Denying homeless persons access to municipal support

Purpose – The purpose of this paper is to focus on the denial of social support to homeless persons and related societal effects of new local governance arrangements. Design/methodology/approach – Analysis of new data and secondary evaluative and comparative data on the policies, administrative structures and management styles of Copenhagen, Glasgow and Amsterdam have brought better understanding of the elements of local governance arrangements that influence the number of homeless persons who are denied access to services and the number of persons sleeping rough who are not eligible for social support. Theoretical explanations for the impact of governance arrangements on these processes and societal effects are considered. Findings – It appears that while the body of research, reports and policy documents on non-eligibility for homelessness services is growing, legal responses at best remain vague, and policies are still in the process of being developed. Modest progress on policy goals, and even more so on policy instruments, leading to less detrimental outcomes, can be explained by centralising and decentralising trends and the relationships between state and society. The latter may also be indicative of how the increased focus on the legal problems of some EU migrants can be explained. Research limitations/implications – The two points in time documented for the case studies are relevant in understanding processes underlying the current circumstances of homeless persons and homeless migrants and offer an interdisciplinary insight into governance and politics, law, and public and health service perspectives. Social implications – Good policy practice, as this paper shows, can lead to a difference in individual lives. Originality/value – Much is unknown about considerations inside government. This paper contributes by combining theoretical and insider perspectives.


Introduction
Over the past decades, homelessness has made its way up the political agenda in Northern European cities (Benjaminsen et al., 2009;Hermans, 2012;Anderson and Serpa, 2013;Boesveldt 2015). The prioritisation of the social needs of nationals/locals among homeless populations has resulted in sub-populationshomeless non-nationals or those with lower levels of needthat are deemed non-eligible. This non-eligibility is a consequence of either not passing the threshold for a mental health need that is seen as a priority for the Netherlands (cf. Planije and Tuynman, 2013;Trimbos Instituut, 2015Boesveldt, 2015), or not meeting the "priority need" or the more recent "intentionality" criteria for the UK.
Non-eligibility also applies to "immigrants who do not 'contribute' to the host member state's welfare system and hence are perceived as a challenge to solidarity spaces which are still predominantly governed by national bonding ties" (Kramer, 2017).
While current EU legal texts give considerable leeway for interpretation by nation states, decentralising trends may increase local discretion on eligibility issues. The many grey areas existing between European Court of Justice (ECJ) case law on European citizenship and the hard texts of secondary EU law give member states considerable leeway to interpret EU law according to their own national interests (Vonk, 2017).
At a national level, the right to social assistance is clouded in legal controversy. Take, for instance, the Dutch law that states that country inhabitants who need support must always receive it. The ECJ has taken a strong stance against the Dutch interpretation of the free movement directive and access to shelters. In addition, the European Committee for Social Rights found that the Netherlands disproportionately denies the right to emergency assistance to migrants (both in regular and irregular situations) by using restrictive criteria to target "vulnerable groups", when, in fact, everyone in the state's jurisdiction has rights to emergency shelter (FEANTSA Complaint No. 86/2012).
While the body of literature on non-eligibility for homelessness services is growing, with legal responses at best remaining vague and policies still under development, the problem of homelessness is increasing. As a result, a considerable number of homeless persons in need of social support services are not receiving them, leading to deteriorating circumstances. But how exactly can we understand which government actions have led to the increases in numbers and barriers to assistance? The findings presented in this paper will provide more insight into the processes taking place and the underlying governance arrangements. Its aim is to clarify what is needed to address this urgent social need and identify the solidarity issues at stake.

