Housing the Poor
Author: Scott Susin
Dissertation School: University of California, Berkeley
Pages: 172
Publication Date: April 1999
Availability:
Available from the HUD USER Helpdesk P.O. Box 23268 Washington, DC 20026-3268 Toll Free: 1-800-245-2691 Fax: 1-202-708-9981 Email: oup@oup.org
Access Number: 10730
Abstract:
In this dissertation, I examine two major policy reforms in subsidizing housing programs. First, housing subsidies have increasingly shifted towards providing subsidies for low-income households to use in the existing housing stock, rather than subsidizing new construction. The first section of this dissertation investigates whether this policy has raised rents for unsubsidized poor households, as many analysts predicted when the program was conceived. The main finding is that demand-side subsidies of the existing housing stock have increased the price of housing for unsubsidized poor households by 13 percent. Second, policymakers have been increasingly interested in providing (or requiring) education, training, job search services intended to shorten stays in subsidized housing, and push recipients into jobs. The second section of this dissertation examines the routes through which recipients exit subsidized housing. The main finding is that employment has a relatively modest effect on the length of housing subsidy spells. Marriage, and other changes in household composition appear to play a more important role. Finally, a simple but important question is just how long stays subsidized housing last. Several studies have suggested that spells are quite brief, with most recipients leaving within a year. The third section of this dissertation shows that even very small amounts of measurement error, which seems to be present in the data I examine, can cause large downward biases in estimates of spell durations. This problem potentially affects any study that uses duration models. I propose a simple technique that greatly reduces the bias in duration models estimated in the presence of classification error. This technique does not require knowledge of the magnitude of the measurement error, or bias the results when measurement error is not present.
Back to Search Result of DDRG Dissertations
