Social capital and co-location: A case study of policing anti-social behaviour

Existing evidence suggests that co-location may be emerging as a preferred model of multi-agency working between the police and a range of partner agencies, yet there is limited evidence available regarding the benefits and challenges of this specific type of initiative. This article draws on an evaluation case study of co-location between the police and a local authority established to improve responses to victims of anti-social behaviour. Co-located officers reported a range of benefits arising from the new arrangements, and there was evidence of deep learning within and across teams. However, by including the experiences of those working outside the co-location more significant challenges became apparent, relating to ongoing relationships between officers and the wider force that we are unaware of from previous research. It is suggested that senior management should pay attention to managing changing relationships that occur in co-location to preserve existing social capital whilst exploiting opportunities arising from newly formed connections.


Introduction
Collaboration between the police and other agencies in the pursuit of effective crime control and community safety is increasingly accepted as mainstream police practice (Higgins et al., 2016), in part as a response to challenges presented by periods of austerity and broader social changes that shift the boundaries of policing beyond crime. As the public sector enters a post-COVID era of potential cuts and restructuring, it is timely to explore some of the organisational changes that may emerge as preferred ways of working. Indeed, policy-drivers are already such that the police are increasingly drawn into myriad multi-agency practices that are often complex, messy arrangements as much dependent on local context as strategic planning. However, while the assumed benefits of partnership working are well rehearsed (see, for example, Berry et al., 2011), and evidence of good practice seemingly endless, less is known about the impact of collaborative practices (Higgins et al., 2016), or indeed whether some types of collaboration are more effective than others (Parker et al., 2018). This article focuses attention on one particular form of collaborative arrangement -namely co-located models of partnership working. In doing so, we draw attention to the inter-and intra-organisational features of co-location, offering lessons for practice in the establishment and initial stages of developing co-location within and beyond the police service.
In order to consider the impact of co-location on the organisations and individuals involved, we draw on theories of organisational social capital that allow us to examine the ways in which the structural, cognitive and relational dimensions of social capital are realised within co-location models. These ideas inform a secondary analysis of interviews conducted as part of an evaluation of a co-located initiative between the police and a local authority in the north of England.
The article adds to an emerging literature about colocation initiatives involving the police by including a much broader range of voices than has been the case to date. We examine data about the experiences of those working within the co-location as well as those who were not part of the co-location. This allows a more rounded, and critical, perspective to emerge regarding the benefits and limitations of co-location than has been reported previously.
We begin by describing the features of co-location before outlining the ways in which this might be understood in relation to a framework for analysing organisational network social capital and policing. After describing the study methods, the findings describe the ways in which the experiences of co-located staff can be understood with reference to aspects of organisational network social capital, identifying benefits and challenges arising from new social relations.

Co-location
The term co-location describes a model of multi-agency collaboration that brings staff from separate organisations into a shared physical space. A key characteristic of colocation is that although resources may be pooled, staff are employed by their original agency (Audit Commission, 1998). As such, co-location allows professional identities to be retained while merging knowledge (O'Neill and McCarthy, 2012). In itself, co-location is simply one of any number of multi-agency collaborative arrangements that might exist, but there is some evidence that it is increasingly seen as a form of 'best practice' in multi-agency working involving the police (Berry et al., 2011;O'Neill, 2014), generating transformative capacity in addressing complex social problems (Crawford and L'Hoiry, 2017) and therefore warrants further attention.
In particular, existing evidence suggests that co-location offers 'added value' to multi-agency working arrangements through the speed with which knowledge can be shared, and action taken, if individuals are in close physical proximity (Crawford and L'Hoiry, 2017;Grace and Coventry, 2010;O'Neill, 2014). The emphasis on informationsharing as a key requisite for managing risk makes any mode of practice that can facilitate this critical in contemporary community safety and crime control arenas. Colocation also supports the development of strong relationships and mutual understanding across agency boundaries between co-located staff, underpinned by mutual trust and respect (Crawford and L'Hoiry, 2017;O'Neill, 2014;Taylor and Bernardi, 2002). These positive outcomes are particularly relevant for the police whose engagement in multi-agency working has been criticised for failing to embrace organisational change (see, for example, McCarthy and O'Neill, 2014). However, this optimistic view of co-location as a route to improved professional practice is not uniformly shared, with issues of accountability (Crawford and L'Hoiry, 2017) and isolationism (Atkinson, 2019) potentially confounding the opportunities for co-location to generate partnership working.
Existing evidence therefore suggests that co-location may be emerging as a preferred model of multi-agency working, but that limitations may exist. Our aim in this article is to explore these elements in more detail than previous work, and in doing so to consider whether a more theoretically informed analysis can yield new insights.