Theoretical framework
In the last 10-15 years, government focus on homeless persons entitled to relief has thrown up the problem of those not entitled to support. Indeed, a former Amsterdam alderman (Guusje ter Horst) responded to this question by stating that it would be better to not even make policies on the issue. However, this is not what happened, and the active emergence of homelessness strategies has highlighted this issue, almost as a by-product (that some have termed "wicked", Rittel and Webber, 1973;Head and Alford, 2015) of the policies themselves.
With the changing situation within the EU, in relation to both EU and non-EU migrants, the issue of migrant homelessness is a growing problem for European cities. According to Gosmé (2014), the interconnection between local actions, national measures and the EU policy arena on homelessness is increasingly evident, to the extent that the EU is now in a position to support member states in addressing homelessness, as the recent refugee crisis has shown. Boesveldt (2015) and Boesveldt et al. (2017) have identified three elements crucial to understanding local governance arrangements and explaining the dearth of services and the resulting societal effects (such as rough sleeping). These are policy, structure and management. A focus on these three elements may eventually lead to specific policies to address support needs and target homelessness.

Policy
Under the policy heading, it is important to focus on the question of the policy model, which concerns the moral and empirical assumptions underlying the policy at stake. Moral assumptions refer to the values considered relevant by policy makers and that in their opinion should be realised (Dunn, 2012;Tirion, 2014). In the context of denying services to homeless people, these assumptions may refer, for example, to the prejudice and mutual distrust that could serve as additional explanatory factors other than legal reasons for refusing homeless persons (Vonk, 2017), or the UK "intentionality" criterion. A person failing to pay their rent can be assessed as intentionally homeless, and therefore not qualifying for services (Boesveldt, 2015).
Empirical assumptions pertain to the causes, features, magnitude and nature of a policy problem. Realistic and scientifically sound empirical assumptions are a precondition for achieving the desired effects (Dunn 2012;Tirion, 2014); for example, the assumption that "becoming part of the group of homeless people who cannot fend for themselves is considered detrimental to the mental wellbeing and resourcefulness of the (relatively) self-sufficient group of homeless people" (de Ridder et al., 2018, p. 12), or "social benefit tourism". EU freedom of movement would result in a disproportionate burden being placed upon both the welfare system and labour market (cf. Guild et al., 2013). Implicit policy models may lead to (explicit) policy goals. The policy element of a local governance arrangement refers to the attempts that are made to serve one or more public interests. Policy goals and policy instruments are relevant in relation to this element (Benjaminsen et al., 2009;Bressers and Klok, 2014;Dunn, 2012;Fenger and Klok, 2014). With respect to policy goals, a distinction can be made between external goals that refer to circumstances in society and internal goals that target circumstances within the administration itself. Internal goals may, for example, be aimed at improving the functioning of the government's own organisation and/or at aligning separate policy sectors such as health, housing, income and justice (Boesveldt, 2015), thus avoiding fragmentation.
Again, when applied to the denial of services to homeless people, one examples is the raising of barriers for outsidersin the form of a local connection test for shelter and support for the homeless under the Dutch Social Support Act (Roorda, 2014), or the UK, where cities can apply a local connection test for housing support (Wallace, 2014). Basing decisions on "self-sufficiency" regarding determination of the extent to which integrated care is offered to the person concerned serves as another example (Runtuwene and Buster, 2014).
Avoidance of fragmentation and involving the local business services department in the objectives set by the care department become relevant as immigrants can become homeless because of irregularities in work conditions or employers not paying workers' taxes.
Policy instruments are methods used by a government body to achieve the desired results. They should accord with the policy goals of the government body to ensure a sufficient level of goal attainment (Fenger and Klok, 2014), which they often are not (Boesveldt, 2015;Boesveldt et al., 2017). As part of homelessness strategies, or alongside them, but with supportive objectives, specific projects and programmes have been introduced aimed at migrants or non-eligible persons, such as repatriation, anti-begging laws and reconnection schemes to help destitute East Europeans return to their countries of origin, to either their families or communities. From 2007 more than 3,000 people have been "reconnected" from Ireland, the Netherlands and Denmark (cf. Fitzpatrick, Johnsen and Bramley, 2012;Mostowska, 2014b). Some member states (e.g. Britain, Denmark and the Netherlands, cf. Kramer, 2017) have started to make use of discretionary powers emanating from recent CJEU case law to actively remove mobile citizens without sufficient resources of their own. Vonk (2017) claims that, when countries resort to forced expulsion measures, these can prove ineffective or breach basic European human rights standards and they remain a sensitive terrain.

