Policy, Place, and Unsheltered Homelessness: Essays on Mobility, Measurement, and Spatial Evaluation

Colin Sides

Advisor: Michael A Clemens, PhD, Department of Economics

Committee Members: Carlos D. Ramirez, Peter J. Boettke

Online Location, #https://gmu.zoom.us/j/98214935408?pwd=LdE3jm033BaZPBq3s8vzPnYasPAd9p.1
May 20, 2026, 08:30 AM to 10:00 AM

Abstract:

This dissertation examines how unsheltered homelessness is measured, distributed, and evaluated when people experiencing homelessness are spatially mobile and administratively counted through place-based systems. Place-based counts may reflect not only changes in homelessness itself, but also changes in where people reside, seek services, remain visible, or are counted. Across three connected essays, the dissertation argues that mobility, measurement, and spatial context are not secondary complications but central features of what place-based homelessness counts actually measure.

Drawing on the housing economics literature, mobility evidence, and spatial econometrics, the first essay develops a mobility-centered framework for homelessness policy evaluation. It argues that when policies alter service access, shelter availability, enforcement regimes, or the relative attractiveness of jurisdictions, conventional place-based estimates can misstate policy effects. A local increase in homelessness may reflect inflows rather than program failure, while a local decrease may reflect displacement rather than genuine reduction. The chapter outlines design principles for mobility-aware evaluation, including exposure mapping, neighbor comparisons, localization tests, and joint interpretation of treated and surrounding jurisdictions.

The second essay estimates local magnet effects within an urban homelessness system. Using tract-level data and a Bartik-style instrumental variables design, the chapter asks whether existing concentrations of unsheltered homelessness increase future unsheltered homelessness in the same area. A location-choice model formalizes the mechanism: unsheltered individuals may be drawn to locations where others are already present through information, safety, mutual aid, or access to services. The estimates imply economically meaningful local persistence and attraction, suggesting that place-based reductions may generate spillover benefits beyond the individuals directly housed or relocated.

The third essay examines whether winter climate helps explain the geography of unsheltered homelessness. Combining HUD Point-in-Time data with January temperature measures, the chapter finds that warmer winter conditions are associated with higher observed unsheltered homelessness, especially in Ninth Circuit and California contexts. However, the evidence does not show a clean break associated with Martin v. Boise, the Ninth Circuit ruling limiting anti-camping enforcement. Winter climate appears to shape where unsheltered living is more feasible and where unsheltered homelessness is more likely to be observed, operating as a contextual condition rather than a singular cause of geographic variation.

Together, the essays show that unsheltered homelessness cannot be fully understood through local counts alone. Place-based outcomes are spatially embedded measures shaped by mobility, environmental exposure, institutional context, and administrative visibility. The dissertation’s central contribution is to clarify how homelessness policy evaluation changes when the location of homelessness is itself part of the outcome.