Introduction
Being granted refugee status does not of itself resolve the difficult situation of many people who flee their homeland in search of security in England. Instead, they find themselves homeless in their new country (Basedow and Doyle, Reference Basedow and Doyle2016; London Housing Foundation, 2004). The research reported on in this paper contributes to our understanding of the relationship between refugee homelessness and immigration policy by addressing the question: Is homelessness concentrated amongst particular newly-recognised refugee groups? We used longitudinal empirical data that spanned the eight months following a refugee’s grant of asylum and binary logistic regression to assess the relative importance of factors associated with refugee homelessness that have been identified in the existing literature. Although longitudinal surveys of refugee homelessness have been conducted in Australia (Flatau et al., Reference Flatau, Smith, Carson, Miller, Burvill and Brand2015) and Canada (Hiebert, Reference Hiebert2009; Mendez et al., Reference Mendez, Hiebert and Wyly2006; Simone and Newbold, Reference Simone and Newbold2014), such analysis has not previously been carried out for England. Our analysis underlines the centrality of immigration policy to the production of refugee homelessness, while also demonstrating the impact of certain broader life experiences and demographic characteristics. It is valuable to know which refugees are disproportionately affected by homelessness so that housing and integration services can target preventative interventions on groups at highest risk. Further, being homeless often goes hand-in-hand with health and other social support needs, so understanding who becomes homeless and why could pinpoint other problems faced by refugees that also need addressing. Our aim was to obtain a systematic empirical account of which refugees are most at risk of homelessness that can underpin future theoretical research. In so doing, we also considered the inferences that could be reasonably drawn from our data about the factors that contribute to homelessness.
The nationality of newly arrived asylum seekers changes as global crises come and go, but in the last fifteen years the number of asylum seekers from the countries considered in this paper has consistently been relatively high. Asylum seekers are people who have applied for refugee status, but whose cases have not yet been decided or are subject to legal appeal. Refugees, on the other hand, are people whose application for asylum has been accepted. Since March 2005, those granted refugee status have no longer been given Indefinite Leave to Remain (ILR), but rather their status is time limited and is reviewed after five years. It is only then that the refugee may be granted ILR. Those failing to meet the criteria for refugee status may be granted Humanitarian Protection (HP) status or Discretionary Leave to Remain (DLR) for a period of three years. ‘Failed’ asylum seekers are those whose applications have been rejected, but are still living in the UK, either because they are awaiting return to their home country or they have stayed without permission. In 2006 (the main period to which the data in which the data analysed in this study related), 9.2 per cent of the 23,608 asylum applicants were granted refugee status, and a further 9.8 per cent either on HP or DLR (Home Office, 2016: Table as.01). For the remainder of this paper all three categories (ILR/DLR/HP) will be referred to as ‘refugees’. All three carry the same entitlements to stay in the UK, and rights to housing, to work, and many of the other rights of full citizens (unless DLR is awarded with ‘no recourse to public funds’). We analysed a survey of refugees who had received their grant eight months previously, although some of them had been in the UK as an asylum seeker for five years or more while awaiting a decision on their case.
Why focus on housing? An extensive literature documents the importance of housing as one of the cornerstones of successful resettlement processes, which has been comprehensively reviewed by Allsopp et al. (Reference Allsopp, Sigona and Phillimore2014). Suitable long-term, safe, affordable accommodation provides the basis for many wider aspects of starting a new life. For example, it provides an address from which to apply for benefits, is often a prerequisite to finding a job, and enables children to have consistent schooling. The quality of housing affects a person’s overall physical and emotional well-being, as well as their ability to feel ‘at home’ (Hickman et al., Reference Hickman, Crowley and Mai2008; Phillips, Reference Phillips and Spencer2006a). Peckham et al. (Reference Peckham, Wallace, Wilby and Noble2004) found that housing was the main area that refugees using Home Office funded projects felt needed improving in their lives. Similarly, Atfield et al. (Reference Atfield, Brahmbhatt and O’Toole2007) found that in refugees’ aspirations for integration, housing was a primary concern. Moreover, refugee housing policy has a wider impact on community cohesion because of (mis)perceptions that they queue jump the allocation of social housing over longer-settled communities (Pillai et al., Reference Pillai, Kyambi, Nowacka and Sriskandarajah2007; Rutter and Latorre, Reference Rutter and Latorre2009).