Structure
The structural element of a local governance arrangement concerns, among other issues, the extent to which responsibilities and budgets in a policy sector have been decentralised to the local level (Fleurke and Hulst, 2006). Administrative traditions vary as to the levels of local autonomy, ranging from decentralised structures (Scandinavia), medium levels (Continental) and more centralisation with lower levels of local autonomy (Anglo-Saxon) (Painter and Peters, 2010). In addition, the extent to which these resources have been divided between local government bodies, private companies and non-profit organisations ( Jessop, 2004;SER, 2010;WRR, 2000) may influence the effectiveness and efficiency of public policy (Benner et al., 2004;Bouckaert et al., 2010;Olsen, 2009). In a centralised structure, decision-making powers are highly concentrated in a single organisation or discipline (or at least in a smaller number). Responsibilities and budgets are not allocated to lower government levels and/or non-government organisations. Concrete tasks may, however, be outsourced to private organisations under clear directions. Complex social issues demand a minimum level of expertise and knowledge, necessary for the formulation and implementation of the policy at stake (Boesveldt et al., 2017). Vonk (2017, p. 11) refers to "local dumping", and the decrease in homeless persons' legal protection as a result of no longer treating behaviour related to homelessness (e.g. begging) as part of national law, instead, increasingly addressing it in local regulations and anti-social behaviour measures (Boesveldt, 2015). Dutch and Belgian local governments impose exclusion orders under administrative rather than criminal law, and in England and Wales this happens under private law. Such non-criminal procedures make it possible to bypass strict procedural requirements (Vos and Duran, 2014). According to Vonk (2017), a strict supervisory and reporting mechanism to the national government should be set up so that the protection at local level is realised. Also, FEANTSA (2016) is calling for stricter interpretations of the rights of economically non-active EU citizens and clarification of concepts such as "genuine chance of being engaged" and "unreasonable burden on social assistance system".

Management
A management style of (Anglo-Saxon) local government characterised by a pluralist vision of the relationship between the state and society offers all relevant organisations the opportunity to influence the quality of support services. All these organisations are offered relatively equal opportunities to influence the way in which social issues are resolved. Local government (Scandinavian), with a corporatist style, assumes a closer link between state and society and some official sanctioning of interest groups by government. Particular interest groups are accorded a legitimate role as representatives of their sector of the economy or society. Only a limited number of actors can play the game, and those that do are bound closely with the power of the state (Frederickson et al., 2010;Pierre and Peters, 2000). Mostowska (2014b) and Stenum (2011, p. 261) describe how in Denmark NGOs were threatened with the withdrawal of their public funding if they helped homeless EU migrants, which then led to active disobedience.
To gain a better understanding of the extent to which elements of local administrative and political approaches to homelessness contribute to the denial of access to social support to specific groups, "denied service coverage for homeless people" (percentage within the metropolitan area of the homeless system who do not receive care from one or more providers) and "non-entitled homeless people sleeping rough" (percentage known to sleep rough within the metropolitan area, specified by non-entitlement to services) (cf. Lauriks et al., 2013;Boesveldt, 2015) have been taken as a starting point for studying different governance arrangements. The theoretical framework is outlined in Figure 1.
This paper focuses on homeless persons' non-eligibility for services. The central research question is: RQ1. Whether possible differences in the administrative-political approaches to homelessness between Northern European metropolitan cities lead to differences in the number of persons not being offered services and in related societal effects, such as sleeping rough? Figure 1 Governance arrangements, processes of service delivery as a quality of services and societal effects for non-entitled homeless persons

Methodology
This section discusses the case studies, the operationalisation of local governance arrangements, service delivery and outcomes, including reliability of the available data sources.

