Many studies use difference-in-differences to identify the impact of an event occurring at a location in space on house prices. Empirical papers often estimate the average effect in a ring around the treatment site. However, in the hedonic model the welfare-relevant parameter is the price derivative with respect to distance. I develop a formal framework for understanding what can be learned about this parameter through difference-in-differences in settings with heterogeneous treatment effects. I show that the derivative of difference-in-differences estimates with respect to distance identifies the relevant parameter under homogeneity but can be biased under heterogeneity. I introduce an alternative assumption. When effects are concave in distance from the treatment site, a natural restriction in many spatial applications, I show a treatment event's welfare impact is bounded. I find in two applications, polluting industrial plants and low-income housing developments, that the estimated bounds are tight.

We characterize the distribution of suburban homeowners' preferences for housing unit density. To measure welfare changes under counterfactual increases in density, we first construct a novel house-level measure of exposure to density and identify its price effects in a boundary discontinuity design. On the borders of municipalities with larger minimum lot sizes, lots are 3,000 ft² larger and houses are $40,000 costlier. We exploit the systematic variation in density exposure induced by these discontinuities to estimate price effects. We then connect these estimates to a structural hedonic model of housing choice to retrieve individuals' preferences for density. Overall, we find an average welfare loss among incumbent homeowners from a 1/2 unit per acre increase in density (equivalent to a 0.3 standard deviation in density) of about $9,500, with significantly larger losses under counterfactual increases solely from rental units. The median welfare loss is only 55% of the average, implying a long, left tail of those with more extreme aversions to density. This tail disproportionately contains households in affluent, low-density neighborhoods.

We investigate the effects of sewer access on neighborhood characteristics in developing-world cities. Because it is more difficult to move sewage uphill than downhill, otherwise similar neighborhoods on opposite sides of drainage basin divides may face different costs of sewer access. We exploit this intuition to identify the effect of sewer access by comparing outcomes for neighborhoods on opposite sides of drainage basin divides. We estimate the effect of sewer access on census tract population density, literacy, and income for Brazil, Colombia, South Africa, Jordan, and Tanzania. On average, sewer access has a large effect on population density and almost none on demographics. These estimates imply that sewer networks are often as important for the economic geography of cities as transportation networks.

I measure the spatial spillovers from increases in allowable housing unit density in the six largest U.S. West Coast cities. I construct a novel dataset of upzoned residential developments by linking historical zoning maps to new construction permits. In most of these cities, zoning maps were extremely persistent over time, leading to few new building developments in upzoned areas. Upzoned development sites decrease nearby single-family house prices by around 4% relative to units slightly farther away, but effects are extremely localized. I find little effect outside of 200 meters from these developments, with most of the effect occurring within 100 meters. Homes near bigger developments have larger price declines.

A Stata package to inflate time series to real dollars. Available on the Boston College Statistical Software Components (SSC) archive.