Exclusion from the labour market and from the housing market are two facets of precariousness. These situations, which may change over the course of a person’s life, often go together but can also be dissociated from each other: working without having housing and having housing without access to employment. The complexity of these situations deserves special attention, especially as they affect a growing population. Using two waves of the Sans-domicile survey (2001 and 2012), the authors analyse types of employment and housing trajectories of users of homelessness support services. The article reveals a diversity of situations to which public policies are struggling to provide answers, particularly for the most marginalized individuals.
1 Housing exclusion has been increasing in large European cities since the 1990s.  The homeless population in France rose by 44% between 2001 and 2012 to reach 81,000 adults and 31,000 children (Yaouancq and Duée, 2014b).  This increase is linked to rising housing costs and growing employment insecurity that affect low-income groups most severely (Siblot et al., 2015).  Some also argue that the protective role of employment is weakening and that it no longer guarantees access to adequate housing, especially for workers in unstable employment (Fondation Abbé Pierre, 2014). In 2012, almost a quarter of homeless people had a job (Yaouancq and Duée, 2014a).
2 It is important to understand how the housing and employment trajectories of individuals excluded from ordinary housing are affected by these developments. How are they faring on the employment and housing markets? The housing sector, much like the labour market which offers different degrees of employment protection, is a highly segmented, hierarchical, and competitive universe (Soulié, 1997; Brousse, 2006; Gardella, 2014). Emergency shelters are open to all, often very basic, and sometimes provide a bed for a few nights only (Bruneteaux, 2006), unlike more selective and attractive ‘social integration’ hostels which can provide more long-term accommodation. Individuals’ living arrangements also determine the social support they receive (Legal, 2015) and may thus affect their chances of finding—or keeping—a job and their prospects of obtaining a home of their own (Lanzaro, 2014).
3 Policies to tackle homelessness in France have nonetheless evolved in the wake of the Enfants de Don Quichotte movement.  Specifically, the ‘staircase of transition’ model (Sahlin, 2005), dominant for many years, has been called into question. According to this model, as individuals become more socially integrated, they move into stabler forms of accommodation before finally obtaining social housing. Since 2009, it has been superseded by the concept of ‘housing first’ (Vives, 2019), the idea that having a home is a prerequisite to social integration. Following the creation of integrated accommodation and orientation services in 2010, new criteria have been defined for the attribution of places in homeless facilities. These include time spent on the waiting list and degree of vulnerability, taking account of age, disabilities, and health problems (Schlegel, 2017; Eloy, 2019).
4 In this context, while homelessness policy reforms have not radically transformed the previous system (Box 1), it is important to understand how homeless people’s situations have evolved in terms of access to housing and how their housing situation ties in with their labour market status. This article studies the diverse situations of users of homelessness services in terms of living arrangements, employment status (current and past), and housing and employment prospects, examining the interplay between these factors and their evolution between 2001 and 2012. Drawing on data from the Sans-domicile surveys conducted by INED and INSEE in 2001 and 2012, this study adopts an original approach by taking account of experiences and future prospects in two areas—employment and housing—and not simply the respondents’ situations at the time of the survey.
Box 1. Changes in homelessness policies between 2001 and 2012
However, while the number of places has increased considerably in recent years, the homeless accommodation system has been slow to set up social integration hostels or to develop more appropriate housing solutions. The share of emergency shelters has continued to increase (Cour des comptes, 2011), and the number of temporary placements in hotels has doubled since 2004. Social emergency measures have even made a comeback in recent years (Loison-Leruste et al., 2020).
5 Section I presents the data, variables, and analysis methods used to reveal homeless people’s diverse housing and employment trajectories. A typology is then constructed, and different groups are identified; people excluded from employment (Section II) are distinct from homeless people in work (Section III). Last, Section IV analyses the changes that occurred between 2001 and 2012 and discusses the links between employment and housing in the current context.
I – Data and method
1 – Two surveys of French-speaking users of homelessness support services
6 The scope of statistical surveys is generally limited to individuals living in ordinary housing, so homeless people are excluded. To reach this specific population, two homelessness surveys (enquêtes Sans-domicile) were conducted by INSEE and INED in 2001 and 2012 among users of homelessness support services and soup kitchens. The scope and questionnaire of the two surveys were similar, so their results could be compared using the weightings provided (Box 2).
