Rethinking Support Allocation: Mental Health and Patterns of Homelessness Support Utilisation

ABSTRACT Homelessness support agencies face substantial puzzles in allocating scarce resources: who gets access and who stays on the waitlist? This paper examines the influence of mental illness on the patterns of consuming non‐clinical homelessness support by persons experiencing homelessness and at risk. We used data from the Initial Assessment and Planning agencies, the ‘front door’ of Melbourne's homelessness system. We observed three main patterns of user behaviour over a 1096‐day period: light (P1), periodic (P2) and regular users (P3). A person presenting with a mental health issue during the first presentation is 3.2 percentage points less likely to be a P1, but 1.5 and 1.8 percentage points more likely to become a P2 or P3. While the results are consistent across different model specifications, in practical terms, the influence of mental health on the trajectory of support use is arguably minimal. The stronger predictors are structural (e.g., being a First Nation Australian) and financial. A client receiving paid income is more likely (by 13.3 percentage points) to become a light user. We conclude with implications for homelessness support and policy, reflecting on Goodhart's law and the plausible category error in pursuing sophisticated vulnerability assessments.