How homelessness should be defined is a fundamental and persistent problem. Relatively little progress has been made toward achieving international agreement in the twenty years since Greve and Currie (1990, p. 28) wrote:“what constitutes ‘homelessness’ and how many people are homeless is a debate which has been running for thirty years or more”. A robust definition of homelessness is a necessary basis for the production of meaningful statistics on the size and characteristics of homeless populations, which are of critical importance for informed policy-making. A definition of homelessness can be judged useful if it allows for accurate and reliable identification and classification of homeless people so that policies can be developed to respond to different manifestations of homelessness and monitor the effectiveness of such interventions. At a more basic level, evidence of the size of homeless populations can play a pivotal role in determining whether the problem is included on a government’s policy agenda in the first place:“it becomes difficult to urge governments to meet the needs of homeless people if the parameters of the homeless population are unclear”(Chamberlain and MacKenzie, 1992, p. 274).

Definitions of homelessness vary considerably across the world and few have a conceptual basis. Definitions produced by government agencies with responsibility for addressing homelessness tend to minimise the population and concentrate on those who are publicly visible. Advocates and non-government service providers, on the other hand, who regard the definition as “the connecting link between the problem of homelessness and agency responsibility”(Minnery and Greenhalgh, 2007, p. 652)(as well as the link to funding), tend to favour broad definitions that maximise the number of people identified as homeless, often by conflating people at risk of homelessness and those who are actually homeless (Widdowfield, 1999). These different framings perform certain functions, but they are unlikely to provide a valid basis for producing accurate homelessness statistics. Hutson and Liddiard (1994, p. 32) observe:“because different professionals have different definitions of homelessness, so they also produce different statistics. In this way, statistics can tell us more about the organisation collecting them than about the phenomena that are being measured”.