This paper contributes to literature on the policy process by presenting new evidence on the shortcomings of, and contradictions between, policies affecting the ‘integration’ of new refugees. We followed Praxis (2002) in adopting a conceptual framework based around three stages of a refugee’s housing ‘journey’: (1) reception in the UK, (2) transition from asylum-seeker to refugee status, and (3) settlement into permanent housing. We focused on the transition period because Perry (Reference Perry2005) and Phillips (Reference Phillips2006b) have identified this as a critical stage when housing and support need to be available. This is because refugees frequently become homeless at the point when they are given the right to stay in the UK (Basedow and Doyle, Reference Basedow and Doyle2016; London Housing Foundation, 2004; Refugee Council & HACT, Reference Perry2009). Consequently, there are extremely high levels of transience among refugees at the transition stage (Phillimore and Goodson, Reference Phillimore and Goodson2008; Phillips, Reference Phillips2006b), with those staying with family and friends moving frequently to spread the ‘burden’ (Duke, Reference Duke1996). They can face destitution if there are delays whilst their benefits claim or national insurance number is processed (Chartered Institute of Housing & HACT, 2012; Fullegar and Smart, Reference Fullegar and Smart2008; Refugee Survival Trust, 2005). Refugees’ lack of knowledge of the housing market (council, social and private rented housing), their rights and responsibilities as tenants, and those of landlords, at the time when they receive their grant also make their search for adequate housing difficult. Additionally, few refugees have savings, and since support for asylum seekers is very limited and they do not have the right to work, finding money for a deposit and basics such as a cooker, fridge or furniture may be a problem (Perry, Reference Perry2005; Refugee Survival Trust, 2005). We next review the existing evidence on the extent of homelessness among the refugee population, before going on to describe government policy on asylum and refugee integration.
Literature review
Previous research has suggested that many refugees are in severe housing need (e.g. Allsopp et al., Reference Allsopp, Sigona and Phillimore2014; Phillips, Reference Phillips2006b). Such understandings have been mainly provided by small-scale local studies focusing on particular refugee groups (e.g. Aden Hassan et al., Reference Aden Hassan, Lewis and Lukes2010; Robinson et al., Reference Robinson, Reeve and Casey2007; Warfa et al., Reference Warfa, Bhui, Craig, Curtis, Mohamud and Stansfeld2006) or particular geographical areas (e.g. Dwyer and Brown, Reference Dwyer and Brown2005, Reference Dwyer and Brown2008; Netto and Fraser, Reference Netto and Fraser2009; Phillimore and Goodson, Reference Phillimore and Goodson2008; Wren, Reference Wren2007), although many of the issues raised in these studies do have a much wider resonance and informed the hypotheses of the present study. According to the homelessness charity Crisis, between 2014/15 and 2018 the number of rough sleepers in London whose last settled base was National Asylum Support Service (NASS) accommodation increased (2018: 340-342), although these figures include not only newly-recognised refugees but also refused asylum seekers. CHAIN (2018) records information on rough sleepers encountered by outreach workers across different agencies in London, but notes that it is extremely difficult to obtain immigration status data from rough sleepers. Furthermore, their report does not include people from ‘hidden homeless’ groups such as those ‘sofa surfing’ or living in squats. In addition, changes in figures may reflect changes in the capacity of agencies rather than changing patterns of user demand. Nevertheless, the figures do suggest high and ongoing levels of destitution among refugees.
Several recent reports analyse the support offered to those who have been granted refugee status, for example No Accommodation Network (NACCOM, 2018), the Refugee Council (2017) and the All Party Parliamentary Group on Refugees (2017). Projects involving housing providers that help refugees also offer some insights that led to the hypotheses underpinning this study (Clarke et al., Reference Clarke, Diacon, Guimarães, Kriteman and Neville2006; Glasgow Housing Association et al., 2011; Jesuit Refugee Service, 2018; Jones and Hussain, Reference Jones and Hussain2010; Mullins and Jones, Reference Mullins and Jones2009; Perry, Reference Perry2009; Refugee Council, 2017; Refugee Media Action Group, 2006). Much of the research is small scale and stems from charities that produce research based on their own practice and draw their samples largely from their own client base, leading to inevitable, yet rarely acknowledged, biases (Allsopp et al., Reference Allsopp, Sigona and Phillimore2014).
