff, 2016). Scholars have suggested a number of different ways to combat racially biased policing (see Harcourt, 2010, on randomization; White, 2010, on improved accountability mechanisms; and Trinkner & Goff, 2016 on deconstructing racial stereotypes). The evidence on how these proposals might be implemented is not exhaustive. If racial bias is ingrained in police practices, it is also necessary to understand how the police themselves deal with racially biased forms of offending. Addressing this question is an Policing, race and the spectre of exclusion Historically, the policing of racial and ethnic minorities has been plagued by biased practices. The literature is vast and there are hardly any studies that do not critically or perhaps somewhat apologetically confront racially biased policing (Engel, 2010). Admittedly, it is not bias per se but its underlying causes that discussions revolve around. Many view bias as a euphemism for discrimination, profiling, intimidation or institutional racism that cannot be reduced to individual factors (e.g. Bourne, 2001). Others contend that bias is a necessary evil which has its roots in police officers' 'working personality' (Skolnick, 2010) or in the social-psychology of stereotyping (Smith & Alpert, 2007).
Recent debates on the rise of predictive policing technologies have added yet another layer to this debate. Facilitated by the ubiquitous surveillance of human actions and transactions, policing by prediction builds on actuarial methods and the analysis of 'big data' (Ferguson, 2017). The profiling of potential offenders is vindicated by means of a premise of objectivity, since it is argued that the metrics of multifaceted data would counteract, or else provide empirical justification for, the disproportionate targeting of particularnon-whitepopulations. One of the more vocal critics of predictive policing, Harcourt (2008), calls this a false premise. Selective targeting, he maintains, will not reduce crime and may even increase it since it is impossible to know for certain the 'relative responsiveness of profiled individuals' to these new policing techniques. In defence of prediction, Margalioth (2008) has argued that 'profiling is (nearly) always efficient in theory' but maybe distorted by 'misguided use' in practice.
In this light, it is tempting to extend Neocleous' (2000) conception of policing as an instrument of class domination to policing as an instrument of racial domination. However, while scholars have long directed a spotlight at racial bias in police practices (see Davenport, Soule, & Armstrong, 2011, on protest policing see; Muniz, 2015, on community policing; White & Fradella, 2016, on stop-and-frisk), little attention has been directed at the question from the reverse angle, namely, how the police themselves deal with racial bias in criminal behaviour. The policing of racist offences has become increasingly salient at a time when right-wing populism, with its emphasis on the exploitation of anti-immigration sentiments, is a getting a firmer hold both in Europe and beyond. Loader's (2006) assertion that the police are 'both minders and reminders of community' thus deserves to be reconsidered in the context of racially motivated crimes.
Loader builds his argument on a critique of what he calls 'ambient policing' strategiesneighbourhood-, quality of life-, community-policing and so onwhich practitioners and scholars alike tend to embrace as pathways to police-citizen partnerships. One of the problems with these strategies, he claims, is the ready acceptance of majority demands for security as benign concerns, often at the expense of the interests of more proactively policed, racialized communities. Policing cannot be understood, Loader reminds us, without referring to its cultural aspects and the production of social meanings. As a mediator of collective identity, the police may, and often do, 'expressly or implicitly denigrate or misrecognize minorityimmigrant or indigenouspopulations', and send 'powerful exclusion signals, [via] practices that indicate that particular groupsteenage boys, or black males or Muslimsare not to be considered full members of society ' (ibid., pp. 211-212). If anything, a democratic ideal for policing should be anchored in recognition and belonging, that is, in the establishment of 'conditions in which individuals possess an effortless, confident attachment to a political community in ways that enable them to feel routinely at ease with the threats posed by the settings of their everyday lives' (ibid., p. 215). These conditions require that the police 'recognize the security claims of all citizens and respond to them in ways that sustain rather than undermine their sense of belonging' (ibid., p. 217).
Given the cultural work that Loader identifies in police behaviour, there is good reason to examine how police officers reason about and grapple with racist incidents in the context of their investigative work. Are there perhaps 'permanent barriers', as Harkin (2015) has recently labelled the constraints on the potential of democratic policing, to envisioning a law enforcement framework that is informed by recognition and belonging? In the following, I first set out a discussion of hate crime policing as an emblematic case of 'social control designed to protect minority populations' (King, 2007, italics in original). This is a rather unconventional task for police forces since it deals with victimization that is specifically due to the victim's actual or perceived identity. Thus, it provides a good opportunity to examine policing in a context in which racism becomes the object of police work.