7 In 2012, to reach the growing numbers of non-French speakers among service users, a self-administered questionnaire in 14 languages was used, but it was much shorter than the main face-to-face questionnaire with an interviewer and did not include any questions on employment and housing trajectories, the central focus of this analysis. Consequently, this article focuses solely on the situation of French speakers (French or foreign nationality ) who were surveyed as users of the same types of homelessness support services in both 2001 and 2012. Moreover, as soup kitchens are covered in this survey, people living in ordinary housing but using meal services are also included. In 2012, 9% of users of homelessness support services were living in rented housing (11% in 2001) and 5% with friends or family (8% in 2001). We decided not to exclude these people because they often live in substandard, overcrowded, and unsafe housing, but also because most have a recent history of homelessness. Keeping these people in the analysis provides insight into certain trajectories leading from homelessness to having a personal home. It is also useful to analyse the situations of these people who have housing but are dependent on homelessness support services, their labour market status in particular. As we are not working solely with ‘homeless’ people under the INSEE definition, we will refer to the study population as users of support services. In all, the sample comprises 4,084 individuals interviewed in 2001 and 4,419 interviewed in 2012, representing, respectively, an estimated population of 61,769 and 75,264 users of homelessness support services.
2 – An analysis of housing and employment trajectories
8 In both surveys, only housing and employment over the 13 months preceding the survey were recorded exhaustively (with a monthly time step). The data from this diary were used in our analysis, alongside other variables giving a more succinct description of the respondents’ housing and employment situations at certain moments in their trajectory: pre-survey employment and housing experiences, situation and stability at the time of the survey, employment and housing prospects, and actions undertaken to achieve these goals (Appendix Table A.1).  Where possible, symmetrical employment and housing variables were used (e.g. total time spent in ordinary housing and total time spent in employment). With regard to living arrangements at the time of the survey, rough sleepers were distinguished from those accommodated in emergency shelters or social integration hostels and likewise from those in other situations (in own home or living with family or friends).
Box 2. The 2001 and 2012 Sans-domicile surveys
a In 2012, the survey’s scope was broadened to cover breakfast services, cold weather shelters, and overnight shelters so that homeless people living in places not intended for human habitation (in improvised shelters or sleeping rough) could be better represented. Accommodation for asylum seekers was excluded, however, in both 2001 and 2012.
9 The descriptive statistics for these variables are shown in Appendix Tables A.2a and A.2b, along with the sociodemographic characteristics of users of homelessness support services (Appendix Table A.2c). The survey population is largely male, although the proportion of women increased from 32% to 38% between the two surveys. The share of foreign-born people also increased (from 35% to 43%), as did that of over-50s (from 18% to 27%).
10 Most users of homelessness support services in our survey were in situations of chronic homelessness (Appendix Table A.2a): 47% had never had their own home in France, a situation more frequent in 2012 than 2001. More than two-thirds had been continuously homeless for the previous 13 months, and some were in situations of extreme precariousness, relying on emergency night-time shelters (10%) and stopgap arrangements (débrouille ) (10%), or sleeping rough (6%). Comparison of the 2001 and 2012 surveys shows that a larger share of users of homelessness support services were actively looking for a place to live, but also that more people were living long term in institutional accommodation, where the proportion resident for between 6 months and 2 years had risen from 24% to 32%.
11 Regarding employment, our analyses confirm that respondents were less often excluded from employment than from ordinary housing (Appendix Table A.2b). Only 11% had never worked, and three-quarters had held a long-term job (6 months or more). However, at the time of the survey, only a quarter were working, and almost half had been continuously out of employment (unemployed or inactive) throughout the 13 months preceding the survey (a larger share in 2001 than in 2012). Those in work had very insecure employment status, and their working conditions were very different from those of the conventional labour market, as we shall see in Section III. Last, half of all homelessness support-service users were looking for a job, and 55% were enrolled at a job centre (Pôle emploi). A larger share of respondents reported looking for a job in 2012 than in 2001, but this is not reflected in the share who actually held a job, which fell slightly between the two dates.
12 With the 19 selected and recoded variables (Appendix Table A.1), we ran a multiple correspondence analysis (MCA) using the 2001 and 2012 stacked data,  followed by agglomerative hierarchical cluster analysis of the main factors of the MCA,  to identify groups of individuals with similarities in their employment and housing trajectories alone (active analysis variables), and to see how they are correlated. The other variables, including sociodemographic characteristics, were then crossed with the clusters obtained to characterize them more fully. We also wanted to see how these types of employment and housing trajectories evolved between 2001 and 2012.
13 Based on the dendrogram, the automatic search for the best partitions and the ease of interpretation of results, we selected a partition with seven clusters (Table 1; Appendix Tables A.2a, A.2b, and A.2c). Axis 1 of the MCA distinguishes individuals in employment from those outside the labour force. Axis 2 distinguishes users of support services according to their past trajectory and the actions taken to improve their situation. On one side, the users have been homeless for a short time, move in and out of housing and the labour market, and are taking steps to find work and housing. On the other, their situation is stagnant (no change in their employment or housing status), and they are not looking for a home or a job. Axis 3 corresponds to the very specific situation of foreigners without work permits. Last, among those in work, Axis 4 distinguishes between service users housed and employed by institutions or in the informal sector (moonlighting, scrap collecting, and ‘odd jobs’) and those in more conventional, though often precarious, employment.