We turn next to what we know from international research about how refugees’ pre-migration demographic characteristics might contribute to their homelessness. A high proportion of refugees are single young men, and they may have difficulty accessing social housing because they are not deemed a ‘priority need’. Private sector accommodation, often in hostels, multi-occupied properties or through sharing, is frequently the only available option to them, so we would expect this group to be more vulnerable to insecure temporary housing. Indeed, Duke (Reference Duke1996) found that the young were most likely to have experienced homelessness. However, for Australia Forrest et al. (Reference Forrest, Hermes, Johnston and Poulsen2012) found sex but not age to be a significant differentiator of housing conditions. On the other hand, many homeless refugee households are female-headed, so sex may provide little protection against homelessness. Forrest et al. (Reference Forrest, Hermes, Johnston and Poulsen2012) found that larger households have greater housing problems, whilst Duke (Reference Duke1996) found that single people were more likely to have been homeless, so the existing literature is inconclusive. With respect to religion, Muslims have been found to be at a particular disadvantage (Lindley, Reference Lindley2002). Abramsson et al. (Reference Abramsson, Borgegård and Fransson2002) found for Sweden that the country migrants come from is important. One reason might be discrimination by landlords towards those from a visible minority, or it could be that rough sleepers would rather stay in locations where there are established refugee communities, and will forgo stable accommodation to achieve this (Cebulla et al., Reference Cebulla, Rice, Tomaszewski and Hough2009). However, Forrest et al. (Reference Forrest, Hermes, Johnston and Poulsen2012) found little evidence of any important differentiation among countries of origin in housing problems in Australia. They concluded that it is important to focus attention on individual characteristics rather than ‘cultural’ backgrounds linked to country of origin.
We will now move on to discuss what is known about how refugees’ post-migration experiences might contribute to their homelessness. Region of residence in the UK is relevant because it is common for refugees to move from ‘dispersal’ areas (especially Northern England) to where they have friends, family or community groups, typically culturally diverse London. Since rents are extremely high in London and parts of the South East this move can result in rough sleeping or hidden homelessness. Also, low income is a risk factor for a range of housing problems, from the inability to afford contents insurance, to the risk of being unable to pay the rent and thereby potentially lose one’s home. Employment is also relevant because landlords prefer tenants in work and the employer may be able to provide a reference. Furthermore, those who lack the language skills necessary to deal with landlords or letting agents will be at a disadvantage. The ability to speak or write English well were not significant differentiators in housing conditions in Forrest et al.’s (Reference Forrest, Hermes, Johnston and Poulsen2012) Australian study, but the ability to read English well was. Poor health is likely to be both a contributor to, and an effect of, homelessness. For example, trauma, torture and time spent in refugee camps are associated with housing problems (Forrest et al., Reference Forrest, Hermes, Johnston and Poulsen2012). Health workers working with new migrants report overcrowding and sharing bedrooms as leading to problems including asthma, depression, other stress-related symptoms and repeated contraction of infectious diseases (Aden Hassan et al., Reference Aden Hassan, Lewis and Lukes2010; Perry, Reference Perry2005). Further, the timescale for Home Office Immigration to process an asylum claim varies and we might expect refugees who spent longer in the UK as asylum-seekers prior to their grant to find it easier to secure housing because of greater knowledge of the social housing system and the housing market, and having stronger social networks. Social networks are extremely important as they can offer refugees practical help, information, ‘know-how’, and emotional support that they can use to find housing (Salway et al., Reference Salway, Platt, Chowbey, Harriss and Bayliss2007). However, securing support from social contacts might be hard for a refugee because of reluctance to ask for help, a fear of creating obligations in social relationships where there are cultural expectations of reciprocity, or mistrust of other community members.
Overall, there are serious shortcomings in the statistical data on asylum seekers and refugees in the UK. Such a lack of research about what happens to people after they have received a positive decision contributes to delays in policy planning for longer-term settlement (Netto, Reference Netto2011b). Correa-Velez and Gifford (Reference Correa-Velez and Gifford2007) argue that the scanty data stems not only from difficulties collecting the data, including the mobile nature of the population and reluctance to be counted, but also from political considerations. Whilst they stress that access to reliable statistical data is fundamental for undertaking any meaningful analysis of asylum-seeking policies, they argue that few scholars have interrogated statistics in this area thus far.
