Association for Public Policy Analysis & Management

A fundamental debate in the design of government programs is whether the government should provide goods or services directly or whether the government should simply subsidize consumers to purchase goods or services on the private market. Whether the government or the private market, receiving subsidy, should be responsible for providing certain basic goods or services to low-income populations continues to be debated in arenas such as primary education, child care, and veterans’ health care. In the case of low-income housing policy, the argument concerns whether government-run project-based rental assistance, public housing, or housing vouchers (formerly section-8) is better at delivering housing stability, neighborhood quality, and favorable outcomes for children and adults.

Low-income housing programs serve more than 9 million persons and total about $50 billion annually in spending. Approximately 40 percent of this is devoted to tenant-based housing vouchers, with the remainder funding several different project-based subsidies, including public housing. The merits of housing vouchers relative to public housing have long been argued by economists with considerable attention devoted to which program is more costly per family served (Olsen 2003). Quantifying the respective benefits of each program has received less attention, with the notable exception of recent evidence on housing vouchers (Jacob, Kapustin, and Ludwig 2015). Critics of voucher policies charge that they unnecessarily subject tenants to the vicissitudes of the private rental market and fail to serve the most disadvantaged households. Advocates of housing vouchers argue that they deliver more housing choices for tenants and don't further concentrate low-income households in poor neighborhoods.

I investigate the effects of public housing admission and housing voucher receipt on housing stability, neighborhood quality, and outcomes for children and adults using a novel housing assistance lottery in Miami Dade County (FL) which generated random offers of public housing or housing vouchers to sample of roughly 70,000 disadvantaged households. The structure of the lottery allows me to estimate both the effects of either type of assistance (voucher or public housing) separately relative to “no assistance,” as well as the comparative effect of receiving one form relative to the other. Using data from this unusual housing assistance lottery linked to administrative data on homelessness, educational outcomes, criminal justice involvement, and health, I estimate the separate and comparative effects of public housing and housing vouchers on a range of important outcomes for up to 8 years after the initial lottery.