The population of social housing stock has been the subject of numerous studies based on qualitative surveys which highlight the processes of categorization, sorting and selection of households, likely to produce discrimination and unequal treatment in access to social housing. As quantitative approaches are rarer, existing research struggles to grasp the relationships between the population of the social housing stock, its structure and the characteristics of demand and allocations. This article proposes an original statistical approach to characterize the population of the social housing stock, and to relate it to the structure of the stock on the one hand, and to the characteristics of applicants and beneficiaries on the other. Tested in six French intercommunities, this methodology firstly shows that the structure of the social rental stock does not fully reflect the population of the social housing stock. Secondly, it reveals that certain applicant profiles are under-represented in allocations overall, in relation to their weight in demand. Finally, it shows that, on an inter-municipal scale, recent allocations continue to house low-income households in priority urban districts, although this is not always explained by the rent levels observed.

Based on an ethnographic survey carried out within six social housing organizations, this article reinvestigates the issue of social housing allocations from the angle of the professional practices of those who implement them. It draws on micro-institutionalist theory, combining contributions from street-level bureaucracy and neo-institutionalism, to overcome some of the pitfalls of "bottom-up" approaches to public action, by articulating different levels of analysis and prioritizing explanations of bureaucratic work. From this perspective, the survey highlights the weight of organizational logics, and qualifies the influence of situational and dispositional factors. Indeed, while agents' social properties shape their conceptions of the job and their registers of actions, they are less decisive when it comes to explaining counter behaviour. Similarly, while interactions can modulate certain allocation decisions, there are few such adjustments, and they are contained by the institutional environment.