Policy context
This section explains how government policy in 2005-7 contributed to refugee homelessness, the period of the data collection for the Home Office’s Survey of New Refugees (SNR). It shows how little had changed by 2019, and why the SNR data continue to be useful to the analysis of immigration policy. The first aspect of immigration policy that is of central relevance to this study is that it leads to delays to housing support, and integration generally, for asylum seekers who subsequently go on to be granted refugee status (Mulvey, Reference Mulvey2015; Phillimore, 2011). This is because amidst a furore over ‘bogus’ asylum seekers the New Labour government introduced the Immigration and Asylum Act 1999 in an effort to deter asylum-seeking and be seen to take action about housing and benefits going to asylum seekers (Bales, Reference Bales2013). Crucially, to achieve this the Act made a sharp distinction between asylum seekers and refugees (Morris, Reference Morris2007). For example, asylum applicants were barred from receiving any benefits, instead being issued with food vouchers. Although vouchers were scrapped, the cash sum paid to asylum seekers (£37.75 per week as of August 2019) is a fraction of the safety net welfare benefits available to refugees. The Home Office have recently attempted to speed up the process of moving refugees to mainstream benefits by: arranging an appointment with a Jobcentre; the inclusion of a National Insurance number, a pre-requisite for applying for welfare benefits, with the Biometric Residence Permit; and the (re)introduction of Refugee Integration Loans to help with upfront payments such as rent deposits. Nevertheless, the recent reports cited above provide evidence that asylum policy is still at least partially responsible for delays to arranging housing for refugees.
The second aspect of policy that is of particular relevance to this study is that the 1999 Act changed accommodation for destitute asylum seekers to a ‘no choice’ of location basis arranged through the newly-created NASS. Destitute asylum seekers were (and still are) provided with temporary housing by housing providers contracted directly by the NASS/Home Office. Where the NASS/Home Office provided accommodation, this was (and still is) almost invariably outside London and south east England, following the government’s policy of ‘dispersal’ in the Act. In 2006, when much of the data collection for this study took place, 95 per cent of asylum seekers (including dependants) housed through NASS were living outside London and the South East (Home Office, 2008:61). However, of those only receiving subsistence assistance from NASS, and making their own accommodation arrangements, 16 per cent were living outside London and the South East, indicating these were refugees’ preferred locations (Home Office, 2008:60).
Third, key to the analysis of the results reported on here is the policy now, as at the time of the data collection, that refugees in NASS/Home Office accommodation have an arbitrary grace period of only 28 days to vacate it and find alternative housing once a positive asylum decision had been made (Doyle, Reference Doyle2014). Responsibility for housing them is transferred from national to local authorities at this time. Refugees have the right to apply for accommodation under the homelessness legislation on the same basis as other UK citizens, but they are not recognised as in ‘priority need’ by mere virtue of being a refugee. Refugees are at high risk of homelessness at this point and their access to permanent social housing depends on local rules. The ‘local connection’ criterion prevents refugees moving to their preferred locations, because, under housing legislation in England, refugees have a local connection with the (dispersal) area where NASS accommodation was last provided for them (Clarke et al., Reference Clarke, Diacon, Guimarães, Kriteman and Neville2006; Phillips, Reference Phillips2006b). For single people without children, the only chance of being recognised as in priority need for social housing is meeting the highly subjective and locally determined criteria for ‘vulnerability’ (Robinson et al., Reference Robinson, Reeve and Casey2007). Thus, the limited availability of social housing is a problem that faces refugees and the general population alike. Refugees ineligible for permanent social housing have to find their own housing in the private rented sector, in hostels, or with friends and family, yet many have no financial assets whatsoever with which to cover deposits for private rental accommodation, nor a job, nor a bank account (Doyle, Reference Doyle2014). There is a shortage of affordable private rented accommodation. This often results in newly accepted refugees sleeping on floors and sofas or, in some cases, living on the streets (Clarke et al., Reference Clarke, Diacon, Guimarães, Kriteman and Neville2006). Civil society and refugee community organisations play an important role in plugging gaps in local authority service provision, more significantly at the transition stage (British Red Cross and Boaz Trust, 2013; Mayblin and James, Reference Mayblin and James2019).