II – Users of homelessness support services excluded from employment
14 The first four clusters of the partition comprise users of homelessness support services excluded from employment at the time of the survey (three-quarters of the sample). Their positions in the world of institutional accommodation, their housing trajectories, and their employment prospects are very varied, however.
Table 1. Synthesis of the different clusters of homelessness service users (in 2001 or 2012)
Table 1. Synthesis of the different clusters of homelessness service users (in 2001 or 2012)Interpretation: Users of homelessness support services excluded from employment but with housing represent 6% of all users surveyed in 2001 and 2012. In the past, individuals in this cluster tended to be more integrated on the labour market than the sample as a whole.
Note: The plus and minus signs indicate that the active variables are either under- or over-represented in each cluster. The + – symbol indicates that tendencies vary within the cluster or across the variables considered. For example, in terms of past housing integration, ‘users excluded from employment but with housing’ are overrepresented in the two extreme categories of the variable ‘Time spent in a dwelling in France’ (i.e. in the ‘never’ and ‘more than 5 years’ categories). ‘Workers on short-term contracts’ are over-represented in the category ‘never had a dwelling in France’, but also in never having slept rough (‘Time spent sleeping rough’ variable).
The proportions observed for each variable are significantly different from one cluster to another (chi-squared test).
Coverage: French-speaking users of homelessness support services.
1 – Users without employment or housing, searching actively
15 The largest cluster (39% of respondents) is that of people excluded from employment or housing but searching actively for both. Practically all these respondents reported looking for work and being unemployed (and not inactive; Appendix Table A.2b). Many had pre-survey work experience: 60% had worked for 5 years or more, and 79% had held a long-term job (6 months or more). If not working at the time of the survey, almost half had moved between employment and non-employment in the previous year. In other words, these people were not disconnected from the labour market.
16 As for housing, 30% had lost their home in the previous year (Appendix Table A.2a). They had pre-survey experience of housing insecurity (living with friends or family, homeless shelter, rough sleeping), but their living arrangements at the time of the survey were generally better than those of the rest of the sample: 30% had been housed for a relatively short time in ordinary housing and 29% in a shelter that remained opened in the daytime. Most of these people were searching for housing: 60% reported search activities, and 36% had applied for social housing. They were still at an early stage of a homeless ‘career’ characterized by long periods of homelessness and a progressive distancing from the support system (Damon, 2002).
2 – Excluded from employment and housing, not searching
17 The second largest cluster, comprising 22% of the survey population, comprises respondents excluded from employment and housing and who, either temporarily or permanently, were not searching for either. It includes many people aged over 50, and 40% are women (Appendix Table A.2c). People with disabilities, retirees, and other inactive people are strongly over- represented. These respondents are among those who received the least support for labour market integration, and almost a quarter had never worked.
18 While a majority had slept rough, sometimes for long periods, and while rough sleepers are over-represented in this cluster, most of these individuals received support (accommodation in ordinary housing or in a centre with daytime opening) because of advanced age, a disability, or their family situation. Regarding both employment and housing, their situation was stagnant and had changed little over recent months. With few employment prospects, these people were also doing little to find an ordinary dwelling.
19 The situations of men and women in this cluster are very different, however. The men, much older than the women, were generally retired and disabled, and 92% were alone and childless. Women are over-represented in the ‘other inactive’ category (often parents with children). Among these parents, 37% were lone-parent families, and 15% were couples with children. The women had much stabler housing situations than the men (55% lived in ordinary housing vs. 11% of men, and only 1% were sleeping rough vs. 22% of men). Accommodation for homeless people with few short-term employment prospects appears divided into two separate channels: on the one hand, women obtain relatively advantageous housing solutions due to their status as a mother or a woman (Loison-Leruste and Perrier, 2019), while on the other, single men are consigned to the most basic facilities or to the street.
3 – Foreigners without work permits, in highly insecure economic and housing situations
20 Comprising just 8% of respondents, all the individuals in this small cluster reported not having a work permit. While excluded from the legal labour market, it is likely they sometimes worked undeclared jobs, but no information on these activities is available.  Their administrative situation is not recorded in the surveys, but we can assume that the majority were undocumented immigrants or asylum seekers. Many were recent arrivals in France, and the proportion living in the Paris region, a migrant entry point (Eloy, 2019), was much larger than among the other respondents. As people most often migrate at young ages, under-30s are over-represented, as are couples with children.
21 Few had a home of their own in France, and their housing situation was especially precarious: 54% had slept rough, and, at the time of the survey, they were over-represented in centres open at night-time only, hotels, and ‘stopgap arrangements’, confirming the difficulty for migrants in obtaining stable institutional support (Dietrich-Ragon, 2017).