Framework for analysing social capital
Theories of social capital offer one means of making sense of the complexity of relationships within and beyond colocation models of multi-agency working doing so, where they draw attention to those 'features of social life -networks, norms and trust -that enable participants to act together more effectively to pursue shared objectives' (Putnam, 1995: 664). In simple terms, we might conceive of any form of multi-agency working representing a network where social capital exists, providing opportunities for realisation of the benefits that are assumed to flow from this, such as: sharing knowledge, improved information flows and increasing trust between staff leading to enhanced efficiency and effectiveness (Brewer, 2013;Kula and Sahin, 2016;Pino, 2001).
Social capital has frequently been used as a framework to examine aspects of policing, most obviously in relation to community policing where the engagement of local residents in crime reduction strategies alongside the police has been found to realise wider benefits (Crawford, 2006;Pino, 2001). Our concern, here, however is with so-called 'organisational' or 'network' social capital, concepts used in much management literature that is more concerned with the benefits social capital can bring to the realisation of organisational goals (Burt, 2000;Kilduff and Brass, 2010;Lee, 2008;Nahapiet and Ghoshal, 1998;Weber and Weber, 2010). As specific forms of (re)organisation, colocation models of multi-agency working offer potential to realise social capital benefits across key dimensions described by Nahapiet and Ghoshal (1998) as structural, cognitive and relational. Structural embeddedness refers to the properties of the network as a whole; relational embeddedness refers to the personal relationships individuals have with each other; and cognitive embeddedness refers to shared representations and meanings. The framework offers a means of examining the potential for social capital in multi-agency working and illuminates those features that might be particularly relevant for successful practice in co-located initiatives. It is to these dimensions that we now turn.

The structural dimensions of social capital
The structural dimensions of social capital refer to the characteristics, or types, of relations that exist -how a network or partnership is organised. A distinction can be drawn between 'bonding' and 'bridging' social capital that have distinctive network characteristics, and hence benefits (Putnam, 2000). In organisational network research the idea of bonding social capital is reflected in the notion of network closure (Burt, 2000) where strongly interconnected social relations create high levels of trust between individuals, generating a sense of belonging, a framework for identity and the establishment of a normative order (Burt, 2000). The consequent 'cohesive networks' that emerge offer opportunities for innovation and sharing of complex knowledge (Kilduff and Brass, 2010).
Bridging social capital, by contrast, refers to outward looking, socially diverse networks that bring people into contact who might otherwise not meet (Putnam, 2000), giving access to wider information and opportunities than might be available through bonding social capital. In organisational network research, bridging social capital references the idea that social capital is generated across 'structural holes' (Burt, 2004) by boundary workers who are able to draw on, and share, the benefits of multiple networks. As such, bridging social capital is believed to generate access to new resources and information that is not available to closed networks, and for building connections that may have otherwise remained unconnected.
At this structural level, co-location models of multiagency working might offer the potential to realise benefits arising from both bonding and bridging social capital by virtue of the overlapping and nested social relations that colocation presumes to occur. By locating staff together, bonding capital should emerge as staff develop relationships with each other and as they bring their own external connections to this arrangement, the extent of bridging capital should be extended. In simple terms, we might envisage co-location generating three new sets of relationships that require development and negotiation, those that develop: between co-located staff themselves; between co-located staff and their 'home' agency; and between co-located staff and wider agencies. Understanding the social relations underpinning these different network forms may offer insight into how the benefits of social capital might be realised by the police in co-located initiatives.

The cognitive dimensions of social capital
The cognitive dimensions of social capital refer to those elements of social relations that support shared understanding including the development of shared language, codes and narratives. These cognitive dimensions of social capital can influence the benefits that accrue in any network based on the assumption that when a knowledge base is shared it becomes much easier to combine information and knowledge to generate new knowledge -to generate synergy (Lee, 2008;Nahapiet and Ghoshal, 1998).
The cognitive dimensions of social capital in multiagency working might depend on shared understanding of the issues being addressed. We might hypothesise that in closed networks, characterised by bonding social capital, cognitive embeddedness will be high: individuals share a common knowledge base and reinforce a sense of belonging through shared stories and narratives. In more open networks, however, we might hypothesise that this type of cognitive embeddedness is not a given and may point to barriers to successful multi-agency working where cognitive dimensions of social capital are absent. Examining the ways in which cognitive dimensions of social capital are embedded in co-location initiatives may give insights into the development of new knowledge across different types of relationships within and beyond the core activity.