At the time of the collection of data used for this study the then New Labour government aspired to improve housing for accepted refugees (as distinct from asylum seekers):
Stable and safe accommodation is an essential pre-requisite to integration … To prevent homelessness among refugees; to ensure that refugees have access to suitable housing, dealing with any problems caused by the move-on from NASS or other accommodation to mainstream support (Home Office, 2005:25).
Specific housing-related assistance to new refugees during the time of the collection of data used for this study included the UK Border Agency (UKBA) Integration Loan Scheme and local authorities’ Supporting People grants. Between 2008 and 2011 the Government funded a short-lived national programme, the Refugee Integration and Employment Service (RIES), which helped new refugees navigate the ‘move on’ period, offering 12 months of practical help, especially with housing, thereby implicitly recognising how difficult this transition can be. A guide to the help available for newly-recognised refugees at the time of writing is available from the Department for Work and Pensions (2018). The present government’s approach to refugees is encompassed in the 2018 Integrated Communities Strategy. Accordingly, the government is trialling Local Authority Asylum Liaison Officers who will:
… support people granted asylum into housing, education and work. We will use the evidence gathered from this work to inform future provision for this cohort, and to ensure that no-one leaves Government supported accommodation with nowhere to stay (HC Deb 20 Jun 2018 153848W).
Also, on paper, refugees can benefit from the duties local authorities now have to all eligible applicants (i.e. irrespective of priority need status) under the Homelessness Reduction Act 2017 (England). It requires local authorities to work with eligible applicants to develop a personalised plan setting out the steps they and the authority need to take to prevent them becoming homelessness. However, despite efforts by local authorities to implement the Act, it has proven financially unsustainable to do so for many councils (House of Commons, 2019).
In summary, asylum and refugee policy has changed since 2005, but is nevertheless still characterised by a sharp distinction between the treatment of asylum seekers and refugees, with asylum seekers entitled to minimal support, no choice in the location of NASS/Home Office accommodation, and only 28 days to leave the accommodation. It is therefore reasonable to assume that homelessness among refugees is as much a valid concern now as at the time of the data collection. Furthermore, the findings of this research have on-going relevance because refugee integration policy is yet again subject to change as new government contracts to provide asylum accommodation will commence in 2019 (HM Government, 2018).
Data, variables and method
Research questions
With regard to the potential link between refugees and homelessness, the focus of our analysis was not just on the question of which newly-recognised refugees are most at risk of being homeless, but also what is it about being a refugee that could play a vital role in bringing about homelessness. Our approach was to combine the selection of variables for inclusion in our regression model by evidence-based theorising with empirical investigation to test for the variables’ relative importance. This paper addresses the following research questions arising from our overview of the existing evidence and policy outlined above:
-
Is homelessness concentrated amongst particular newly-recognised refugee groups?
-
What is the relative role played by pre-migration demographic factors, post-migration life experience factors, and government policy on asylum in accounting for any patterns found?
Analytic strategy and data
Regression analysis has been established as a method for assessing the factors associated with refugees’ housing, for instance by Forrest et al. (Reference Forrest, Hermes, Johnston and Poulsen2012). The SNR allowed us to explore the contribution of a wide array of the potential contributory factors identified in the literature, which we describe below. We divided the variables into three conceptually driven blocks informed by the literature and policy review: pre-migration demographic factors; post-migration life experience factors; and government policy on asylum. Binary logistic regression was used to estimate the associations between homelessness and the independent variables. Odds ratios (OR) and standard errors (SE) were calculated. Analyses were weighted to correct for survey sample design and non-response bias. The source of data was the Home Office UK Border Agency’s SNR which ran between December 2005 and March 2009 and which was obtained from Home Office et al. (2010). The overall aim of the survey was: (1) to collect information on the characteristics of new refugees at the time of their asylum decision and (2) to provide data on the integration of new refugees in the UK over time. A postal baseline questionnaire was sent to all new refugees aged 18 or over who were granted a positive decision of asylum, HP or DLR between 1 December 2005 and 25 March 2007. It was a longitudinal study in which respondents were surveyed four times in total: one baseline questionnaire and three follow-ups after 8, 15 and 21 months. The baseline questionnaire collected information on the characteristics of refugees at the time of their asylum decision, including their previous education and employment, English language ability, physical and emotional health. The three follow-up questionnaires were used to collect information on how these refugees integrated in the UK over 21 months in terms of English language skills, employment and housing. However, although 5,678 refugees participated in the first survey, a response rate of 70 per cent, the number had decreased to 939 respondents by the third follow-up. For this reason, this study only uses data from the baseline questionnaire and first follow-up, as the later samples were considered to be too small for valid analysis. It builds on previous research using the SNR by Cebulla et al. (Reference Cebulla, Daniel and Zurawan2010), Daniel et al. (Reference Daniel, Devine, Gillespie, Pendry and Zurawan2010), and Cheung and Phillimore (Reference Cheung and Phillimore2017) to provide more detailed empirical evidence on housing.
