The determinants of homelessness and the targeting of housing assistance

This study combines data from the Survey of Income and Program Participation with the National Survey of Homeless Assistance Providers and Clients to estimate a conditional probability model of homelessness. The results suggest which factors are important predictors of homelessness and argue that gender, age, and race of the household head, are important determinants of whether a household is homeless. Using the results of the model predicting homelessness to simulate the effects of the removal of housing subsidizes from the subsidized population indicates that subsidized housing could be targeted better toward those most at risk of being homeless. (C) 2004 Elsevier Inc. All rights reserved.