Study population
The case studies are Copenhagen (548,443 inhabitants[1]), Glasgow (580,000 inhabitants) and Amsterdam (811,185 inhabitants). Although there were differences in admission criteria for homeless services among the three areas, the population defined/not defined as a target group for the social support system, which therefore could be used to compare the performance of the social support systems in this study, comprised single-person homeless households, with homelessness defined as either sleeping rough or requesting support from housing services for the homeless (i.e. night shelters or emergency/temporary housing). Almost every country in the EU defines this population as homeless (ETHOS definition of homelessness and housing exclusion, cf. Amore et al., 2011), including the UK, the Netherlands and Denmark[2]. However, any estimation of homelessness is based on specific criteria and assumptions (see Busch-Geertsema, 2010;Busch-Geertsema et al., 2016), and the use that homeless persons make of specific homeless services is often a way of identifying those in housing distress and consequently essential for accurate counts for data collection purposes (Waegemakers Schiff et al., 2016). To get a better picture of the homeless population not entitled to services, it is important to study additional data, such as that on the usage of Winter initiatives (that have virtually no threshold) or regarding the non-takeup of services.

Measures of local governance arrangements
Data on local governance arrangements were collected from policy documents and semi-structured interviews with key stakeholders (also see Boesveldt, 2015 for details on these policy documents).
Stakeholders included policy makers, civil servants, service-provider professionals and service users. Interviews were transcribed and policy documents were coded on the three core elements, according to the categories "EU nationals", "Danish nationals", "thresholds", "public mental health criteria" (i.e. referring to criteria related to access to shelters, based on a homeless person's mental health, with more serious health conditions increasing eligibility) and "immigrant", and analysed using qualitative content analysis methods (Atlas.ti v7.1, 1993-2016).

Measures of service delivery and outcomes
Data on process and outcome were collected from publicly available homeless monitors, evaluation studies and municipal registries. These were helpful in getting a better understanding of the situation in each city and were a source of identifying how data change in relation to policy change. In Amsterdam, the periodical monitoring of individuals in the Public Mental Health Care (PMHC) system is conducted by the Public Health Service of Amsterdam (GGD) and published in reports on the status, inflow, throughflow and outflow of PMHC clients every six months (Runtuwene and Buster, 2013). Additional data were collected from the client registration system (GGD Amsterdam, 2014) in which all relevant service providers register their clients for coordination and case management purposes. Runtuwene and Buster (2014) describe how Amsterdam policy between 2012 and 2013 changed along with local connection, now also basing decisions for access to homeless services on levels of "self-sufficiency". In 2012, the total number persons reporting at central access was 1,647, while in 2013, it was 1,138. This is related to a change in registrations and workflows, rather than persons reporting. From 2013, only persons with a low level of self-sufficiency were referred to central intakepreviously, it was everyone with housing need. Moreover, Buster and de Wit (2011) explain that the estimate of the number of outdoor sleepers was based on several assumptions that they were unable to verify. This estimate is therefore more an indication of the order of magnitude of the number of outside sleepers than an accurate reflection of the total number. In addition, it is assumed that it also gives an indication of the trend. It is therefore important to note that the estimate relates to the number of outside sleepers on an average night and that this group has a varying composition over a longer period.
For Copenhagen, the national evaluation of the Danish homelessness strategy (Rambøll and SFI, 2013) was examined. Baptista et al. (2012) comment that in Denmark data on homelessness are extensive as national counts were carried out in 2007, 2009 and 2011 and because there is a national client registration system covering almost all homeless shelters providing data for comprehensive annual statistics. Nevertheless, SFI assesses the count of homeless without permanent or legal residency as more uncertain than the count of other groups. This is because homeless people without permanent or legal residency are unable to use public homeless programmes and so are not registered there, just as they often do not possess CPR (ID) numbers, which are methodically used in the mapping to avoid double counts. Evaluators point out that they are only certain about 164 individuals sleeping rough, of whom possibly only 129 are entitled to homeless services (Rambøll and SFI, 2013).
The Scottish Government Homelessness Statistics Unit is responsible for monitoring the homeless population in Glasgow. Data on the caseload of the Homeless Mental Health Team and discharge and referral data are published quarterly (www.gov.scot/Topics/Statistics/Browse/Housing-Regeneration/ RefTables). Additional data on the number of homeless people in Glasgow and the services they received were requested from the team leader of the Homeless and Asylum Health Services.
In 2013, the number reporting as homeless in Glasgow declined 10 per cent in comparison to the preceding year. The locality of the preventative housing options (HO) pilot showed a drop of 21 per cent. The monitoring report states that there will have been additional presentations at some of these service points which were satisfied through the provision of information and advice or HO assistance and says that further information on these cases will be available once the new PREVENT1 statutory return is in place. Even so, there is concern that the HO pilot functions as a gatekeeper and so policy outcomes of the previous homelessness strategy are blurred as a result (cf. Fitzpatrick, Pawson, Bramley and Wilcox, 2012;Anderson and Serpa, 2013). Fitzpatrick, Pawson, Bramley and Wilcox (2012) noted that administrative changes associated with the increasingly robust implementation of homelessness prevention activities have undermined the value of statistics as an indicator of trends over time in "acute housing need" because the widely adopted HO approach has narrowed the scope of official statistical recording. Table I presents data on non-entitlement to services and the resulting societal effects of governance arrangements on homelessness in Amsterdam, Copenhagen and Glasgow. The relationship between policy and these outcomes is also described on the basis of the interviews.