Variables
There are a number of criteria against which refugees’ housing situation could be evaluated. Ager and Strang (Reference Ager and Strang2004) and Forrest et al. (Reference Forrest, Hermes, Johnston and Poulsen2012) used owner occupation as a marker of resettlement. However, the policy shift in the UK towards offering more temporary time-limited forms of protection under HP/DLR status makes it hard to obtain a mortgage, so choices are limited to the social or private rented sectors. Homelessness is clearly undesirable, and has been suggested as an indicator of poor integration (Ager and Strang, Reference Ager and Strang2004). Therefore, we used as our dependent variable ‘situation of homelessness in the previous six months’. Homelessness takes many forms, not just rough sleeping. Consequently, this variable was defined as the respondent having lived in at least one of the following in the previous six months: Hotel or bed and breakfast; hostel; homeless shelter; living on the streets; and ‘other than a house or flat’.
We now turn to the independent variables we used – the factors that influence housing outcomes for refugees suggested by the existing literature – which are listed in Figure 1. We included country of origin (although it is unclear whether this was interpreted by respondents arriving from third countries as country of birth). We also included a variable on education because it is often assumed that difficulties are greater for people with less education. Only refugees from England were included because the Scottish Government’s approach to refugee integration is different from that of Westminster (Mulvey, Reference Mulvey2018). The explanatory variable used to capture income was employment. The variable that captured self-perceived language capability drew on the questions: ‘Compared to a native English speaker, how well can you speak English?’ and ‘Compared to a native English speaker, how well can you read English?’. The model encapsulated health status in the question: ‘How is your health in general?’. This was self-reported on the 5-point scale Very good/Good/Fair/Bad/Very bad. Since it is hard to assess the quality of networks, the variable for social network was defined as having no family or friends in the UK, with the emphasis on those with no contacts at all. The samples from the SNR were sufficiently large to examine whether there were notable contrasts between the above sub- groups.
Figure 1. Independent variables used in the model.
