Association for Public Policy Analysis & Management

During the past two decades, public and scholarly discourse around antipoverty policy has shifted from an emphasis on providing income subsistence to one of asset development. Since the mid-1980s, multiple generations of U.S. Department of Housing and Urban Development (HUD) programs have tried to blend housing assistance with a variety of supportive services designed to improve the economic wherewithal of recipients. In this so-called “housing plus” approach, the provision of basic shelter is augmented by services to support both resident families and the larger community development initiatives in which subsidized housing in located.

The primary programmatic manifestation of this reoriented subsidized housing policy has been the Family Self-Sufficiency (FSS) program, authorized in 1990. Our research conducts an impact evaluation of a variant of the FSS program: the Denver Housing Authority’s (DHA) Home Ownership Program (HOP). To assess impacts, we rely upon parameter estimates from quasi-experimental methodologies that permit one to draw causal inferences with confidence provided that treatment and control groups share a common distribution and are balanced on variables that influence selection. Our identification strategy is compares two levels of treatment. Our “control” group consists of participants exposed to a low or moderate intensity of the HOP. The low intensity group qualified and enrolled for HOP but completed less than 12 months of the program. The moderate intensity group enrolled and completed 12 months or more of the HOP program but did not move to the second phase, additional programming called the Home Buyer’s Club (HBC). We combined these two groups due to small sample size. The “treatment” group includes “high intensity” participants who completed HOP and participated in the HBC.

We evaluated the impact of HOP on four outcomes: 1) earnings growth during HOP, 2) earnings growth after HOP, 3) self-sufficiency, and 4) homeownership. We modeled the transition from low-moderate controls in HOP to the treatment group in HBC using eligibility variables (e.g., savings, employment) and individual characteristics (e.g., gender, race). The earnings outcomes constituted a sub-sample. A caliper of 0.5 standard deviation was set to drop the tails of the earnings impact variables after HOP because they did not share a common distribution in the propensity score. The impact parameter is called the average treatment effect on the treated, estimated by a difference in means across matched samples. Because the earnings outcomes did not have covariate balance, we adjusted estimates using differences in differences regression to address remaining bias.

HOP participants who were intensely treated, gained considerably in all four realms compared to matched low-intensity and moderate-intensity control groups. The HOP high-intensity treatment increased annual earnings during HOP by $3,213, increased annual earnings after HOP by $4,523, increased the probability of becoming self-sufficient by 0.42, and increased the probability of becoming a homeowner within five years of program enrollment by 0.29. All these impact parameters proved statistically greater than zero at the .01 significance level. We draw implications of our findings for the design of FSS strategies.