Neighborhood Selection of Public Housing Residents in the Housing Choice Voucher Program: Quasi-Experimental Results from Chicago
Date of Original Version
Abstract or Table of Contents
Millions of families are supported nationwide by housing subsidies which have traditionally tied them to a place - a public housing unit. Based on promising results from Gautreaux program and midterm evaluation of national MTO experiment, it is deemed reasonable by the housing policy researchers to relocate all households in public housing projects to rent-subsidized units in open market via Housing Choice Voucher Program (HCVP). Policy design based on voucher-based housing subsidies requires detailed knowledge of the preferences and relocation choices of subsidy recipients. However, little is known about these decision processes. In this paper we seek to understand the intricacies of family relocation decision given an opportunity to use housing vouchers. Data for this study are derived from an initiative in Chicago dating from 1997 in which families on the waiting list for the then “Section 8” program were purged in a management review and the waiting lists repopulated. Families on the waiting list were then chosen to receive Section 8 vouchers via a lottery.
We use a logit model to identify correlates of destination outcomes. Unlike most studies on housing relocation, our choice set is comprised of census tracts, based on a belief that families originating in public housing have limited exposure of distant destinations and therefore do not take decisions based on larger aggregates. Consistent with our belief, we find that a focus on Census tracts leads to large but tractable models, and that most households relocate to tracts close to their origin address. We find that age, sex and employment among others affect the relocation choice. Interestingly we find that differences exist in neighborhood selection of public housing households when compared to other households in the same program. We also find truth to the claims of geographic clustering.