Model strengths and limitations
Methodologically, this study has certain strengths. For example, other UK surveys (e.g. the Labour Force Survey, 2011 Census) have not asked about refugee status. However, it also has limitations. For example, there was no information on immigration status (i.e. refugee/HP/DLR). This is relevant as they vary in how temporary they are, which could influence the willingness of landlords to accept them as tenants (CIH & HACT, 2012). A limitation of the SNR dataset is that unaccompanied child refugees were not included, nor were refugees who had claimed asylum in another EU country, nor those who entered under resettlement schemes whereby they were granted refugee status before leaving for the UK (e.g. the Gateway Protection Programme). The meaning of housing ‘integration’ has been discussed elsewhere (Phillimore and Goodson, Reference Phillimore and Goodson2008; Phillips, Reference Phillips2006b), and there is a lack of consensus surrounding the most useful terminology to use. Furthermore, the meaning of ‘home’ is culturally contested and will differ across different groups and different places (Kissoon, Reference Kissoon, Temple and Moran2011). In view of this we were careful not to over-simplify the term ‘housing’ and to appreciate the psycho-social significance of the term ‘home’, whilst considering its inter-relationships with the other indicators of integration (Ager and Strang, Reference Ager and Strang2008; Phillimore and Goodson, Reference Phillimore and Goodson2008). Many studies attempt to capture the dynamics of migrant housing circumstances by examining ‘housing pathways’ or ‘housing careers’. The basic argument is that the outcome of the housing search process results from the interplay of a number of factors. Some of these were included in this model, including a household’s characteristics. But the SNR did not have data on some other factors, such as: a household’s cognitive resources, and their preferences; the dynamics of the local housing market; difficulties faced in the housing search, and the strategies used to overcome these barriers (Murdie, Reference Murdie2008). The effect of housing on well-being depends on the quality of the neighbourhood and wider area in which housing is located (Netto, Reference Netto2011b). It is therefore a drawback of the SNR is that there is no data about neighbourhood. Lastly, the data did not allow use of an approach that focuses on relationships and interactions, perceptions and choices, events and decisions, and the actions of individuals and collectives (Netto, Reference Netto2011a; Robinson et al., Reference Robinson, Reeve and Casey2007). However, quantitative data analysis of large-scale surveys can be complementary to seeking to understand these types of relationships. It is important to qualify our results by observing that the models described in this paper were linear additive, assuming a similar layering of effects for different groups, and not incorporating interaction effects. While this was an effective research strategy at this stage, there is clearly scope for further investigation of more complex models, with considerable scope for exploration of interactions and differential effects between groups.
Results
Our primary research question was: ‘Is homelessness concentrated amongst particular newly-recognised refugee groups?’. Our secondary research question was: ‘What is the relative role played by pre-migration demographic factors, post-migration life experience factors, and government policy on asylum in accounting for any patterns found?’ The results of the logistic regressions are presented in Table 1. An odds ratio (OR) of greater than 1 indicates a variable that is associated with an increased likelihood of having been homeless and an OR of less than 1 indicates a variable that is associated with a decreased likelihood of having been homeless. It is clear from our results that the odds of experiencing homelessness are systematically structured around a set of identifiable factors outside of the control of the individuals directly affected. While statistical associations cannot in themselves establish causation, they can suggest inferences about likely relationships when underpinned by a meaningful rationale. We sought to establish such a rationale through the above literature review in order to provide a coherent framework within which we could make sense of our results, including identifying where the greatest prospect for policy intervention lies.
Table 1. Results of binary logistic regression models for homelessness. Model – ‘situation of homelessness in the previous six months’
Constant = 3.482
-2 Log likelihood = 1199.004
N = 1023
Nagelkerke R Square = .285
* p < 0.05,
** p < 0.01,
*** p < 0.001
1 Income Support or Jobseekers Allowance
From the review of existing evidence, we expected to find relationships with government policy on asylum, and this is clearly confirmed: our results underline leaving NASS accommodation is a powerful predictor of subsequent homelessness. The OR for the NASS coefficient was 0.530 (95% CI 0.348-0.712). This suggests that those who were not in NASS accommodation were about half as likely to become homeless than those who were. But leaving asylum support accommodation is not the only trigger for rough sleeping among refugees. With respect to the timescale for the Home Office to process an asylum claim, the OR for the ‘5 years or more’ coefficient was 0.412 (95% CI 0.123-0.701). This suggests that refugees who waited less than six months were about 2.4 times more likely to become homeless as those who waited five years or more. This is not an argument in favour of the Home Office delaying a decision. Rather, it highlights the importance of access to local knowledge. We might expect refugees who spent longer in the UK as asylum-seekers to find it easier to secure housing because of having greater familiarity with the housing market and stronger social networks to draw on for practical help and advice.
The pre-migration demographic results were broadly in line with expectations since our analysis further revealed that, even after taking into account asylum policy, family type and age do matter. First, with respect to family type, the OR for the ‘Couple, no children’ coefficient was 0.336 (95% CI 0.04-0.632). This suggests that couples without children were about a third as likely to become homeless as single people. The OR for the ‘Couple, with children’ coefficient was 0.465 (95% CI 0.255-0.675). This suggests that couples with children were roughly half as likely to become homeless as single people. Thus, our results confirm the well-established vulnerability to homelessness of single adult households. On the other hand, living as a couple, either with or without children, appears to operate as a protective factor against homelessness. The direction of causation may be that the process of getting recognised as homeless and in priority need by a local authority is not an easy one, and general strains on housing services prevent assistance to single people. Second, with respect to age, the OR for the ’45-64’ coefficient was 0.408 (95% CI 0.076-0.74). This suggests that refugees aged 18-24 were about 2.5 times more likely to become homeless as middle-aged refugees. Although some ORs for country of origin were statistically significant, no conclusion could be drawn from them because of the size of the CIs.