The relational dimensions of social capital
The relational dimensions in the framework refer to the quality of social relations between individuals. Specifically they include: trust, norms, obligations and expectations, and identity. These features are important in definitions of social capital where they concretise 'what' social capital 'is' or consists of -the types of relations that are assumed to exist.
Relationships that are high in trust are generally seen to be more productive than those where trust is low (Fukuyama, 1995). Trust, in this context, generally refers to the ways in which individuals are perceived as reliable, competent and acting in good faith (see Nahapiet and Ghoshal, 1998: 254). Brewer (2013) found that in multiagency initiatives where levels of trust were high, more informal information exchange occurred that encouraged creativity in dealing with issues and generated new relationships between the police and partners. By contrast, where levels of trust were low, organisations still worked together but they relied on formal processes and procedures which were not as productive as the informal exchanges that occurred in high trust relationships.
Norms represent a degree of consensus within a network -agreed ways of operating that are shared and understood (Nahapiet and Ghoshal, 1998). The establishment of norms is said to facilitate information exchange, reduce the time it takes to access information and allows for the informal exchange of information (Lee, 2008). Obligations are best understood as a form of reciprocity in which there is an expectation that others will 'return a favour' at some point in the future. The final feature of relational embeddedness in the framework refers to identification where the establishment of a shared (group) identity enhances and incentivises cooperation. The importance of shared identity for the establishment and sustainability of some forms of social relations makes intuitive sense in the context of multiagency working but may be a source of tension in co-located initiatives where agencies retain strong employment (and professional identity) links to their 'home' agency. This may be of particular interest for co-location efforts involving police officers deeply embedded in an occupational culture that is often resistant to organisational reform (den Heyer and Mendel, 2019).
The relational dimensions of social capital bear most similarity to features of multi-agency working that underpin models of best practice, encompassing trust between partners, common goals and so on. Some have described the levels of trust and accrual of tacit knowledge between the police and other agencies in multi-agency working as a form of social capital (Crawford and L'Hoiry, 2017;O'Neill, 2014), drawing attention to some of these relational dimensions. Examining co-location through the lens of relational elements of social capital may offer further insights into the ways that these are manifested in different types of multi-agency working arrangements.

Social capital and the police
The police might be characterised as an organisation rich in bonding social capital. The classic solidarity feature of police occupational culture aligns closely with the conceptual underpinnings of the social capital literature, referring as it does to the close bonds and high levels of trust that bind officers together in a common cause. The density of rules and procedures that surround the operation of policing, and the evidence surrounding the socialisation of police officers into the informal norms that govern police officer identity, all point to an organisation that exhibits clear elements of social capital. Kula and Sahin (2016) argue that social capital among police officers improves their effectiveness, echoing more general findings regarding the benefits of social capital for organisations.
Following the social capital thinking, multi-agency working requires police officers to engage in different types of networks; to share information and knowledge; to trust partners and establish new ways of working. Examining police engagement with multi-agency working through the lens of social capital should provide a more nuanced account of the barriers and benefits that accrue to the police from these arrangements. Previous studies examined social capital within police organisations (Kula and Sahin, 2016;Langbein and Jorstad, 2004), thereby limiting understanding of the wider multi-agency setting; others have explored multi-agency working with the police but focusing on single elements of social capital such as trust (Brewer, 2013), or have a focus on social capital within communities (Crawford, 2006;Pino, 2001). Furthermore, studies of co-location involving the police also tend to be limited to perceptions of those working 'inside' (Brewer, 2013). An exception to this is Atkinson (2019) who identifies limitations of co-location for those 'outside' the initiative that impeded face-to-face communication between officers and other agencies.
The theoretical and conceptual discussion so far suggests that social capital offers a theoretical lens on the potential and actual benefits of multi-agency working that emerge from the relationships between agencies engaged in partnership. Nahapiet and Ghoshal's (1998) framework offers a means to explore these networks through structural, relational and cognitive dimensions, providing a more thorough analysis of the benefits and challenges of multiagency working than is often the case. By examining these issues through a case study involving police officers and local authority housing officers, we examine the 'actual' experiences of a co-location initiative from multiple perspectives, allowing a more rounded approach to both the study of co-location and of social capital in multi-agency working to emerge.