Hate crimes and the promise of inclusion
A couple of decades have passed since hate crime entered the dictionary of law enforcement. Most jurisdictions do not define hate crime as a distinct category of offences but instead generally refer to penalty enhancement statutes, to hate crime motives as an aggravating factor in relation to sentencing or to administrative measures such as collecting statistics (King, 2009). While legal definitions vary, hate crime is commonly understood as a criminal act that is partially or wholly motivated by bias or prejudice towards the victim's actual or perceived identity (e.g. Walters, 2014). For some, such a definition necessarily involves a 'norm of sameness' (Jenness, 2003), since it applies regardless of the majority or minority status of the victim. For others, this would involve turning a blind eye to the structural inequalities that make hate crime victimization an asymmetrical phenomenon (Bourne, 2002). Evidence suggests that the bulk of hate crime victims are members of minority populations, and the most frequent targets tend to be members of racial and ethnic minorities. The emergence of the hate crime concept and its gradual adoption into law also owes a great deal to social movements and civil rights activism focused on counteracting the violent intimidation of historically subjugated groups based on their skin colour, religious faith, sexual orientation, disability and so forth (Jenness & Grattet, 2001).
The peculiarity of hate crime law lies in that it 'attempts to reverse the exclusive logic of punishment by exploiting punitiveness towards the promise of social inclusion rather than exclusion' (Mason, 2014, p. 296, quoted in Russell, 2017italics in original). Putting hate crime into the context of punishment theory, Russell (2017) claims that hate crime law can be considered as a deviation from the symbolism of traditional criminal law, which seeks to 'affirm existing hierarchies of legitimacy and social order'. What the penal construction of hate crime does, she argues, is to '"turn the law against itself", in the sense that the violence of law is called upon to protect those groups that have historically been subjugated or invisibilised by it' (ibid., p. 319).
However, the existing literature on hate crime policing shows that the symbolic content of hate crime law does not neatly produce instrumental effects in the context of its implementation. One of the major issues here pertains to the (in-)capacity and the (lack of) commitment of the police as regards hate crime investigations (Hall, 2012). One common tendency among police departments is to under-prioritize crimes motivated by prejudice. Boyd, Berk, and Hamner (1996, p. 827) have observed that while a completely dismissive attitude was not widely shared, 'many interviewed officers, detectives, and commanding officers expressed resentment over the demands placed on them by the new [hate crime] policy, seeing it as another bureaucratic imposition that takes officers away from their "real" crime-solving duties'. It is also not uncommon for police officers to consider racist violence and intimidation as insignificant or simply 'yobbish', or to associate victim claims with 'ulterior motives' such as seeking additional benefits (Bowling, 1999). Recent analyses have also shown that hate crimes overall are less likely to attract police attention and to result in arrest or clearance compared to non-bias crimes (Lantz, Gladfelter, & Ruback, 2019).
Studies on specialized hate-or bias-crime units present a more nuanced picture. For one thing, the law enforcement response to hate crimes seems to be patterned by organizational and demographic factors. Jenness and Grattet (2005) argue that police departments that prove to be 'pervious', that is, relatively flexible in terms of policy innovation, are more likely to adopt a specific hate crime policy into their repertoires. Moreover, the policing of bias crimes can itself be racially biased. King (2007) has shown that U.S. police agencies are less likely to comply with the Hate Crime Statistics Act in regions with a larger black population, even though this association is confined only to the southern states. King concludes that hate crime policing is 'not a colour-blind process', since the law is 'possibly enforced the least where they are at highest risk of occurrence' (ibid., p. 218). Police responses might also be more attentive to certain forms of interracial conflict and victimization (Nelson et al., 2016).
A further point is that even when detectives purportedly follow rigorous guidelines and checklists in order to identify, say, a racist motive, controversies arise in the construction of hate crimes, prosecutions are rare and investigatory outcomes are usually less than satisfactory (Boyd et al., 1996;Mason, McCulloch, & Maher, 2016). In one ethnographic study, Bell (2002) observed that bias-unit detectives in the so-called Centre City experienced both pressure to enforce hate crime laws (by advocacy groups) and pressure not to enforce them (by white working-class residents). Jacobs and Potter (1998), in their work on the NYPD, have argued that police officers' 'unique responsibility for deciding whether particular crimes ought to be labelled "bias related"', 'complicates and contributes to the politicization of police operations'. This may partly explain why police departments tend to assign more resources to hate incidents characterized by extremist attacks or brutality, while less effort is invested in cases of everyday racism, assault and intimidation. Lewis' (2014) critical discussion of the cultural politics of hate crimes, for example captures how devoting excessive attention to 'hyper-violent murders committed by clearly guilty perpetrators who belonged to white power groups' (p. 5) might foster indifference to mundane forms of racial violence, under the guise of tolerance and colour blindness.
