Essays in Urban Economics

Abstract

This dissertation is comprised of three essays, each dealing with topics in Urban Economics and Applied Microeconomics. In the first chapter, I use a dynamic discrete-continuous choice model to examine the decision to invest in home improvement. In each time period, households face the decision of whether to reoptimize their housing consumption, either through home improvement or by selling their property. The structural model is estimated using a uniquely rich micro-level dataset that encompasses each property in Minneapolis for a period of almost 20 years. I reconstruct the optimal investment policy and choice probabilities as a function of a property's housing quality and neighborhood quality. Then, I explore a counterfactual scenario in which the cost of investment in home improvement is subsidized and show that such a policy would be effective in increasing the predicted level of investment. I find that policies aimed at encouraging home improvement can be a cost-effective tool to leverage private investment in housing renovation, and to promote urban revitalization without displacing the residents of low income neighborhoods. In the second chapter, I investigate how the development of new apartment buildings can affect local property prices. On one hand, increasing the supply of available residential units is expected to lower the price of other housing options in a given area through a substitution effect. On the other hand, apartment building development may produce aggregation economies and other spillovers increasing the desirability of a given neighborhood and in turn property prices. Estimating the net effect under a standard parametric framework is complicated by the non-linear interaction of geographic and temporal distance from the site of construction. I apply a new econometric technique developed by Diamond and McQuade (2019) to non-parametrically estimate the effect of apartment building construction on nearby residential property prices, transforming transaction prices into a price gradient that is a smooth function of time and distance and then numerically integrating over the estimated derivatives to measure changes in property prices. I find that property prices increase with distance from the site of a new development, suggesting that the substitution effect might be stronger than other potential spillovers, although the overall effect varies heterogeneously across different types of neighborhood. In the third chapter, Mark Ponder and I examine the effect of the introduction of light rail transit in Minneapolis. We focus on decomposing the overall impact on local property prices to assess what share is attributable to the direct effect of improved access to public transit and what share is attributable to an indirect spillover effect through the increase in local amenities. After assembling a rich spatial dataset encompassing every residential property in Minneapolis and hundreds of thousands of businesses and neighborhood amenities, we use machine learning techniques to estimate a hedonic pricing surface. We extend the method of Boosted Smooth Trees introduced by Fonseca et al. (2018) to a high-dimensional dataset and to incorporate instrumental variables, allowing us to control for endogenous changes in amenities. Our results indicate that the price of properties located within a half mile of a light rail station increased by around 11.3%. The direct impact of access to the light rail itself is estimated to increase local housing prices by 5.5%, while the estimated spillover due to changes in amenities is quantifiable at 5.8%.