Methods
The starting point for this case study was a formal partnership, underpinned by contractual agreements and funding between the police force and local authority. The agreed aim -to develop better responses to victims of anti-social behaviour -could have been met in any number of ways and so decisions made by senior management at this initial stage were critical in shaping the type of collaboration that emerged. The key elements decided at this point were as follows: Staffing -the partnership was established on the basis that police officers and local authority housing officers would form the core staff of the collaboration focusing on personal anti-social behaviour with a newly formed neighbourhood enforcement team to deal with environmental issues.
Co-location -it was agreed that these three distinct teams would share the same office space and that this would be located in the main council building, although staff retained primary employment links to their 'home' agency. Hub and spoke model -the agreement from the outset was that this 'core' staff (the hub) would facilitate communication with other agencies who had an interest in the management of anti-social behaviour (the spoke). Weekly meetings were used to bring external agencies together with hub staff to review cases and discuss problem solving.
The research team was recruited to undertake an internal evaluation of the operation of the hub with a view to identifying areas for improvement and good practice. A mixed methods approach comprised qualitative and quantitative elements, although only the former is reported here.
The qualitative data collection was designed to capture a range of perceptions of the hub, comprising senior managers, staff who had been co-located, professionals working outside the hub and hub clients, and thus the final sample of participants includes those internal and external to the hub operation (see Table 1) .
In total, the data includes views from 42 different people connected to the hub operation in some way. Twelve indepth interviews were undertaken with individuals working external to the hub; three participants were interviewed in both waves of data collection. Views of those external to the hub were also sought through focus groups convened among officers from Safer Neighbourhood teams (SNTs). These comprised one sergeant, four police constables and three police community support officers (PCSOs) (eight individuals in each wave). From within the hub operation, focus groups covered the three co-located teams: hub police officers (n¼6); hub housing officers (n¼5) and hub enforcement officers (n¼9) giving a total of 20 staff (98% of the workforce). The focus groups in wave 1 and wave 2 of data collection comprised the same 20 staff.
All of the staff involved in the hub were invited to participate in focus group discussions. These were conducted by home agency/role so that the groups were homogenous in composition. A total of eight focus groups took place covering the three main staffing groups: police officers, housing officers and hub enforcement officers. The main aim of the focus groups was to discuss how different staff groups were experiencing co-location and any operational issues that senior managers could address going forward. The method proved useful in allowing participants to discuss the benefits and limitations of the co-location as well as identifying points of difference between colleagues that allowed them to consider each other's perspectives on the initiative. Across the two waves of data collection, every member of co-located staff took part in the focus groups or a follow-up one-to-one interview if they were unable to attend the group. The latter was important to ensure that all staff felt their voice had been heard in the evaluation process. None of the individual interviews revealed new themes or issues that had not emerged during the focus groups. To complete interviews with the hub staff, a one-to-one interview was conducted with the hub manager. It was appropriate to interview this person separately given their managerial role which spanned all the separate groupings within the hub.
One of the aims of the evaluation was to consider the impact that the hub was having on key stakeholders 'beyond' the hub. These comprised three main groupings: individuals from 'home' agencies who were intended beneficiaries of the hub's creation; senior managers who had supported the development of the hub -the visionaries; and those whose role in other agencies gave them a vested interest in the work of the hub in responding to victims. Topic guides were devised to explore how people understood the purpose of the hub, any benefits/limitations they had noticed from the hub development, any improvements they felt could be made, and how their work with regard to anti-social behaviour and victims had changed as a result. In addition, focus groups were held with police officers from the SNTs who were the main point of contact for the hub police officers and the communities across the city. All PCSOs, neighbourhood police officers and sergeants were invited to participate, but in the event only those who were working shifts on the day of the focus groups took part. No follow-up interviews of these officers were undertaken. The focus groups were convened to explore similar issues to the one-to-one interviews for stakeholders 'beyond' the hub. Two waves of data collection took place: one when the hub had been running for six months and a second wave one year later. Four respondents were only interviewed at stage one: these were all managers and senior managers whose views on 'why' the hub was established and how it was intended to operate formed the context to the evaluation. Three respondents were interviewed only at stage two: two of these were from agencies whose work overlapped with the remit of the hub and the purpose here was to examine the impact of the hub on supporting victims of anti-social behaviour. A third had not been employed at stage one but was one of the recommendations of the first stage evaluation and so it was deemed important to consider whether the involvement of a mental health link worker had made a difference. Three respondents were interviews at stage one and stage two: these were individuals holding roles that had reported the greatest impact at stage one and they were interviewed again to consider whether those changes had held a year later. Focus groups were convened at both first and second wave for all hub staff and SNTs.
Interviews and focus groups lasted between 40 and 90 minutes and were audio-recorded and transcribed. The data reported here are derived from a secondary analysis of the original data from the evaluation that used a constant comparison method to identify key themes relating specifically to the experience of co-location.
The range of stakeholders involved in the evaluation offered an opportunity to explore aspects of co-location from 'internal' and 'external' perspectives, while the longitudinal element afforded opportunities to examine changes over time. A key limitation of the data is the lack of wider input from agencies working with the hub that were not police officers or representatives from within the council. Although we approached a range of agencies to participate, take-up was slow and/or non-existent. Observations at multi-agency hub meetings where these agencies were invited to give input into cases suggested that there was limited engagement with the work of the hub. We reflect further on this omission in our findings.