Results
The following section presents findings related to the possibility of a relationship between local governance arrangements and these outcomes. Notes: a Group at risk is the number of inhabitants; b group at risk is the number of homeless persons in catchment area of healthcare system; c group at risk is the number of homeless with SMI

Amsterdam
In Amsterdam, a relatively small percentage of the population (0.2 per cent) is reported as being homeless. What stands out is the small but stable amount of homeless persons served by one or more homelessness services (30-40 per cent of the total), and it appears that this overlaps with suffering serious mental illness. This leaves 60 per cent of the registered homeless population not receiving homeless services. From the average amount of persons found to sleep outside on one night (11-12 per cent of all homeless persons), the majority have no access to services.
Policy goals. Amsterdam states that from 2010 regional bonding was an important admission criterion for Social Care (i.e. homelessness services). It says that access to social care is only granted if individuals meet all of the following conditions. They: If a person meets the above criteria, it is most likely they will be successful. These criteria also prevent people in high numbers turning to Amsterdam for social care and the already high pressure on the facilities and housing market becoming disproportionately large (City of Amsterdam, 2018, p. 28). Amsterdam policy on the Social Support Act also states that minor problems should be dealt with by non-intrusive solutions and people should take responsibility for themselves.
The connections made in these texts, for example between the criterion and the prevention of "people in large numbers", can also be categorised as parts of the Amsterdam policy model.
The 2015 Amsterdam policy report states that in the first semester of 2015, 6 per cent (35 persons) of applications for homeless services came from outside the Netherlands. Eight persons (3 per cent) from outside the Netherlands met the non-self-sufficiency criteria ( Table 2, pp. 4-5).
Policy instruments. In the Amsterdam case, the number of policy instruments increased after 15 December 2010. Additional so-called "orientation trajectories" for non-national homeless people as well as Barka's reconnection trajectories were subsidised by the city. During orientation trajectories opportunities for self-sufficiency in the Amsterdam context and in the homeless person's country of origin are explored and compared. The trajectories are provided by non-Dutch social workers to facilitate communication with foreign homeless people and gain access to professional networks in the social workers' home countries.
The policy response to the increase in persons without public mental health issues had been to finance additional social work services within the city districts for persons rejected by the municipality based on not having made multiple requests for help in different areas (cf. Runtuwene and Buster, 2013). In 2015, the Winter initiative was extended beyond the Winter period for homeless persons without Dutch nationality, or residing lawfully as a foreigner but not suffering from severe health needs.
Structure. In Amsterdam, the municipal government expected that a big trend towards decentralisation would result in several inefficient outputs (i.e. extra costs, e.g. for additional instruments). The policy to refer persons requesting homeless services to a place where that person is "most likely to be successful" in the Amsterdam interpretation is part of the "National Accessibility" regulation: If the assessment shows that a client should not be taken care of in Amsterdam but in another region, then transfer from Amsterdam to that other region takes place. The scope of intensity of this transfer extends from referral to personal guidance to the municipality concerned. This depends on the degree of self-reliance of the client. In this way, Amsterdam follows the national accessibility and regional ties to the social security system of the VNG (Association of Dutch Municipalities). In this guide, arbitration also provides for the situation that a client cannot go anywhere. (Fourth Social Support Act Regulation 2015 in Amsterdam, 2017, p. 21).
Adjacent or foreground provisions, such as business services departments or employers' organisations that may have a role in the prevention of migrants who arrive in the Netherlands with a worker's status becoming homeless, are not mentioned.
