Homelessness and Incarceration: A Reciprocal Relationship?

Objectives: Examine whether exits from incarceration lead to homelessness and whether homelessness leads to incarceration. Methods: This paper uses a unique longitudinal dataset which follows disadvantaged Australians over 2.5 years and provides very detailed information on their housing circumstances. Although studies consistently report a positive association between incarceration and homelessness, little is known about the causal relationship between them. We advance in that direction by exploiting the longitudinal dimension of our data in two ways: (i) employing individual fixed effects models to deal with time-invariant unobserved heterogeneity; (ii) lagging key independent variables to minimise reverse causality issues. Results: Our results show that homelessness does not increase the risk of incarceration. In contrast, incarceration does increase the probability that an individual will become homeless, but not immediately. Exploiting details of the accommodation calendar 1-24 months after release, we find a modest immediate effect of incarceration on homelessness (a 3 percentage points increase), which increases 6 months after release (to around 12 percentage points) and persists for a further 11 months with respondents most often staying in precarious housing arrangements (boarding houses or with friends with no alternative) rather than becoming literally homelessness. Conclusions: Our study shows the importance of having adequate coverage for post-release programs to break the link between incarceration and homelessness. Specifically, we find that the critical period for ex-inmates starts 6 months after release suggesting that this may be the time when support programs are currently lacking and would be most efficient.