Findings
As a model of multi-agency working, it was not surprising to find benefits of co-location mirroring existing evidence in key areas. However, the case study also gave insight into how these benefits affected those working outside the colocation raising questions about challenges created by co-located initiatives. We begin by outlining the improvements in service delivery that the hub generated, before going on to examine aspects of practice that underpinned these. The latter highlights differences in experience from officers working within the hub and those external to it that raise questions about the trade-offs that occur in co-located initiatives.
A central aim for the hub was to improve the service for victims of anti-social behaviour, and especially for those classified as high risk. Improvements in informationsharing, knowledge exchange and expanding networks/ connections were producing beneficial outcomes for the management of cases in multiple ways.
First, information-sharing facilitated by a single database of cases allowed for a daily review of all reports of anti-social behaviour to the police and council.
Second, physical proximity to appropriate colleagues meant that the service provided to victims was less fragmented and responses were quicker than in the past: They don't get the police turning up and saying 'you need to ring the council' but they get a police officer and someone from the council at the same time working on solutions together. (Police inspector E1) Third, the co-location had expanded the range of enforcement-and tenancy-related action that could be taken by bringing together teams with different powers to deal with anti-social behaviour. This had positive consequences for both the management of crime and support for victims. The recognition that pooling resources and powers was a positive outcome of the co-location was summed up by one of the hub police officers as follows: [Working with the council] gives us far more tools to deal with asb [anti-social behaviour]. If we just isolated and didn't work with the council, we're very limited apart from when criminal offences have been committed, we're very limited as to what we can do to assist. But working alongside the council we can use their powers if you like to get a better result. (Hub police officer E1) Finally, there were efficiency gains arising from the colocation, and the reorganisation of neighbourhood policing that had facilitated the initiative, whereby two service delivery officers replaced seven safer neighbourhood beat managers. The rationale for this was that the hub 'removed the treacle' created by dealing with anti-social behaviour. The reallocation of high-risk and complex cases also meant reduced paperwork for SNTs, which was welcome, and coordinated information-sharing allowed the service delivery officers to access information about anti-social behaviour cases if they needed to.
Many of these service improvements were realised from very early on in formation of the hub and continued to be discussed at the second stage of interviews, suggesting that the early quick wins from co-location were embedded in day-to-day practice over time. The features of the co-location that facilitated these positive outcomes mirrored findings from previous studies, clustering around improvements in information-sharing, knowledge exchange, increasing range of expertise to support cases drawing on shared networks and contacts, and so on. However, these practices brought tensions and trade-offs that give some insight into the limitations of co-location to which we now turn.
Physical proximity enabled coordinated informationsharing between police officers and other teams within the hub. The nature of these exchanges was both formalthrough a single database used to record all incidents of anti-social behaviour, and informal -simply getting to know each other and engaging in informal discussion about cases. However, improvements in information-sharing within the hub were not matched with maintenance of prior relationships with home agencies. For the police this was particularly problematic because tensions arose between the hub officers and those working in the SNTs, where normative orders had been disrupted affecting the nature of information exchange. Hub officers were able to task PCSOs in neighbourhood teams with actions such as completing victim risk assessments, or collecting witness statements, which disrupted the usual control hierarchy and generated frustration for hub officer when tasks were not completed: You can see some PCSOs who don't like doing [victim risk assessments] -they'll do anything they can not to do a VRA [victim risk assessments] so then you spend your time emailing them asking them to do it . . . if you ask the sergeant to have a word then you're like the big bad hub. (Hub police officer E1) SNTs countered these concerns by questioning what the hub officers' actions were on cases and why it was legitimate for them to be tasking others to undertake some actions, highlighting the ways in which teamworking across different parts of the force had been lost: What are they actually doing day to day with each of these jobs? That's what we're not sure about. (SNT officer E2) Informally, there was a disconnect between hub police officers and their neighbourhood counterparts once the former were no longer part of the daily briefings that formed a critical part of information exchange. SNTs wanted more face-to-face and informal communication: I'd like for them to come and see us more -just spend a day in our office speaking to us, sharing information. (SNT officer E2) But this view was not shared by hub officers: [S]ometimes you just don't' know what's going to come in on that set of shifts, and to sit at another station for four hours just so PCSOs see us is overkill. (Hub officer E2) Capacity to share information with contacts and agencies working outside the hub had started to realise the 'spoke' model that had been envisaged as the hub officers assumed expertise and knowledge of cases across the city. However, where new relations developed with non-police agencies, relations with police officers beyond the hub deteriorated. Officers in SNTs felt they had lost control of cases, were unable to feed intelligence into the hub and that their local knowledge was not valued: They seem to go in all guns blazing and just take over. For me personally I'd prefer them to come to me -I mean OK I'm not always available but if I am -just get in touch and say 'what are you aware of' or what approach would you take instead of just going in with their approach