Management. Amsterdam, with its more homogeneous and long-standing relationships and institutionalised coalitions, has to a certain extent absorbed the issue of non-entitlement to homelessness services, rather than relying on adjacent or foreground provisions, such as business services departments or employer's organisations having a bigger role.
In the Amsterdam case study, the clearest relationship between policy and outcomes can be identified. Primarily, there is the high level of serious mental health issues, making homeless people eligible, plus the reason for the majority of those sleeping outside could most likely be for reasons of non-entitlement. Policy instruments have attempted to mitigate the most detrimental outcomes, most recently by the provision of extra places, but elements of structure at local level are also responsible along with negotiation and management of the issue.

Copenhagen
The Policy goals/instruments. Copenhagen policy focused increased attention on persons entitled to homelessness services between 2009 and 2013. The number of homelessness provisions expanded, but primarily for persons suffering from addiction. Homeless persons falling under the responsibility of the mental health region were not served by these new services. According to service providers, a stricter policy with regard to offering shelter to homeless migrants had been introduced around 2006. Homeless shelters could lose funding if they failed to meet the new rules. At the time of the interviews (2012), the rules were felt to have become less strict, but were still not state funded: "We have to turn to private foundations for funding. The Municipality still says we cannot use state money or municipality money" (Voluntary Services Provider).
Copenhagen respondents also link sleeping outside specifically to immigrant homelessness. Since "foreigners" cannot "join the system", they are told to sleep on the street. Reference was also made to "another kind of homeless people" when respondents talked about immigrant homelessness. The following quotation expresses a service provider's discomfort with the current situation: They are not allowed to stay in public-financed shelters. They are only allowed to stay in the emergency shelters that have been set up. The emergency shelters normally run from December to February-March, so it's only in the very cold months that they have the right to some kind of housing or shelter. Often, they sleep on the floor on mattresses. (Voluntary Services Provider) Structure. The Copenhagen structure, which had been expected to serve as the example of a decentralised setting, appeared more multi-level than was anticipated. The Copenhagen policy objective regarding the "non-Danish national", with "outreach and relationship-building work and in close interaction with social centres, shelters/temporary housing and home municipality" as the proposed policy instrumentation, is not mentioned in the national strategy. Respondents claim that the national and local strategies have not successfully addressed the recent increase in immigrant homelessness. "You can say Copenhagen has a problem that was not been dealt with in the National Strategy on immigration. That was the big difference in my opinion" (Statutory Respondent).
Management. It is felt increasingly both by the City of Copenhagen and third-sector providers therewhich at the time of the study (2012/2013) were formally forbidden by law to help non-Danish homeless persons apart from when certain exceptions were madethat a specific policy is required. In terms of management, in Copenhagen a relatively close linkage has been observed between state and society. This means particular influential parties tend to have a rather independent position in relation to the local authority and achieving a balance within the (ultimately national) network is, therefore, part of the municipal effort. This independent mindset is valued by the homelessness sector as is the idea that direct societal influence on policy is a democratic mechanism. It is in this management setting that the introduction of a specific policy dealing with homeless migrants in 2014 was realised.
In the Copenhagen case, clear linkages can also be seen between policy objectives and outcomes. Increased policy attention for Danish and Copenhagen homeless persons has revealed the issues of mental health and migrants among the homeless population. Elements of structure explain differences in the attention paid to the migration issue and how service providers were eventually successful in overcoming restrictions on providing shelter to homeless migrants.