22 They had more often held a qualified job in the past than individuals in other clusters (university and high school graduates are over-represented in this category), so they doubtless experienced a radical break in their career when they migrated, with a loss of status after arriving in France (Cordazzo and Sembel, 2016). A large majority received no support in seeking employment. They also had little chance of finding a place to live. Fewer than 10% had an ongoing social-housing application, and only 19% were looking for housing. These low proportions can be explained by their undocumented status and perhaps by their lack of knowledge about available support.
4 – Users excluded from employment but with housing
23 Another cluster, comprising 6% of users of homelessness support services, concerns people with their own home (98% were tenants). These people were most often born in France and generally lived alone. Most were men, and over half were aged over 50. They are under-represented in the Paris region, where low-cost housing is scarce.
24 Despite having their own home, these people had good reason to use support services. Many had had severe housing difficulties in the past, with 40% reporting experience of sleeping rough. They had found a home thanks to social housing, one of the main escape routes from homelessness (Brousse et al., 2008). Almost a quarter lived in rent-subsidized public housing (HLM), and 12% lived in a dwelling managed by an association or an institution (CHRS, hostel, etc.). 
25 Those unable to obtain social housing often lived in substandard, poorly equipped, or even unsafe dwellings.  Many had severe socio-economic difficulties. While a large majority of people with housing held jobs in the past, most were inactive at the time of the survey because of their advanced age or health problems (26% were retired, and 21% had disabilities). This status as an older adult or disabled person, a form of capital recognized on the social welfare ‘market’ (Soulié, 1997; Schlegel, 2017), explains why some could obtain social housing and certain specific benefits targeting older or disabled populations. However, their low incomes were often insufficient to cover the rent (36% had payment difficulties, and 17% had severe difficulties). For people on very tight budgets, attending soup kitchens is a way to ‘make ends meet’ (Marpsat, 2006).
26 As their housing situation was the ‘culmination’ of a long trajectory rather than simply an intermediate stage, these people were not looking for another home, either because they had just moved into their current lodgings, or because they knew they could not afford anything better. Likewise, in view of their age or health problems, most were not looking for work. They received little support from social workers, as they were not in a situation requiring encouragement to find a home or a job.
III – Workers without their own home
27 A quarter of users of homelessness support services were in employment at the time of the survey. Certain features distinguish them from workers with their own home: their job was often obtained with the help of public intermediaries or associations and, above all, their employment conditions were much more insecure (Dietrich-Ragon and Remillon, 2016). These workers can be divided into subclusters based on their employment and housing conditions.
1 – Workers on the margins of employment
28 The first cluster of workers, comprising 7% of the sample, concerns individuals on the margins of employment. They worked for specialized structures that employ vulnerable people (21%), for private individuals (26%), or were self-employed (15%), and 89% did not have a standard employment contract (‘other’ contract). This suggests that they worked either in the informal sector or had a contract of the type used by work integration social enterprises (structures d’insertion par l’activité économique). The share of people in this cluster who had earned money in the month preceding the survey is lower than among other worker clusters, signalling that their work did not always correspond to a paid job, but rather to labour performed in exchange for accommodation. Their accommodation was also more precarious than that of the other workers (stopgap arrangements, centres with daytime opening).
29 The proportion of workers with previous work experience is higher in this cluster than in the others: 90% had already held a long-term job, and 77% had worked for a total of more than 5 years. Some had also lived in stable housing, although half had slept rough, and around the same share had never had their own home in France, reflecting the fact that foreigners are over-represented in this group.
30 This worker cluster is the one with the least favourable housing and employment prospects. Their situation had been stagnant for a long period, and they were not taking steps to improve their prospects: 71% were not looking for a place to live, 51% did not want to move from their current accommodation, and a large majority did not have an ongoing social-housing application. Only 41% were looking for job, although the one they currently occupied was highly precarious and low-paid.
31 This cluster in fact comprises two different profiles. The first is that of single, older men whose characteristics, as described above, suggest that they worked for the centre which gave them a roof. The second is that of foreigners in a very precarious situation for both housing and employment. In both cases, the short-term prospects for improvement were limited.
2 – Workers close to ‘conventional’ employment
32 The next cluster in the typology comprises 11% of the sample and includes workers most integrated in ‘conventional’ employment, of whom 89% worked in the private sector as unskilled manual or clerical workers. More than 80% worked 35 hours or more per week, and half had an open-ended contract. Even so, they did not earn enough to rent a place to live, a situation linked in some cases to the high rental costs in the Paris region where these respondents are over-represented. The majority had already been in long-term employment and had substantial work experience, so they were not looking for another job. Many had moved between employment and non-employment in the last 13 months. Young, low-educated men aged 30–49 and born in France are over-represented.