The above discussion raises important questions about the policing of racist offences. It seems that such offences are likely to be placed towards the lower end of the hierarchy of police work. Even if this could be refuted, the promise of inclusion is obfuscated by practices of selective enforcement and by the politicized nature of bias labelling. It would however be wrong to understate the notable advancements that have been made in hate crime policing in various countries in the past two decades (Chakraborti, 2014). At the same time, even though progress has been made, further issues may surface, with Mason et al. (2016), for example describing the possible risks of police forces using over-and under-inclusive checklists to identify hate crimes. At any rate, the bulk of the literature builds on observations from the U.S.A., the U.K. and Australia with little input from other contexts. To what extent these findings are generalizable to a broader geographical context thus remains unclear.
In this regard, the Swedish case provides an intriguing example. Although Sweden has an institutionalized police response to hate crime, research on the policing of racism is sparse (for notable exceptions see Holgersson, 2018;Shoultz, 2015). The Swedish population with a non-European background has increased markedly in recent decades, while a number of Sweden's urban districts have become segregated along racial and ethnic lines. Recent reports even point to the notoriety of racial profiling in police stops of nonwhite individuals (Mulinari, 2017). The populist far-right has been expanding its electoral base, and neo-Nazi demonstrations have raised public concerns. Thus the policing of racially motivated hate crimes has become an ever more salient issue in Sweden, not least for police-minority relations.
Against this background, the overarching purpose of this study is to explore some of the prevalent perceptions and patterns of reasoning within the Swedish Police with regard to the investigation of racially motivated hate crimes. I am particularly interested in how racist hate motives are constructed, or dismissed, from a police decision-making perspective. As a secondary aim, I also want to address the question of what kinds of justificatory claims are used in relation to contested issues such as under-enforcement and low prosecution ratesquestions that have commonly been emphasized in previous research (Mason et al., 2017;Perry, 2008).

A brief overview
The concept of hate crime entered the public discourse in Sweden in the 1990s (Tiby, 1999). Legislation on hate crime has a longer history however. Sweden has three reference points regarding hate crime in its criminal code. Crimes that are committed with a motive to aggrieve an individual or group based on race, skin-colour, national or ethnic background, faith, sexual orientation or transgender identity are covered by a penalty enhancement rule (Chapter 29, Section 7) which first came into force in 1994. The second reference point, the offence of agitation against a national or ethnic group (hets mot folkgrupp), dates back to the late 1940s. It was introduced in the wake of the Second World War as a measure against racist and anti-religious propaganda and has subsequently been integrated into the criminal code (Chapter 16, Section 8). Finally, the notion of unlawful discrimination offence (olaga diskriminering) entered the criminal code in 1970. It basically refers to cases in which people who would like to obtain a service at a business facility such as a shop, restaurant or bar are discriminated on the grounds of their identity (Chapter 16, Section 9) (Granström, Mellgren, & Tiby, 2016).
Statistics on hate crimes can be traced back to the early 1990s when they were produced by the Swedish Security Service (Säpo). In 2006 the task was handed over to Brå, which has since published hate crime statistics based on incidents reported to the police and also on data from victimization surveys. Since 2012 the presentation of policereported hate crimes is based on a random sample that includes around 50 percent of a subset of offences 1 reported to the police. The relevant offence reports are then filtered by Brå, and those meeting the definitional criteria for hate crime are selected (BRÅ, 2017, p. 27). Among these, the most common hate crime motive identified is racial prejudice, as is the case in many other jurisdictions. What is striking is the large discrepancy between self-reported exposure to hate crime and the number of hate crimes that are reported to the police. As regards crimes with a racist motive, recent estimates indicate that less than 20 percent of incidents are reported to the police (ibid.). Whilst not being a specifically Swedish phenomenon, this discrepancy can be attributed to factors such as the troubled relationship between minorities and the police, a shared belief that reporting to the police would not make much difference, or shortcomings in police knowledge, education and training, which could possibly increase the likelihood of secondary victimization (Holgersson, 2018;Shoultz, 2015;Wigerfelt, Wigerfelt, & Kiiskinen, 2013).