cos it might not always be right for these people. (SNT officer E1) The consolidation of information about anti-social behaviour meant there was a risk that the hub became a 'silo' that was increasingly separated from the agencies that dealt with neighbourhood crimes. For housing estate managers in the council, this was frustrating because they had access to police officers in the building but were still not informed about criminal activity in social housing properties: My frustration with the hub is that all they are getting [from an address] is the ASB [anti-social behaviour], whereas from our point of view if a drug raid has happened the chances are there are lots of other things going on at that property that we would want to know about but we wouldn't necessarily know about that drug raid [unless the SNT told us]. If there's a lot of criminal activity going on from a property it is a breach of tenancy, so we want to know about it, but from the ASB officers point of view they're only looking at the ASB so we still don't get that information. (Housing estate manager E1) In SNTs this focus did not always make operational sense: You can't divide those jobs up and say 'well there's a bit of ASB [anti-social behaviour], you go deal with that, and here's a crime, you go deal with that. It needs a holistic approach. (SNT officer E2) Sharing of networks and contacts between the police and colleagues within the hub gave police officers direct access to expertise more quickly and easily than in the past, such as social services and private landlord teams within the council. Police officers were also able to bring their own contacts in specialist areas such as domestic violence and child sexual exploitation from across the force. However, not everyone was prepared to engage in these broader arrangements or to access emerging networks around the management of anti-social behaviour: If I know how to deal with something I won't speak to them about it unless they've requested something, because I'll just get on with it. (SNT officer E2) Some estate managers want to tackle everything themselves because it's their patch and they're sort of the sheriffs of that patch, so I think some of them think they don't need the hub to help. (Council solicitor E1) Furthermore, some agencies that had previously worked closely with the police and/or housing officers across antisocial behaviour, the establishment of the hub had disrupted networks: It seems to have just put an extra person in between . . . so my communication with the housing manager direct has gone almost. (SNT officer E2) I would prefer to go to one of the cops in the hub that I know, but that's not the process anymore. (Youth Offending Team police officer E1) Some respondents working outside the hub were concerned that the range and number of agencies involved in weekly meetings to discuss cases was dwindling by the second round of interviews. There was some agreement that this was rational and efficient -agencies did not attend unless cases were relevant to them -but at the same time, this limited capacity for the hub team to gain additional knowledge or understanding from different professionals.
Knowledge exchange and developing expertise was supported in co-location through consistency of staff and clarity of hub aims. By co-locating agencies with a clear remit to respond to high-risk victims and review cause for concern cases, a core staff group was able to embed new knowledge in day-to-day practices. A consistent and dedicated team of officers was crucial in establishing positive relations with other agencies and underpinned developing expertise and emerging trust between the police and other agencies. However, the co-location did not eliminate tensions around professional identity and types of knowledge that threatened to undermine some aspects of improved outcomes for victims. In particular, the introduction of mental health support worker into the hub generated conflict over cases that manifested as a welfare/enforcement divide with mental health support workers prioritising treatment and support to victims and perpetrators, and hub police officers concerned this could undermine enforcement activity: Well I was at court to give evidence at an eviction trial and [the mental health support worker] turned up defending the person the council was trying to evict. So half of the hub was there to try and enforce the eviction because of the behaviour and the other half was there trying to plead for the judge not to evict. (Hub police officer E2) Physical proximity and clarity of aims could not overcome competing professional knowledge claims and priorities, despite agreement that mental health issues were a core concern in many high-risk anti-social behaviour cases. In some instances, mental health support workers were excluded from information-sharing about cases to protect enforcement activities: They've restricted [name] level of access [to the shared database] now because she can see all the prosecution's case, and if she's gonna go to court and defend these people she'll know all the details. (Hub police officer E2) Housing estate managers similarly felt their capacity to work on cases had been undermined: The hub officers don't want that, they really don't want the estate managers getting in the way . . . my manager is saying that it's still [our] case and [we've] got to keep ownership of that case, whereas the ASB [anti-social behaviour] officers would say 'we're dealing with this the way we want to deal with it'. (Housing estate manager E2) The perceptions of safer neighbourhood officers that those in the hub knew no more than they did and sometimes less, added to questions about how information about cases was shared, culminated in questions being raised about the legitimacy of having police officers in the hub at all: They're not using their police powers to do the role they are doing . . . you've got six warranted officers sat in an office that aren't doing police powers based roles. (SNT officer E2) The analysis revealed a much broader and more nuanced experience of co-location than those reported previously. Similar to earlier studies, we found staff within the colocated facility reporting a range of benefits arising from the new arrangements, and evidence of deep learning within and across teams that at least met if not exceeded the vision of senior management. However, by including the experiences of those working outside the hub more significant challenges were identified relating to ongoing relationships between the hub and 'home' agencies that we are unaware of from previous research.