Our research questions arose from the literature review and policy context sections of this paper, and our statistical analysis has, we would argue, broadly confirmed the factors associated with refugee homelessness identified by previous studies. Our analysis of the SNR would tend to support the contention that eviction from NASS accommodation with inadequate social support is the biggest factor in homelessness, in line with the literature, without discounting the possibility of individualistic factors in specific cases. It is possible that the balance of underlying contributory factors may vary between different refugee groups, with pre-migration factors more important in some cases and government policy causes more important in others. However, a contribution of this quantitative study has been to ascertain the relative importance of these different factors.
Concluding remarks
This concluding section draws out the implications of this analysis for potential policy responses. Given the link between homelessness and other poor outcomes in refugee integration, such as worklessness and illness, early preventative measures could ultimately prove highly cost-effective, not to mention the moral imperative to act on behalf of this extremely vulnerable group. The extent of refugee homelessness, and its social consequences, leads us to the view that reform of national and local policy is urgently needed. As a result of our analysis, we are able to propose a number of changes to national policy that we believe have the potential to greatly ameliorate the homelessness problem among refugees. These can be summarised as follows.
-
1. Greater flexibility in implementation, with refugees being able to stay on in Home Office provided housing for longer periods than the arbitrary 28 days, and landlords’ costs covered by housing benefit.
-
2. Provision of support to refugees leaving Home Office housing, which recognises the different stages of refugee housing need, and includes access to housing advice, rapid access to housing benefit and a rent deposit loan scheme.
-
3. More Home Office resources to support the voluntary sector in their work with refugees in crisis and to prevent street homelessness.
We also propose the following in relation to the role of local authorities.
-
1. Local Authorities should uphold their statutory homelessness prevention duty by creating dedicated pathways for new refugees to access temporary housing and progress promptly to settled housing without having to use emergency homeless services when their asylum support ends.
-
2. Local authorities should particularly invest in support for young single people, who are disproportionally affected by homelessness.
Twenty-eight days is frequently not long enough to find housing. That being the reality, the implication of this research for frontline practitioners working with asylum seekers is to prepare them for the process that will follow should they be granted refugee status. Housing and homelessness services can support asylum seekers by providing information about private sector rents locally, the level of housing benefit they are entitled to, and the local social housing allocation criteria. Services can mitigate risk by working with clients to plan an interim means of support, such as schemes run by charities, churches, food banks and night shelters. Once a positive decision has been received, it is a matter of urgency for frontline services to help refugees apply for a National Insurance number, apply for work or benefits, make a homelessness application to the council, and look for private rented housing. The uneven distribution of risks by family size and age identified in this paper can be marshalled to develop policies that target the most vulnerable groups, while being quite clear that such policies can never be expected to predict with perfect accuracy all those who would otherwise become homeless. Thus, whilst practitioners should avoid making assumptions based on a refugee’s family type and age, agencies might seek to develop partnerships with community groups among whom homelessness is especially prevalent. Nevertheless, those living in private rented accommodation or with family and friends are likely to be ‘hard to reach’ in order to do prevention work before they present as sleeping rough.
All the available data points to the conclusion that the asylum support system is failing many refugees, especially young single people, just when they are accepted as being in need of protection. Indeed, it is the ‘transition’ point that demonstrates most clearly the failings of the current system, which is designed, in theory, to meet the needs of some of the most vulnerable and traumatised, many of whom arrive in the UK with scant resources having been forced to flee. Whilst a range of factors are associated with homelessness among refugees, the asylum support system of itself is a major contributory factor. Our empirical findings, coupled with our rationale on causation, clearly suggest that homelessness is predictable under present policies but not inevitable. Therefore, action to address the housing problems faced by refugees moving on from the accommodation provided for asylum seekers should be considered as a high policy priority, albeit that associations between homelessness, household size, and age also present intervention opportunities.