This study
Since the present work focuses on the police perspective, it draws primarily on interviews with police officers of varying rank and position, and civilian investigators employed within the Swedish Police Authority. The number of participants interviewed for this study is 32. Since individuals are not required to be police officers with a formal police education to work as investigators (utredare) within the Swedish Police, it is useful to make a distinction between civilian investigators and police officers. The latter term denotes personnel who are officially employed as polis, and who undertake various tasks in yttre tjänst (patrolling, emergency response, neighbourhood policing and so forth) or who work on investigations.
In planning the sample of interview subjects, I assumed that interviewing only hate crime investigators would risk missing a substantial amount of information about situations in which an incident might initially has been suspected to constitute a potential hate crime, without then being passed on to specialized hate crime investigators. I therefore decided to interview not only hate crime investigators but also police officers whose work is focused on other tasks, such as patrolling, neighbourhoodliaison and administration. The intention was to provide a basis for developing a holistic perspective in relation to the questions I wanted to explore. One limitation associated with this approach, however, is that it involves interviewing officers who are not well versed in, and have relatively few experiences of, hate crime. Indeed, this turned out to be the case with some interviewees, largely because of my sampling strategy. However, even these interviews were able to provide information about how the Swedish police at large perceive and reason about their work to combat racial hatred. Thus, what may at first seem like a limitation may also be viewed as providing an opportunity in relation to the aims of the current study.
The interviews were conducted at multiple locations in Sweden between June 2017 and October 2018. Initial contacts were established during a personal visit to an exhibition on hate crime at the Police Museum in Stockholm. Further contacts were then established by means of a snowballing strategy. In practical terms, this strategy is associated with both strengths and weaknesses. It nonetheless allowed me to reach investigators who specialize in hate crime in several different parts of Sweden. In sum, I interviewed 12 fulltime or part-time investigators with specialized knowledge on hate crime. As regards police officers who work with other tasks, however, experiences of hate crime varied considerably. This group included some who had first-hand experience, and others whose experience was rather limited or second hand. I established contact with these officers based on my communications with previous interviewees.
At the outset, I intended to conduct interviews on an individual basis. However, the time and workplace constraints experienced by some participants made the use of joint interviews, with two or more colleagues, a more feasible option. Thus, on eight occasions group interviews were held with at least two participants. Among those interviewed, 18 were male and 14 were female. The age of the interviewees ranged between 25 and 64. While some interviewees had started their career in the police force quite recentlywhich was mostly the case among patrolling officersothers had been in the organization for many years. With the exception of two, all interviews took place at police stations. In one of the two remaining cases, the interviewee preferred to meet at a cafe, and in the other, the participant was interviewed by phone.
Participants were informed, verbally and in writing, about the purposes of the research. The interviews were semi-structured and questions were directed at investigating working principles, processes and professional reflections with respect to hate crimes specifically motivated by racism. Key issues covered included the different stages of hate crime investigations, education and training, relations with the victims and the reporting tendencies of complainants. The length of the interviews varied between approximately 15 min and 1 h-and-10 min. With the exception of the phone interview, all interviews were recorded and transcribed by the author. All markers, e.g. names and locations, which might possibly identify the participant, were removed from the material. In the quotations from the interviews presented in the results section, names have been replaced by pseudonyms.
The results presented below are based on a repeated, close reading of the interview material. For the purpose of this article, particular attention has been paid to the constructions of racist motives and to the justificatory claims made in relation to these constructions. Quotes from the interviews have been translated from the Swedish. On a few occasions, I use parentheses to denote an unfinished or unintelligible sentence ([. . .]), and to add my own clarification of expressions ([author's note]) in a given sentence.

Results
Until recently, the police capacity for investigating hate crimes was limited to one specialized unit and a handful of crime investigators in Sweden. In 2014, the Swedish Police was instructed by the government to prioritize and invest more resources in combating hate crime (Polisen, 2017). Two additional units were subsequently established in the West and South police regions, and more investigators were recruited across the country to focus specifically on hate crimes. Specialized units are not only in charge of criminal investigations but also work to develop capacity and conduct liaison tasks. Capacity development involves both educational activities, such as attending university-level courses on topics such as racism and Islamophobia, and the instruction of front-line police officers about hate crime. Liaison tasks involve establishing contacts with civil society groups and organizations and representatives of minority communities, and holding meetings with them mostly on a yearly basis. The Swedish Police have also introduced an interactive, but non-compulsory, educational package on hate crime that is available to all police staff.
