Association for Public Policy Analysis & Management

According to the Annual Homeless Assessment Report (AHAR), 578,424 people were homeless on a given night in January 2014. Typically, resources for the homeless have centered on shelter and transitional housing; however, an innovative housing service has emerged in recent years. Rapid Re-Housing (RRH) provides time-limited financial assistance and case management to help persons who are homeless move quickly into housing, minimizing the time they spend being homeless and decreasing the risk of returning to homelessness. There is an urgent need for an evidence base to support the national policy approach on RRH, as it is one of the Department of Housing and Urban Development’s (HUD) top strategies for ending homelessness.  To date, there have been very few studies addressing the impact of RRH on future housing stability, length of shelter stay, and risk of repeat homelessness. Applying rigorous rapid cycle evaluation methods to understanding the effects of RRH will inform a broad range of national and local policies addressing the prevention and elimination of homelessness.  This study utilized innovative evaluation methods to understand what would have happened to RRH households if they had not received the RRH intervention.  Homeless Management Information System (HMIS) datasets were accessed in each of five sites across the United States (State of Iowa, San Diego, Phoenix, Memphis and Oakland County, Michigan) and a propensity score matching (PSM) method was applied to create comparison groups of households that received RRH and households that look the same but did not receive RRH. The goal of the PSM process was to create treatment and matched comparison groups that have no visible pre-treatment differences.  The PSM model included demographic variables that measured race/ethnicity, age, sex, education, marital status, and number of children, as well as contextual variables including disabling conditions, income levels, and receipt of TANF, SSI or SSDI. Matched on these variables, the treatment and comparison groups were expected to be statistically similar on each of these critical dimensions. Individual propensity scores were calculated and study households were matched by their probability score for enrollment into RRH.  Specification tests were run to indicate whether the matching procedure succeeded in balancing all covariates.  Return to homelessness and length of time in homelessness were estimated for each group. T-tests of differences in means of characteristics at the time of program enrollment assessed whether the propensity score matching model resulted in strong matching outcomes.  Results showed a significant reductions in households returning to homelessness among those who received RRH compared to those who did not.  Length of stay in shelters was also significantly less among those who received RRH compared to those who did not.  This study demonstrates the impact of RRH from five different sites across the country and strengthens the evidence-base of RRH’s effectiveness in ending family homelessness.  Policy-makers and social impact investors can use this data to help support to the United States Interagency Council of Homelessness’s goal to end all family homelessness by 2020.