33 Their housing trajectories are marked by severe marginality. Half had slept rough, and few had spent a long period in ordinary housing in France. Their housing difficulties were partly linked to personal events—one-third of the people in this cluster lost their home after a separation, 7% following domestic violence—and many had experienced difficulties in childhood (parental illness, family conflict, etc.). Some also reported financial difficulties such as eviction due to rent arrears, debts, etc. At the time of the survey, most were living in dwellings provided by an association or in hostels with daytime opening, i.e. the most comfortable forms of accommodation. This is the cluster with the highest proportion of social-housing applicants, as they satisfy the conditions of entitlement and were helped with their application by social workers.
34 In short, after a history of major difficulties, these individuals appear to be on a generally upward housing trajectory that may ultimately lead them to a home of their own.
3 – Workers on short-term contracts
35 The last cluster, comprising 8% of respondents, groups individuals with an especially precarious employment status. Young people, foreigners, women, and lone-parent families are over-represented, as are individuals with above-average qualifications and residents of the Paris region.
36 Contrary to the previous cluster, many of these respondents worked in the non-profit (33%) and public (21%) sectors, where they held skilled and unskilled manual jobs and unskilled clerical jobs. Most were on fixed-term contracts, and a large majority worked part-time. Their employment status was unstable. A majority had moved between employment and non-employment over the last 13 months, and they are over-represented among people with limited overall work experience. Last, 57% were looking for a different job.
37 Additionally, they had rarely lived in ordinary housing in France, or only for short periods, but had less experience of rough sleeping than the other support-service users. At the time of the survey, the people in this cluster were living in the most stable types of accommodation. This gave them access to social support, which is reflected in their active efforts to find a place to live, even though only 43% had applied for social housing.
38 Overall, this cluster comprises respondents who have never been fully integrated in the labour or housing markets but whose sociodemographic characteristics (presence of children in particular) give them greater access to institutional support and hence more encouragement to search actively for employment and housing.
IV – Contrasting trajectories and outcomes
39 Our typology reveals the diverse housing and employment situations of users of homelessness support services. After highlighting the dividing lines between the different user profiles, this last section explores the interdependence between housing and employment situations, and examines how the overall picture evolved between 2001 and 2012.
1 – A multifaceted group
40 Users of homelessness support services form a composite group presenting large contrasts in terms of relative distance from the housing and employment markets. This population cannot be viewed from the simplistic angle of exclusion (Rullac, 2005). Those concerned are sometimes integrated on the labour market, albeit in irregular and precarious jobs, and most move in and out of employment or are actively looking for work. The same is true for housing. However, one feature shared by these people is their highly institutionalized trajectory. They receive institutional support for both housing and employment, and in some cases have never lived or worked independently.
41 This analysis also points up the need to consider past trajectories and future prospects when seeking to understand their contrasting situations. In this respect, the clusters can be differentiated across two key dimensions. The first concerns past integration on the labour and housing markets. People who had a job and a home in the past and have lost them for some reason contrast with those who have never had housing or employment in France. For the former, homelessness is generally not the direct consequence of job loss, but of an accumulation of difficulties such as poor health (Peretti-Watel, 2006), lack of qualifications, relationship breakdown and gender violence (Loison-Leruste and Perrier, 2019), and social isolation (Firdion and Marpsat, 2014). In fact, users who have already lived for at least 3 months in a dwelling that they owned or rented (53% of respondents) rarely mention employment as a factor in the loss of their home (in the 2012 survey, which includes this variable, 13% mentioned this factor). Family problems are much more frequently mentioned, notably separations—the main reported reason for leaving a dwelling in all clusters. In reality, as already pointed out by Firdion and Marpsat (1996), the media figure of the senior manager who loses his job and suddenly finds himself in the street is the exception rather than the rule. Those who have never been integrated on the labour or housing markets are often young people and migrants whose sole life experience in France is that of casual jobs and temporary accommodation, and who are at the start of their employment and housing trajectory. For them, exclusion from housing is not a question of ‘rupture’, but rather of ‘non-integration’.
42 The second dimension separating the clusters concerns the objective prospects of finding employment and housing, and the action taken to that end. Some people face major obstacles (undocumented status, disability) that preclude any short-term improvement in their situation. Moreover, while the variables describing actions taken (enrolment in a job centre, social-housing application, search for a place to live) might suggest that the respondents are acting on their personal initiative, the institutional environment in which they evolve, and especially the type of structure that houses them, play an equally, if not more, decisive role.  In other words, homeless people’s degree of investment in the search for a job or a home largely reflects the type of structure that receives them, depending on their employment situation, nationality, sociodemographic characteristics (Dietrich-Ragon, 2021), and level of vulnerability as reflected in health problems, etc. (Schlegel, 2017).