Before a reported incident is passed on to hate crime investigators, it has to be marked as a suspected hate crime at some point after the complaint has been registered. This means that Police Contact Centres (PKC) and front-line officers are key actors in relation to the initial identification of hate crime incidents.
As a police officer I can tell you the following. First we take notes about the incident. At the same time, we ask some extra questions [at the crime scene] to find out whether the incident may involve a hate crime. We write everything down. Once we arrive at the police station, we type our notes into the computer. All the information we collected, and our interrogation notes, are included in this document. While drafting the report, a check-box pops up in the digital platform we use. It asks you, 'Is it possibly a hate crime?' You answer by putting a cross in the YES box or the NO box. (Patrol officer Roger) It should be noted that front-line officers' decisions are no guarantee that all such marked cases will automatically be sent to hate crime investigators. The incident may be confirmed, or dismissed, as a probable hate crime case at a local police station or by officers who are specialized in the underlying offence category. It is also possible that a complaint which has not been marked initially, might subsequently be identified as a likely case of hate crime. In other words, there are multiple checkpoints before hate crime investigators begin their inquiries. While this multi-stage process may work as a filtering mechanism, false dismissals at the onset may also result in the jettisoning of a number of cases that would have otherwise been dealt with as hate crimes.

Making hate a crime motive: when is racism (dis-)counted?
Both qualitative and quantitative studies on hate crime policing commonly make the observation that an under-prioritization of hate incidents has been a shared tendency among various police departments in different contexts. Even though remarkable progress has been noted, this tendency does not seem to have entirely disappeared. The level of attention that is (not) focused on hate crimes, as was mentioned earlier, is influenced by factors such as police competency and organizational features, as well as by different types of (interracial) conflict. In Sweden, the vast majority of hate crime reports are written-off either directly or after a preliminary investigation (BRÅ, 2017). While this does not seem to be specific only to hate crimes, it would not be entirely unexpected to find that the perception that 'nothing happens anyway' would produce a negative incentive for victims to file a complaint. Nonetheless, it would be overly hasty to conclude that most cases are dismissed simply because hate incidents are trivialized by the police. There appears to be a particular way of reasoning, however, among the officers and investigators interviewed for this study. Rather than making an explicit effort to downplay hatred, most interviewees tend to adopt a relatively narrow approachsomewhat reminiscent of what Mason et al. (2016) have labelled an under-inclusive definition of hate crimein their constructions of racist motives. The following excerpt from an interview indicates that even in cases where racial prejudice seems self-evident, a link between motive and action needs to be established.
'Why do you think you have been assaulted this way?' we typically ask. Let's say molestation. We want to know what exactly has happened. So you'd get a picture [from the victim's description]: 'Well, I was just standing there. He saw that I was wearing Romani clothes and said: "Aah, you fucking Romani!"' Then we ask: 'Was it a coincidence that you were molested, or was it like . . . of all the other people around, he picked you because you had . . . clothes. Because you are like dark-skinned? [Inaudible] We let them explain. We don't ask many leading questions but try to keep things open. Ehm . . . 'What is it like . . ., do you know this person? How did it happen?' That kind of open-ended questions, to find out why this person was victimized. 'Was it coincidence, by chance, or did he choose you because of who you are?' (Investigator Anna) We might thus infer that the purpose of asking questions about 'coincidence' and 'by chance' is to demonstrate that a racist motive exists, given that there is a criminal offence. Two things would follow from this reasoning. The first, and arguably less controversial, of these is that in the absence of a criminal offence, a racist incident would not be investigated further. The second is that not every offence that is perceived by the victim as a hate crime would be considered so, unless the motive-action link can be establisheda difference from the U.K. model. Further evidence from the interviews indicates the narrow approach that the Swedish police tend to employ in their work against hate crimes involving racism. Below are two quotes which illustrate how the interviewees base their explanations of the complex nature of hate crime investigations on the 'heat of the moment' argument (Walters, 2014, p. 15).