Glasgow
Glasgow not only had the highest number of homeless persons (1-1.2 per cent of the Glasgow population), but also the sharpest decline (−0.2 per cent) over the strategy period. Serious mental issues were less frequent among homeless persons (10-12 per cent), who were also more often served by one or more homeless service (60-65 per cent). Figures on sleeping outside are high (5-8 per cent of the homeless) compared to the other case studies, but relatively low in comparison to the Glasgow homeless population. It is unknown what percentage of those sleeping outside do so because of non-eligibility.
Policy goals. In Glasgow, in 2011, towards the end of the 2009-2012 strategy period, key strategic priorities were formulated. One of which was to meet the target for the abolition of priority need. Priority need is one of the tests a homeless person must pass to determine their entitlement. With the abolition of these criteria, the only distinction left is the intentionality criterion. A closer look at the municipal statistics at the time shows that, of all homelessness applications, half received additional support within the statutory homelessness provisions (60.5-65.6 per cent), while those that did not (39.5-34.5 per cent) are categorised as: potentially homeless; neither homeless nor potentially homeless; applicant resolved homelessness prior to assessment decision; no duty owed; or contact lost before duty discharge (Table 2.3 of the Scottish Government Homelessness Statistics Unit, 2013/2014). The search for partnerships and cooperation is mentioned in reference to the issue of non-eligibility.
Policy instruments. To address the needs of persons not being offered services within the statutory framework, Glasgow ran a pilot for the HO instrument as part of its 2002-2012 strategy. In this approach, housing associations work together in a specific area and local casework services are supplemented by commissioned services. In HO, the focus is away from applications for homelessness and on how people can stay in accommodation (e.g. through mediation) and how new private sector housing supplies can be obtained. In evaluations, the pilot succeeded in reducing the number of homeless households the city still has a duty to house. One respondent felt that this approach had provided the solution to the tension between the duty to house all non-intentional homeless people and the lack of housing: "Housing options has […] filled that policy aspirational gap from 2003 to 12" (Authority Respondent) (Boesveldt, 2015).
Glasgow provides a specific service for asylum-seeking homeless persons (the Hamish Allan Centre) and partly subsidises (with the Big Lottery) drop-in centres providing services to non-eligible persons.
Figures on non-eligible persons sleeping outside are not available. During interviews in Glasgow, rough sleeping is referred to as "relatively small" but with "a tendency" to rise if not addressed. Reference in this context is made to the "complex-need" cases within the homelessness population, and not specifically to non-eligibility, or immigrant homelessness.
Structure. Centralised and detailed policy also addresses the issue of rough sleeping by EU migrants and persons refused asylum. In the 2015-2020 strategy (p. 20), the city's dependence on this policy area in other governance levels is emphasised: The law governing the rights and entitlements of people subject to immigration control and those who have no recourse to public funds is complex and evolving. People in these situations have limited rights to support or assistance and are at high risk of homelessness and destitution. Decisions at UK government level to restrict entitlement to welfare benefits for people whose claim for asylum has been refused and certain EU migrants are likely to increase pressure on the HSCP (Health and Social Care Partnership) and undermine our attempts to end the need to sleep rough.
Management. The Glasgow 2009-2012 strategy also includes explanatory elements of the context around more detached relationships between state and society. For example, policy objectives that address immigration state under objective 23: "In the constantly changing area of 'People from Abroad' seeking advice and accommodation in Glasgow, [we] evaluate our current approach to the services provided to refugees and migrant workers and implement recommendations". In referring to "accommodation" and "migrant workers", the municipality is framing the issue in broader and more specific terms than homelessness. In this way, it also signals expectations of support from the policy areas responsible for the housing and work status of vulnerable migrants.
In Glasgow, it is also possible to identify a relationship between a decreasing number of homeless persons and preventative policy being set. In terms of structure, this authority placed emphasis on structures to address needs identified that require its involvement. A more corporate management style continues along these lines.