2 – Interdependence between employment and housing situations: the case of homeless workers
43 We also find that integration and active searching for employment and housing often go together. People with stable employment in the past have also had stable housing; those looking for a job are also searching for an ordinary dwelling. More specifically, a detailed analysis of the position of homeless people on the labour market challenges the assumption that holding a job is no longer a guarantee of housing market integration. First, while the figure of the homeless worker does exist, it corresponds to a very specific set of people in highly precarious work, often in the form of ‘odd jobs’ rather than true employment. Second, for those not defined as vulnerable, i.e. who are neither older adults nor disabled, holding a job is a prerequisite for obtaining social housing (Lanzaro, 2014; Chauvin, 2020).
44 The homeless-worker category is nonetheless a reflection of certain changes in the housing and employment markets, and, in particular, of rising property prices in large cities where a large share of respondents live.  While in other contexts these workers would probably have found a home of their own, they are ‘disqualified’ in a hyper-selective market (Ballain and Jaillet, 1998). This point ties in with Sassen’s observations about ‘global cities’ (such as Paris) characterized by high concentrations of jobs at the extreme ends of the social scale: high-earning professions at one end and low-paid workers on the other, mainly comprising immigrants from countries of the Global South (Sassen, 1996) who often cannot afford high city rents. Thus, the figure of the homeless worker (along with that of the foreigner without a work permit, sometimes in an undeclared job) appears to be a product of our large metropolitan cities. In these particular contexts, certain precarious jobs are now occupied by people with no fixed address and with no housing choices other than institutional accommodation. We see that this institutional system fulfils an important function in cities and on the labour market, as hotels and hostels provide a roof for workers whose income and employment status are incompatible with prevailing property prices and the private rental market.  The institutional accommodation sector can therefore be seen as a sort of welfare benefit for workers on low incomes, provided in the form of low-cost accommodation subsidized by the state.  The ‘precarious worker in institutional accommodation’ thus becomes the paroxysmic figure of the ‘assisted precarious worker’ identified by Paugam and Martin (2009), whose labour income is topped up through welfare benefits and, in this case, though low-cost accommodation. This situation has major implications for the institutions providing accommodation to vulnerable populations. Faced with a shortage of affordable homes for the lowest-income households, they often take in people who have a job and who have the means to pay a reasonable rent (Fondation Abbé Pierre, 2014).
45 This figure of the worker in institutional accommodation also raises questions about whether social support designed to help homeless people find or stay in employment is the right approach to the problem. The fact that a non-negligible share of respondents have a job could stem from an injunction to regain their independence as part of a logic of ‘activation’ to achieve labour market integration ‘at all costs’ (Lanzaro, 2014); employment is seen as key to initiating administrative processes and obtaining a home. In particular, women close to the workers-in-precarious-employment category, often of immigrant origin, are placed under a dual injunction: both to be ‘good’ mothers (Loison-Leruste and Perrier, 2019) if they have children, and to find employment. This often leads them into precarious part-time and low-paid jobs. Efforts to comply with both employment and housing norms (Bresson, 1998) may create other forms of inequality, combining inequalities of social status, gender, and race.
3 – Towards a three-tier system
46 While the clusters seem to remain quite stable over time,  evolving situations come to light when changes within each one are examined (Appendix Table A.3). In the cluster of users excluded from housing and employment who are actively searching, the main change is that respondents were more strongly encouraged to find a job and a place to live in 2012 than in 2001. This is linked to the increasing emphasis placed on ‘activation’ within the social welfare services (Barbier, 2008). These respondents had more often been living in their current accommodation for a long time, doubtless making it easier for them to look for employment and housing in a context of relative stability and with the support of social workers. However, a larger share of respondents in 2012 had never had a home of their own. It looks as if individuals with this profile find it more difficult to gain a foothold in the ordinary housing sector and spend a long time in institutional accommodation where they are encouraged to take steps to move on.
47 The cluster of individuals excluded from employment and housing and not searching also changed. In terms of sociodemographic characteristics, the share of young people fell sharply, while the proportion of over-50s rose from 26% to 47%. This is consistent with research findings that point up increasing precariousness among older adults (Coulomb, 2015). The share of couples with children also increased from 6% to 15%. As in the previous cluster, respondents stayed in the same accommodation for longer periods. The proportion sleeping rough or in a shelter with night-time opening decreased, while the share living in a dwelling and, to a lesser extent, in a hotel, increased. The share of respondents in situations of long-term labour market exclusion (due to their age or family situation) who were living in stable accommodation appeared to stabilize, while homeless people considered fit to work, such as young people, tended to move into the cluster of users excluded from housing and employment and actively searching.