Today you often hear stories like, someone shouts svartskalle 2 at someone. It's not a crime to do so. But admittedly inappropriate. It's not okay to say things like that, and yet it would never lead to a prosecution. But say, someone shouts 'fucking svartskalle' and hits the person. Then yes . . . It can be . . . It is a crime [. . .] Then you ask questions and the victim tells you what happened. 'I was passing this person and he just hit me'. We also ask the suspect about what happened. 'Well, this person was passing and spilled beer on me. I became damn angry. So I hit him, and I happened to use that expression'. (Investigator Mattias) You know the whole issue revolves around aggression. An effect of aggression itself. So that's the difference in this example. You shout 'svartskalle'. That's an effect of aggression [. . .] 'I didn't hit him because he was a svartskalle. Actually, we were just arguing with each other. He spilled beer on me. He happened to spill his beer. So I was like . . . I hit him. At the moment I hit him, I also said this word. But that's not why I . . . [hit him]' Like I said, we have to be able to distinguish between two things here. 'Was it an effect of aggression that you shouted "fucking svartskalle" in that moment? Or was the reason that you hit this person, because he is a svartskalle?' (Investigator Linda) It goes without saying that due to a lack of evidence, witnesses or an identifiable suspect, police might not be able to carry the criminal investigation of a hate incident further. Most interviewees claimed that many cases thus end up being described as 'one individual's testimony being pitted against that of another (ord mot ord). However, even if the underlying offence can be established, the police may subsequently realize that the expressed racial slur cannot be directly linked to the action itself because it has to do with 'something else'. Another interviewee confirmed this: It may be something that at the outset makes you think like: 'This is a hate crime'. Take molestation or insulting behaviour for example. You recognize some racist expressions and say to yourself: 'Well, this one I would describe as a hate crime, or potentially a hate crime'. But then, when you look into the case more closely, it may become . . . Perhaps during the interrogation [. . .] It may turn out that what has been said [. . .] has something to do with a prior issue. Maybe a dispute which is about something completely unrelated. So we can talk about an expression in affect. In such a situation, the [hate] motive would not count even though at the beginning there was so . . . say, such a powerful indication of . . . I mean this expression, this powerful expression which is so unmistakably racist or something similar. In fact, few things are self-evident and clear. What you think at the beginning may change later on. (Investigator Gunilla) To put it another way, if an investigator arrives at the conclusion that there is no necessary link between a racial slur and the offence itself, the hate motive will most likely be dismissed. It is certainly plausible to argue that the boundaries for when racism becomes the driving force behind an offence may often be blurred. However, as far as the interview material is concerned, the police appear likely to exercise discretion in a way that gives more credence to explanations based on banal aggressions, relative to everyday forms of racism which may actually underlie that aggression. Clearly, there may be plenty of cases in which the motive of the offender is relatively unambiguous and leaves little room for doubt. To take one example, the quote below shows that the link between motive and action can also be communicated by means of other markers such as symbols, clothing or simply by 'doing': I read about a case which involved someone who poured milk over a dark-skinned person. A glass of milk. Pondering about such an act you may wonder: 'Okay, what's the point of [using] milk?' For it could simply have been water, but that person chose milk. Just milk, but not water or anything else. So it was very obvious. I mean, the intention of the person who did this thing. The motive behind it, I mean. (Investigator Sara) However, the boundaries of racial prejudice, as well as the lines between when an offence is racially motivated and when it is not, might also be intangible, not least from a police perspective. In fact, what needs to be explored is not only how the police perceive and act in these less straightforward cases. It is equally important to understand what is viewed as difficult or unclear with regard to racist motives and their investigation from a policing point of view. In the following excerpt, an interviewee describes a scenario which can be read as an example of 'white flight' (Kruse, 2005). It is notable that the story is explained in terms of economic rationality rather than racial prejudice or bigotry: Sometimes things are straightforward, but they can also be much more complicated. Suppose you bully a person and say things like . . . Let's say some people start moving into a residential areapeople with an Arab background. Residents of that neighbourhood may actually have nothing against the Arab identity. But if more and more people from a different ethnic background move in, the residents believe, house prices will plummet. It [the neighbourhood] would no longer be attractive. In fact, they [the residents] do not have anything against them [ethnic Arabs] but they just want the [house] prices to remain high. That's why they do such things. Then you may ask yourself: is there really something to dig up here because the issue is not really about being against [an ethnic group]? They just want to keep the house prices high [. . .] The motivation is not to intimidate others because they have another [ethnic identity] [. . .] It may depend on many things which are often difficult to prove. (Investigator Kalle) What is considered an expression of material concern may arguably be viewed as a display of racial hostility, rendering the distinction between a complicated as opposed to a straightforward case contestable. This reasoning squares with the abovementioned accounts that indicate that a racial slur may not be counted as a hate motive if the perpetrator used that 'unmistakably racist' expression 'not because he or she has anything against . . . ' but as a result of aggression or a prior dispute about some other issue. I would argue that this adds yet another layer to what Loader (2006) has articulated as the cultural work of policing. More importantly, in using their discretionary power to draw the lines of racial bias in relation to an offensive behaviourlines which are conceivably permeable, indefinite or obscurethe Swedish police seem to prefer a restrictive use of bias labelling. The narrowness of this approach may prevent the development of an 'inflated image of the problem' (Mason et al., 2016) or the over-stigmatization of offenders as 'haters' (Walters, 2014). However, it might also signal that the concerns of racialized communities about, and their frequent subjection to, everyday racism are not being addressed adequately. How serious it may get if hate-victim claims are not well-recognized was perhaps most notoriously experienced in the 'Pilkington case' in Leicestershire 2007 (for a summary see Walker, 2011). The fact that the bulk of hate incidents reported to the police are dismissed or, where investigated, not labelled as hate crime, may at least partially be a result of this approach.
