Abstract: Lower income commuters are more likely to ride and reside near public transit within cities, but do they also benefit more from faster transit travel? Combining survey data on travel behavior with web-scraped data on roads, establishments and counterfactual travel times for millions of trips across 49 large US cities, I estimate a structural model of travel mode and residential location choice. I characterize the heterogeneity across income groups and cities in commuters' willingness to pay for access to faster transit and the expected increase in transit ridership in response to marginal transit improvements. I find that richer transit riders sort more aggressively into the fastest transit routes and are, on average, willing to pay more for faster commutes. Improvements in transit speed are most effective at generating transit ridership and welfare gains where transit is already fast (relative to driving), in cities with a greater share of rail-based transit and where the gains are larger for higher-income commuters. Transit improvements benefit low-income commuters more where transit is already relatively slow, in cities with more bus transit, and where the overall marginal gains are small.
Abstract: What are the implications of mass transit improvements for residential income segregation within cities? This paper models a stylized city where heterogeneous households choose where to live and how to travel given a spatial distribution of travel times and a competitive housing market. I characterize when and where marginal improvements in transit access reduce income segregation instead of exacerbating it. I show that a planner trying to maximize the city's transit ridership is incentivized to improve low-speed transit (e.g. buses on shared lanes) where it reduces income segregation but improve high-speed transit (e.g. subways) where it increases income segregation. These results are consistent with recent changes in transit ridership and neighborhood incomes in US cities.
Abstract: We develop a methodology to estimate robust city-level vehicular mobility indices, and apply it to 154 Indian cities using 22 million counterfactual trips measured by a web mapping service. There is wide variation in mobility across cities. An exact decomposition shows this variation is driven more by differences in uncongested mobility than congestion. Under plausible assumptions, a one-standard-deviation improvement in uncongested speed creates much more mobility than optimal congestion pricing. Denser and more populated cities are slower, only in part because of congestion. Urban economic development is correlated with better (uncongested and overall) mobility despite worse congestion.
Abstract: Housing is the most important asset for the vast majority of American households and a key driver of racial disparities in wealth. This paper studies how residential segregation by race eroded Black wealth in prewar urban areas. Using a novel sample of matched addresses from prewar American cities, we find that over a single decade rental prices soared by roughly 50 percent on city blocks that transitioned from all White to majority Black. Meanwhile, pioneering Black families paid a 28 percent premium to buy a home on a majority White block. These homes then lost 10 percent of their original value as the block became majority Black. These findings strongly suggest that segregated housing markets cost Black families much of the gains associated with migrating to the North.
Abstract: We provide a novel approach to estimate the deadweight loss of congestion. We implement it for road travel in the city of Bogotá using information from a travel survey and counterfactual travel data generated from Google Maps. For the supply of travel, we find that the elasticity of the time cost of travel per unit of distance with respect to the number of travellers is on average about 0.06. It is close to zero at low levels of traffic, then reaches a maximum magnitude of about 0.20 as traffic builds up and becomes small again at high levels of traffic. This finding is in sharp contrast with extant results for specific road segments. We explain it by the existence of local streets which remain relatively uncongested and put a floor on the time cost of travel. On the demand side, we estimate an elasticity of the number of travellers with respect to the time cost of travel of -0.40. Although road travel is costly in Bogotá, these findings imply a small daily deadweight loss from congestion, equal to less than 1% of a day’s wage.
Abstract: Despite the “1/N problem” associated with profit sharing, the empirical literature finds that sharing profits with workers has a positive impact on work team and firm performance. We examine one possible resolution to this puzzle by observing that, although the incentive to work harder under profit sharing is weak, it might be sufficient to motivate workers to report each other for shirking, especially if the workers are reciprocally-minded. Our model provides the rationale for this conjecture and we discuss the results of an experiment that confirms that profit sharing is most effective when peer reporting is possible.
Selected Research in Progress
Mobility and Congestion in World Cities: Evidence from Google Maps with Victor Couture, Gilles Duranton and Adam Storeygard(see slides)
The Impact of Public Transit on Congestion and Pollution: Evidence from Jakarta’s MRT with Arya Gaduh, Alex Rothenberg and Yao Wang
Public Transportation and the Rise of the Segregated Metropolis in the United States with Allison Shertzer and Randall P. Walsh
Public Transit and the Working Poor with Jason Cook, Sierra Hall and Hugh Macartney
Abstract: Suppose a set of sites on a grid are initially infected (by some disease). Once infected, a site remains infected forever and a new site is infected if at least r = 2 of its neighbours are infected. Given n initially infected sites, is it possible to distribute them on an n-by-n grid such that the entire grid is eventually infected? Is this the minimum number of infected sites to have this property? Furthermore, what is the minimum number of initial infections needed to eventually infect the n-by-n grid if r = 3 or r = 4? Suppose a grid G is eventually infected given a set A of initially infected sites. How large can A be so that no proper subset of A will eventually infect G? This paper explores contemporary combinatorial works on Bootstrap Percolation theory to answer the questions above.