48 In the cluster of foreigners without work permits, the share of couples with children increased from 12% to 21%. This shift towards a population of migrant families has implications in terms of accommodation provision. These families cannot share accommodation with the traditional users of homeless shelters (Le Méner, 2013), so are often housed in hotel rooms where they have no access to educational or social support (Observatoire du Samu Social, 2014). Use of hotel accommodation increased massively: in 2012, 29% of respondents in the cluster were living in hotels, compared with just 3% in 2001. The share of rough sleepers increased from 4% to 10%. Clearly, the situation of foreigners without work permits deteriorated between the two surveys, with an increase in the share receiving no support and in the share accommodated in hotels. Their housing trajectories were also much more precarious. In 2012, 65% of foreigners without work permits had experienced a period of rough sleeping (41% in 2001), which for 17% lasted more than a year (7% in 2001), and they had more frequently been continuously homeless in the year preceding the survey. Conversely, the foreigners without work permits surveyed in 2001 had more frequently lost a home in the year preceding the survey, and a larger number were looking for a place to live. These findings show that in 2012, foreigners without work permits who used homelessness support services were even more precarious than in 2001, and their prospects for improving their situation were limited.
49 The changes affecting the cluster of respondents with housing but excluded from employment were similar to those observed among respondents excluded from employment and housing and not searching, with an increase in the proportions of over-50s (and hence of retirees) and of women. There were also more couples with children and lone-parent families, reflecting the priority status of these categories for the attribution of social housing. In other words, access to stable housing for people excluded from employment appears to be slightly more dependent on criteria of age and family situation.
50 Generally, the relative size of the clusters comprising support-service users in employment fell slightly between 2001 and 2012. A larger share was taking steps to find a place to live and even a new job. The share of individuals in these clusters who had never had a dwelling in France or who had had one for less than a year increased sharply, as did the share of workers who had remained homeless over the previous 13 months, reflecting the increased housing difficulties encountered by precarious workers and their reliance on institutional accommodation. Some signs of improvement in institutional accommodation are visible, however. In the cluster of workers close to conventional employment, the share housed in a dwelling increased. That said, a growing share of workers at the margins of employment (including many foreigners) was living in the most precarious forms of accommodation, hotels in particular.
51 While access to stable housing improved in the majority of clusters, this was not the case for migrants, whose living conditions deteriorated. A growing number of respondents had never had their own home, and many appeared to spend more time than in the past on the margins of ordinary housing; they form an entire population increasingly trapped within the circuit of institutional accommodation. In this context, a three-tier housing system appears to be emerging: an emergency sector that accommodates unemployed migrants and single men; a stable accommodation sector for people with little or no short-term prospect of rejoining the labour market but who belong to categories targeted by housing policies (people with disabilities, retirees, women with children); a social integration sector that provides accommodation, generally on a short-term basis, for people with greater reintegration potential and who are encouraged or required to take steps to move out of the social welfare circuit (Soulié, 1997).
52 Combining information on the housing and employment trajectories of users of homelessness support services provides a means to distinguish subpopulations in terms of their present and past employment and housing situations, but also their integration prospects. We began by describing four groups of users of homelessness support services excluded from the labour market. They form a heterogeneous population with a generally tenuous attachment to the labour market and very contrasting housing and employment prospects that reflect their past trajectories and their sociodemographic characteristics. Three groups of poor workers (at varying distances from conventional employment) excluded from ordinary housing were then identified. While some of these workers were on the margins of the labour market, others occupied unskilled jobs that did not pay enough to cover the high rents charged in large cities.
53 Beyond these contrasting situations at the time of the survey, there is a dividing line between support-service users with respect to past integration on the employment and housing markets: while some had experienced a sudden break or loss of status in their housing and employment trajectory, a large share had never worked or lived in their own home. Another dividing line is that of housing and employment prospects, which are again very heterogeneous. Some people appeared to exist in limbo (migrants especially), while others had more opportunity to improve their situation.
54 Last, the analysis of changes between 2001 and 2012 reveals several different trends. First, the migrant population seems to live increasingly in survival mode, be it for employment or housing. Secondly, individuals’ sociodemographic characteristics appear to weigh equally in the balance alongside employment status for obtaining the stablest housing solutions. In addition, between 2001 and 2012, policies to ‘activate’ people experiencing homelessness and to increase the number of places in shelters and hostels had unequal effects, leading to the reinforcement of a three-tier system that only partially overlaps with employment status. Support for people in situations of long-term labour market exclusion differs from that provided to individuals with ‘reinsertion potential’, or to migrants without work permits, who have seen their living conditions deteriorate.
55 In general terms, this analysis suggests that the problem of homelessness cannot be addressed through accommodation policies alone. Solutions must also take account of policies on employment and migration, two areas closely interlinked. Further surveys are needed, especially to reach people not making use of homelessness support services. Likewise, the situations of non-Frenchspeaking foreigners deserve greater attention. INSEE’s next Sans-domicile survey scheduled for 2025 will doubtless provide new information on the interplay between housing and employment situations among populations living on the margins of the housing system.
AcknowledgementsWe would like to thank Bénédicte Garnier (Statistical Methods, INED) for her methodological guidance, Rose Kedami Kougbla for comparing the two survey questionnaires during an internship at INED, and the reviewers of this article.