Justificatory claims: dismissal of cases and low prosecution rates At this point, it would be beneficial to further explore police perceptions about claims of under-enforcement, the dismissal of cases and low prosecution ratesquestions that are commonly raised in the existing research on hate crime policing. It has been noted that 'the need to prove beyond reasonable doubt that the offender was motivated by prejudice' often has a negative effect on the willingness of police officers to label incidents as cases of hate crime, and feeds into reluctance to do so (Hall, 2010). Thus, it is no huge mystery why 'very few offenders accused of a hate offence are ever convicted or sentenced as "hate offenders"' (Walters, 2014, p. 27). A common theme in the interviews corroborates this observation. In the quote below, liaison officer Nils points to the perceived dilemmas of bias-labelling. Another theme that runs through the interviews involves an emphasis on the police's function as law enforcers. While a certain amount of criticism has been directed at the under-policing of hate incidents, a number of interviewees stressed the legal boundaries of police work. Recent debates about the police's stance on neo-Nazi rallies might help to illustrate this point. In particular, street demonstrations by the Nordic Resistance Movement (Nordiska motståndsrörelsen NMR) have raised concerns about the underenforcement of hate crime provisions, specifically those pertaining to the hets mot folkgrupp offence. NMR members usually wear identical clothes, and carry shields and flags with Nazi symbols such as the Tyr rune 3 on their marches. When it comes to the Tyr rune, the Chancellor of Justice has already stated that this symbol alone is not enough to be labelled hets mot folkgrupp. Then it becomes totally meaningless for us to act on . . . because we would never have sufficient grounds for doing so. (Investigator Gunilla) The emphasis on the law-enforcement role of the police may be interpreted as an effort to deflect much of the attention directed at the discretionary powers of the police. 'I don't think people are well-informed about a lot of things like how we and other authorities work, who's responsible for what, who drafts laws and so on', said one investigator. With respect to the neo-Nazi marches, the legal boundaries of police work were often mentioned in relation to law making and policy making, as processes external to the police. 'This is a political decision', one interviewee noted with regard to NMR demonstrations, and added: 'It is up to politicians who have the power to decide on the right to demonstrate, how wide its scope should be and which groups have the right to use it and so on'. Another interviewee stated that it was 'too bad that we should be made to answer for something which is not our decision'.
