Competition and homelessness

The homeless are among the worst off members of society - not only are they extremely poor and often detached from family and community, but they also disproportionately suffer from a range of vulnerabilities including mental illness, developmental disabilities, domestic violence and substance abuse. And while alleviating poverty is argued by many economists as one of the most important goals of their discipline, the poorest of the poor in the United States receive little attention in economic research. If we take seriously the incentives of homeless individuals and those who serve them, how can economics help us better understand and improve the efficiency of homelessness assistance? How does competition among service providers and governments affect the provision of services to the homeless?;The homeless shelter market, together with other social services in the United States, is comprised of many service providers which are paid trillions of dollars annually by governments and private donors to help other people. A vital and frequently debated question is how to reward service providers in a way that maximizes performance. In the first chapter, I present a model of social service markets with a key distinguishing feature---providers compete for users to help by adjusting service quality. For example, homeless shelters must bid up service quality for homeless individuals in order to attract them to their shelter. I show that popularly used performance-based incentive schemes---relative to optimal incentive schemes---distort the allocation of service quality across heterogeneous users, thus reducing aggregate performance and resulting in potentially undesirable distributional outcomes. Applied to homeless shelters, the model explains new evidence of spending patterns across users, and when applied to teacher and physician markets with intrinsic incentives, it offers a new explanation for longstanding puzzles of variation in both teacher and physician quality. Having shown in the first chapter that competition for users undermines popular incentive schemes for social service providers, in the second chapter I quantify this effect in a major segment of the homeless shelter market. Using novel data from federal funding applications by transitional housing programs for the homeless, I estimate a homeless production function that allows for simulation of housing outcomes under several popular performance-based incentive schemes. I find that popular incentive schemes which do not account for shelter competition for homeless people reduce aggregate outcomes by more than 1% relative to optimal schemes. Distributional effects are more pronounced, with the lowest income users receiving 13% less spending under incentive schemes which do not account for competition.;In the third chapter, I turn away from service providers which compete to attract the homeless, to communities which compete to drive them away. Survey data indicate that homeless individuals are highly mobile, and thus, communities are incentivized to substitute away from long term shelter (where lengthy stays are guaranteed) to emergency shelter (where stays are granted on a nightly basis) so as to preserve the possibility of free decreases in the number of unsheltered via out-migration. I show that even when out-migration by unsheltered homeless individuals is random, communities inefficiently substitute from long term shelter to emergency shelter, and furthermore, decrease the total number of shelter beds. Using national survey data on individuals experiencing homelessness, I document a high degree of mobility, with additional evidence consistent with long term shelter forming ties with the homeless, providing support for the key feature of the model.