Table A.1. Variables constructed to describe housing and employment trajectories
Table A.1. Variables constructed to describe housing and employment trajectories(a) The variables ‘housing trajectory over the last 13 months’ and ‘employment trajectory over the last 13 months’ were constructed using the detailed monthly diary. Only variables that were identical in 2001 and 2012 were included in the analysis. Appendix Tables A.2a, A.2b, and A.3 list the categories of the different variables in detail.
(b) This variable was constructed using the question ‘Do you want to stay in the place where you slept last night?’ Respondents could answer ‘for as short a time as possible’, ‘for a while but you’d like to go somewhere else soon’; and ‘for as long as possible’.
Note: Cells in boldface = highest value for each category. Cells in green = above-average values.
Coverage: French-speaking users of homelessness support services.
Coverage: French-speaking users of homelessness support services.
See the reports of the European Observatory on Homelessness set up by the European Federation of National Organisations Working with the Homeless.
This statistic and all those cited in this article are drawn from the 2001 and 2012 surveys (called Sans-domicile) among users of homelessness services for accommodation and hot meals (INSEE–INED; ADISP [distributor]). As stated by INSEE, ‘a person is considered homeless in the survey if on a given day he or she spent the previous night in a place not intended for human habitation [definition of rough sleeper] or in free or low-cost accommodation’ (hotel or dwelling paid for by an association, room or dormitory in a shelter, etc.).
In Paris, where housing costs are highest, apartment sales prices rose by 185% between 1998 and 2008 (Gallot et al., 2011). According to the Paris region rental price observatory, rents increased by 50% over 10 years.
In 2006, the Enfants de Don Quichotte organization (EDDQ) was created to condemn the living conditions of the homeless, some of whom were led to sleep on the street and die of hypothermia. In December 2006, the organization erected 200 red tents on the banks of the Canal Saint-Martin in the 10th arrondissement of Paris and called on Parisians to support its work for the homeless. On 25 December 2006, EDDQ drafted the ‘Canal Saint-Martin Charter for Universal Access to Housing’, which included the idea of making the right to housing enforceable across France. The media pressure exerted by EDDQ members throughout the country pushed the government to adopt a bill introducing the enforceable right to housing (‘DALO’) and comprising measures in favour of social cohesion.
Non-French-speaking foreigners represent 18% of homeless adult support-service users (Yaouancq et al., 2013). Their origins are very different from those of French speakers, most of whom are African (77%). Non-French speakers are more often from Eastern Europe: 48% are from non-EU countries (particularly Armenia, Russia, Georgia, and Kosovo) and 18% from recent EU member countries (Romania, Bulgaria, and Poland) (Mordier, 2016). As non-French speakers are omitted, the vulnerability of foreign homeless people is probably underestimated (people with a poor command of French are least visible to the administrative services and often do not claim the support to which they are entitled).
While the 2001 and 2012 questionnaires are similar, they do not totally overlap, so for comparative analysis we used variables available in both. When necessary, we recoded and regrouped the response categories to harmonize them across the two surveys.
The term débrouille is used in reference to the economy of débrouille (expediency) and combine (ruse) studied by Pascale Pichon (2010). In this analysis, the term refers to the following living arrangements: people paying for a hotel room, living in a squat, living with family or friends or in a caravan, and a few other exceptional situations.
We began by running two separate MCAs for the 2001 and 2012 surveys. Both produced practically identical types, making it possible to stack the data and obtain a single typology.
The cluster analysis (Ward’s method) is performed on the first 10 factorial axes so that the data can be ‘smoothed’ by eliminating certain random variations and thus improve the partitioning by producing more homogeneous clusters.
The questionnaire included a filter, which meant that these people did not answer the module on employment, but 27% reported having moved between employment and non-employment during the previous year. These jobs provided them with little income, however. People in this cluster had the lowest incomes in the sample.
A large share of these people continued to use the services they frequented while homeless, either for meals or to meet up with old friends and acquaintances (Pichon, 2010).
A quarter of people in private-sector accommodation shared a common toilet, 21% had no shower or bath, 43% had no kitchen, and 48% reported problems of damp.
After controlling for the current situation (in terms of housing especially), individual characteristics have little impact on the probability of applying for social housing (Dietrich-Ragon and Remillon, 2016).
The Paris region is over-represented in the three worker clusters.
This is a recurrent function of poor-quality housing. In the post-war period, substandard dwellings were already used by manufacturers to house their low-paid workers (Duriez, 1979).
This is indeed the original function of social housing: ‘Social housing exists alongside the social security system to complement and improve employees’ incomes, which are still too low; they represent a form of complementary wage’ (Flamand, 2012, pp. 89–90).
The only clusters for which 2012 is over-represented are those of foreigners without work permits and of users excluded from employment who are actively searching.