Discussion
A central theme of this discussion has been the importance of social relations underpinning multi-agency working, and in particular the value of exploring these through the lens of social capital. The second part of our analysis examines the findings in relation to the three dimensions of social capital described above -namely those structural, cognitive and relational features -as a means of identifying broader learning from our case study example.

Structural embeddedness
As anticipated in the social capital literature, the hub represented features of bonding social capital through the physical proximity of staff that facilitated team-building, a sense of belonging and common purpose. Some of the positive benefits arising from the hub's activities were attributable to bonding social capital, notably aspects of innovation and sharing complex knowledge between police officers and the housing enforcement team. There was evidence that co-location was also breaking down interprofessional barriers that some respondents had experienced previously, adding to a sense of increased trust between co-located staff. However, tensions around knowledge claims and professional identity undermined bonding social capital with mental health workers, suggesting that simply co-locating different professional groups will not necessarily generate bonding social capital without attention being paid to the congruence of normative practice.
Bridging social capital was also realised through new connections between hub staff and agencies they had not worked with so closely in the past. However, networks were slow to expand and there was little evidence of any expansion of networks between the first and second round of interviews. Without this bridging social capital, colocated initiatives might find that they become 'silos' and unable to capitalise on the new resources or emerging/new knowledge as this emerges in a particular field.
Furthermore, the case study showed changes in the relations between co-located staff and their home agencies, which could be characterised as a shift from cohesive networks (bonding) to working across structural holes (bridging). The frustration that neighbourhood police officers felt about the failure of hub officers to visit local stations more often or to engage in face-to-face informal communication reflected the way in which previously cohesive networks had broken down. Hub officers increasingly perceived the neighbourhood teams as one of many agencies or networks that they were working with. It is possible that, over time, these new relations will spawn bridging social capital that is beneficial to both neighbourhood and hub officers but at the time of the case study, this transition had not occurred.
Analysing social relations in the co-location through the lens of structural embeddedness identifies challenges that future initiatives could seek to overcome, particularly around how to sustain and incentivise the ongoing development of bridging social capital through connections with external agencies; and how to manage the transition from bonding to bridging social capital for co-located staff and their home agencies.

Relational embeddedness
The co-location realised relational embeddedness within the hub that aligned with aspects of the social capital literature, in particular high levels of trust and reciprocity between police/housing officers that underpinned communication and enforcement activity along with normative practices that defined what the hub 'did' and how. However, there was also evidence that some elements of relational social capital were not being realised because trust was not so high between the predominantly enforcementoriented police/housing officers and mental health workers. This was manifested in removal of mental health workers' access to shared information about cases and (therefore) different norms between these teams. Initially, at least, these actions were supported by hub management although by the end of the study attempts were being made by management to seek resolution to some of the conflicts that had arisen. The difficulty here may be how far early manifestations of mistrust between frontline staff can be overcome through managerial efforts. The social capital theory literature would seem to suggest that without efforts to foster trust and reciprocity between staff members, such efforts could remain at the level of rhetoric rather than practice.
Furthermore, relational social capital was diminished between the hub officers and neighbourhood teams, suggesting that the benefits of social capital were not being realised (or had been lost) in these ongoing relationships. Trust started to diminish as hub officers and their neighbourhood counterparts began to question each other's reliability to undertake tasks; the reciprocity that defined neighbourhood policing teams also excluded hub officers so that some PCSOs became resentful of hub officers 'telling them what to do' as new norms were established within the hub about how cases should be dealt with. Ultimately, the identity of hub officers as 'real police officers' began to be questioned by some of their colleagues in neighbourhood teams. This shift did not appear to be linked to rank per se (PCSOs were generally comfortable with taking orders from a range of fellow officers), but more about the separateness of the hub as an entity where trust and reciprocity were eroded to the extent that each group of officers began to question the other's motives and capabilities. The key issue here seemed to be a lack of reciprocity rather than adherence to strict rank hierarchies.
Clearly these elements of social capital are closely linked and overlapping, and attempts to present them in this way are open to interpretation, but the evidence from this case study did seem to suggest that as new ways of working became established in the hub, prior relationships began to suffer. The challenge for co-location, then, is how to ensure trust, reciprocity and identity are retained between staff within organisations when individuals are moved into new extra-agency roles.