Conclusion and discussion
The aim of this paper has been to provide clarity on what is needed to address the social needs and solidarity issues at stake in denying homeless persons access to municipal support. For this reason, relevant elements in governance that are helpful in explaining the delivery and non-delivery of municipal support have been discussed. Interviews and secondary data then illustrated a better understanding of what is at stake. While taking into account methodological limitations in the registrations of persons being denied services, the effect of policy interventions on client outcome levels could be observed and the wider context of the local level was examined through structure and management variables.
It is clear that there is a risk of double exclusion when both migrant and self-sufficient homeless persons are turned down when applying for services in Amsterdam, or migrants and persons with mental health needs in Copenhagen. For migration policy, there was a lack of integration between departments, as well as between multi-level structures, creating complex configurations that eventually gave way to more relaxed policies in Amsterdam and Copenhagen. Meanwhile, in Glasgow, the centralised and detailed policy with respect to rules and targets created a clear task at the local level, with the degree of detail leaving little room for discretion by local civil servants with regard to interpreting policy. In terms of management, where solutions in Copenhagen and Amsterdam come from existing provisions of homelessness services, Glasgow addresses parties outside its homelessness provider network and makes more use of national and legal institutions.
Research into barriers towards services shows that there have recently been increases in such obstacles (Roorda, 2014;Wallace, 2014) connected to this study's data on client outcomes that relates to these policy changes. In addition, the finding that centralised structures and the more distant and pluralist relationship between state and society seen in Glasgow may explain the start made in addressing the wider solidarity spaces at stake by referring to "accommodation" and "migrant workers" which is regarded as more effective (see FEANTSA, 2016;Vonk, 2017). Indeed, findings on the closer relationships between state and society in Copenhagen and Amsterdam are an important contribution to the literature on policy instruments, such as repatriation, anti-begging laws and specific projects (Fitzpatrick, Johnsen and Bramley, 2012;Mostowska, 2014b), since these provide a better insight into understanding the changes that occurred in Copenhagen where providers no longer risk being fined for helping non-Danish homeless persons, or in Amsterdam where orientation trajectories were added to the repatriation offer and extra emergency beds provided.

Recommendation
The most important message to take from this study is that, even though it seems that defining persons in terms of their self-sufficiency, health problems or legal papers may seem the right thing to do when considering the provision of homelessness services, it may well be that targeting the issues causing this situationrather than temporarily solving it with homeless servicesis most important. Therefore, it is recommended that there be a focus on prevention, workers' status and the responsibilities of health services at the most relevant and effective point, which is most probably at the legislative, national level.
Notes 1. Inhabitants figures as at the end of the local strategy period.
2. Homeless families, youngsters and elderly were excluded because these populations are relatively small; they are provided with different services sometimes included in social services; social services for these groups were therefore considered separate from services for the homeless.