Essays in Urban Economics

Mobile devices and online services allow capturing an unprecedented amount of information about human behavior. In this dissertation, I use these new types of data to understand how the built environment affects social life and businesses in cities. In Chapter 1, I provide the first causal evidence that the provision of urban parks promotes opportunities for racially and ethnically diverse encounters. Utilizing a novel dataset featuring individual GPS tracking data for more than 60 thousand Twitter users in the New York metro area, I introduce a measure of racial diversity that captures one's level of exposure to diverse others in places visited daily. My empirical strategy relies on using the variation in the timing of park construction works across the city (that temporarily limit the capacity of said parks) to identify the impact of the effectively accessible parkland area on the individual exposure to racial diversity. My results show that for White and Black/African American residents additional 10 acres of parks within a 5 km radius from home increase individual chances of encounters with the members of other groups by 1 p.p. The effect is sizable: for reference, transitioning from the current state to the random mixing scenario would require a 9 p.p increase in diversity for an average Black or African American individual and a 3.5 p.p increase for an average White person. I also provide evidence to suggest that park accessibility affects the diversity of White and Black residents differently: for parks located closer to home, the effect appears to be more pronounced for Whites than Blacks. Chapter 2, written jointly with Dmitry Sedov, investigates the role of sports facilities in generating consumption spillovers for the local businesses. The construction of sports facilities, which can cost hundreds of millions of dollars, is often subsidized by public sources. In many cases, subsidies are allocated on the premise that sports venues benefit the local economy by bringing new customers to nearby businesses. We pin down the size and the spatial distribution of such spillovers using daily foot traffic data from mobile phones covering major sports league facilities and the surrounding commercial establishments. By employing the fixed effects and the IV estimation strategies, we show that the spillover benefits are heterogeneous across sports and business sectors. We find that 100 baseball stadium visits generate roughly 29 visits to nearby food & accommodation businesses and about 6 visits to local retail establishments. While the estimates for football stadiums are comparable, basketball & hockey arenas do not appear to generate significant spillovers for the surrounding businesses. Using our spillover estimates, we also compute an upper bound on the additional local spending induced by each sample arena. The median value of the additional spending turns out to be substantially smaller than the corresponding median subsidy to sports facilities in our sample. In Chapter 3, I examine the contribution of parks to social ties between neighborhoods in New York City. Although the role of public spaces in facilitating social interactions in cities has been widely discussed by social scientists and urban design scholars, data sets from online social networks present unexplored opportunities to quantify this link on a larger scale. I use data on friendship links between Facebook users across New York City zip codes to show that two neighborhoods with a higher density of green spaces between them are more likely to have stronger social ties. In particular, when controlling for demographic differences and zip-code level fixed effects, I find that a 1 p.p. increase in the percentage of land allocated to parks between two given zip codes is associated with a 1.2% higher chance of online social connection between their residents. Comparing the effects of park density for different types of parks, I further document that the presence of community parks, flagship parks, and playgrounds are all significant predictors of higher social connectedness between zip codes. Notably, the largest estimated effect is for playgrounds, indicating a 33% higher probability of connection per 1 p.p. increase in density