A third theme focuses on the symbolic virtues of hate crime policing viewed against the backdrop of a very low prosecution rate. For that matter, the number of hate incidents reported to the police in Sweden increased considerably from 2226 in 2000 to 6415 in 2016. At the same time, the clearance rate dropped from 9 percent in 2007 to 4 percent in 2016, and it is hard to confirm any substantial change in the distribution of offences with a hate motive during this period (BRÅ, 2008(BRÅ, , 2017. Many interviewees were aware of the unwillingness shared by racialized victims to file a complaint because doing so would most likely 'not lead anywhere'. As Kalle, an investigating officer, put it: 'But what happens in the end? A lot of people come to realize that nothing has happened. You would say: "I was assaulted, the police came, it led to a trial, and he got acquitted"'. In this regard, Sweden is by no means an outlier. That is probably why police officers and investigators also frame their work to combat hate crime in terms of its symbolic effects. In a follow-up phone interview, Fredrik, a relatively experienced officer, stated that the outcome of an investigation, i.e. whether a case would be dismissed or prosecuted, is not the most important issue to be concerned about. Reporting to the police is meaningful in itself, he insisted, as it can reveal that a particular type of criminality is becoming more common and that something needs to be done about it. His reasoning was echoed by another interviewee: There is yet another thing to think about when it comes to reporting [crimes] to the police. It concerns why people should file a complaint beyond reasons about justice, I mean, reasons related to obtaining redress within the criminal justice process. Reporting has a statistical effect because every complaint that is registered becomes part of the statistics on reported crimes. So even if your complaint eventually gets dismissed, it would still show that we have a problem in our society. (Investigator Gunilla) Ever since the birth of the hate crime concept, much debate has centred on the symbolic effects of legislating against criminal acts motivated by bias or prejudice. Grattet and Jenness (2008) suggest that the reporting of hate incidents to the police could in itself be viewed as one form of instrumental effect. This follows from the assumption that 'hate crime reporting likely reflects commitment by the police to enforce hate crime initiatives' (King, 2008(King, , p. 1370. As far as the Swedish case is concerned, the increase in hate crime reporting in the last two decades might indeed be an indication of growing public sensitivity to hatred. Nonetheless, it would be preferable to maintain a healthy scepticism, until we understand how the presumed symbolic and instrumental effects of policing are experienced by those at the receiving end of law enforcement. Given the perceived dilemmas of bias labelling and the use of discretion, police decisions on reported incidents would be likely to become politicized, as has been noted by Jacobs and Potter (1998). This would pave the way for additional problems for hate crime policing, both when hate crime provisions are enforced and when they are not enforced. However, a recourse to the symbolic virtues of hate crime policing may not always help to alleviate the likely alienation of racialized communities from police decisions, since the perception that complaints 'don't lead anywhere' appears to be so well established that is endorsed even by the police themselves. The question of being 'tough on hate?' which was recently raised by Lewis (2014) thus remains a valid one to address.

Conclusion
One of the major challenges to contemporary policing is that of ensuring the equitable treatment of racial and ethnic minorities. Police continue to have a troubled relationship with non-white communities in western democracies and beyond, and the presence of racially biased over-and under-policing is perhaps the most pervasive, as well as the most contested, phenomenon infecting this relationship in one way or another. The current study has addressed this question by revisiting the presence of racial bias on the other side of the equation. Much research has focused on the causes and consequences of racial bias in police behaviour, while little attention has been paid to situations in which racial bias itself becomes the object of police work. This article provides insights into the possibility of, and the inherent limitations associated with, the 'promise of inclusion' as a distinct feature of hate crime policing.
In Sweden, the police seem to adopt a rather narrow approach which involves a relatively restrictive use of bias labelling for offences that may involve racial hatred. While police officers do not trivialize hate crimes as being insignificant, there is a discernible tendency to refer to actions taken in the 'heat of the moment', or the significance of a prior but unrelated dispute, as explanations for commonly experienced situations that might just as well be considered as manifestations of racial hostility. Arguably, the use of this approach might reflect a legitimate concern regarding the police's duty to demonstrate that the offence, if an offence has been committed, was indeed the result of a hate motive. This might militate against over-policing of hatred, and prevent over-stigmatization of offenders as bigots. But it is also quite likely that more mundane, everyday forms of racism and the harms that these inflict primarily on racialized people will be ignored. By endangering recognition and belonging, and consequently the promise of inclusion, police relations with racial and ethnic minorities will thus be undermined.
As it can also be inferred from the study, justificatory claims around the dismissal of cases and low prosecution rates lay bare the conundrums for hate crime policing. On the basis of the above discussion, it would be possible to formulate a couple of policy implications. First, it would make sense to suggest that a less restrictive, and perhaps more inclusive, approach to what is counted as a racially motivated offence might help (i) to combat incidents of hatred that lie beyond the tip of the iceberg and (ii) to meet some of the pressing concerns experienced by racialized communities about hate crime victimization. Second, improved communication channels with victim complainants would also be beneficial in terms of addressing their questions, demands and frustrations related to criminal investigations and bias labelling.
Since the current study is based primarily on the perceptions of police officers, no inferences can be made about how police decisions and behaviours focused on addressing racial hatred are experienced at the receiving end of this issue in the Swedish context. This has implications for future research in at least two respects. First, the symbolic, let alone the instrumental, effects of police work cannot be taken for granted without developing an understanding of their consequences for those affected by racist victimization. Second, what matters is not only the content of the police decisions made in relation to reported incidents, but also the treatment of victim complainants in their contacts with the police. This dimension has far-reaching implications for both procedural justice and the ideal of democratic policing.