Cognitive embeddedness
The social capital literature draws attention to the importance of shared understanding and a common knowledge base in order for benefits to be realised. Achieving cognitive embeddedness within co-located initiatives helps to generate the conditions under which creative and innovative responses can be realised by pooling ideas that emanate from a shared definition of the causes of particular problems. This was indeed the case between police/housing officers whose shared understanding of enforcement allowed them to deal with perpetrators and victims of anti-social behaviour in flexible and creative ways. However, where cognitive embeddedness was absent, as with mental health workers, opportunities for creative solutions were stifled and staff felt isolated and undermined. In turn, this affected the quality of bonding social capital among the team where suspicion developed instead of reciprocity.

Conclusion
The complexity of anti-social behaviour makes it fertile territory for multi-agency working and the co-location was able to navigate through this complexity by bringing together multiple enforcement agencies while linking in to wider networks for additional support. Indeed, we identified a range of benefits arising from the co-location in this case study that confirms previous research findings. However, evidence from professionals working outside the hub revealed a more complex mix of benefits and challenges emerging from co-location. Among these respondents, the positive aspects of the co-location related almost exclusively to the impact on workloads as the hub picked up the most complex and time-consuming cases.
A major finding here is that as the benefits of co-location became embedded in new ways of working among hub staff, relations with home agencies deteriorated, particularly for police officers. Tensions arose between the hub officers and the SNTs who performed regular policing duties across the city. Where neighbourhood officers felt they had the same level of knowledge about anti-social behaviour, and in some cases deeper understanding because of their proximity to the victims affected, there seemed to be little for them to gain from the hub apart from reductions in workload when, and if, hub officers took on the paperwork that surrounded complex cases. For co-location to be successful in their eyes, there needed to be more obvious and direct benefits for them. Rather than being a lack of clear process or rules, the data showed that changing social relations lay at the root of tensions between the hub and neighbourhood officers. These relational obstacles seemed somewhat impervious to shared occupational cultural norms, suggesting that (re)organisation of police structures may be a mechanism for developing deeper cultural changes.
The experiences of officers involved in this co-location suggest that police management need to consider the relationships between co-located staff and the wider force to ensure that professional identities are not diluted and that trust and solidarity does not deteriorate. Our data suggest that it is in the period of transition from wider force to colocation that the greatest disruption occurs, and therefore where opportunities lie to ease relational disjuncture.
Our analysis has utilised an organisational network social capital approach to explain the benefits and challenges arising from the social relations that underpin co-location. Our findings suggest that co-location did not naturally allow the benefits of both bonding and bridging social capital to emerge. The evidence was stronger for bonding social capital within the hub generating a range of positive outcomes; but bridging social capital was slow to develop and indicated that efforts to ensure co-located initiatives remain outward-looking may be crucial to avoid a silo mentality developing. The relational elements of social capital helped to explain the manifestation of tension between co-located staff and home agencies, giving insight into the domino effect that can occur once one aspect of relational embeddedness is lost: once operational norms changed and previous ways of working were disrupted, it was not long before trust between officers started to erode and questions of professional identity were raised. In tandem, cognitive embeddedness proved to be a useful device to examine the benefits and challenges of bringing different ways of knowing to bear on complex problems. In sum, we would suggest that managing changing relationships in colocation to preserve existing social capital whilst exploiting opportunities arising from newly formed connections requires careful planning and consideration of: how colocated staff might be affected, what types of professional knowledge are likely to produce the most beneficial outcomes, and how these are perceived by those beyond the co-located initiative.

Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding
This work was supported by North Yorkshire Police through the Police Innovation Fund.

Ethical approval
This work was subject to ethical approval by the Ethics Committee of the Department of Social Policy and Social Work at the University of York