No abstract

We investigate determinants of driving speed in large U.S. cities. We first estimate city-level supply functions for travel in an econometric framework where the supply and demand for travel are explicit. These estimations allow us to calculate an index of driving speed and to rank cities by driving speed. Our data suggest that a congestion tax of about 3.5 cents per kilometer yields welfare gains of about $30 billion per year, that centralized cities are slower, that cities with ring roads are faster, and that the provision of automobile travel in cities is subject to decreasing returns to scale.

This paper documents and explains the striking rise in the proclivity of college-educated individuals to reside near city centers since 2000. We show that this recent urban revival is driven almost entirely by younger college graduates in larger cities. With a residential choice model, we quantify the role of jobs, amenities, and house prices in explaining this trend. We find that the rising taste of young college graduates for non-tradable service amenities like restaurants and nightlife accounts for more than 40 percent of their movement toward city centers. Complementary data shows a corresponding rise in young college graduate expenditures on and trips to non-tradable services. We then link changes in both consumption and urbanization to secular trends of top income growth and delayed family formation amongst young college graduates.To measure consumption amenity density, we pair a geo-coded census of establishments in 2000 and 2010 from the National Establishment Time-Series (NETS) with a dataset containing travel times between these establishments and census tract centroids by foot from Google Maps. 8 We calculate indexes measuring four types of consumption amenities: two non-tradable services (restaurants and nightlife) and two types of tradable retail (food and apparel). We also measure consumption amenity diversity as an inverse-Herfindahl index using the most refined industry classification available in the NETS (at the SIC8 level, e.g., Korean restaurants). Finally, we use the smartphone visit data, described further in Couture et al. (2019), to calculate an amenity quality index that captures the presence of restaurant chains preferred by a given age and education group.Our primary measure of housing costs for 2000 and 2010 is the Zillow House Value Index for two-bedroom homes, which measures median house prices at the zip code level. In robustness checks, we use alternative house price indices, rental prices using HUD's Fair Market Rent Series for one-, two-, and three-bedroom homes (available at the county level), and the median age of the housing stock from the 2000 census and the 2008-2012 ACS, to measure one aspect of housing quality and new housing developments. 9 We complement these three main tract-characteristic datasets with information on public amenities (transit times, violent crime per capita, school district rankings) and natural amenities. Our measure of transit performance at the tract-level comes from Google Maps in 2014, and is the average travel time of a five-mile trip from a tract centroid to a random set of NETS establishments. We measure violent crime (murder, rape, robbery, and aggravated assault) at the police district-level using the Uniform Crime Reporting Program (UCR) data for 2000 and 2010. We measure school quality using within-state rankings of school districts in 2004 and 2010 from SchoolDigger.com. 10 There are typically multiple tracts within a particular police and school district. We match these areas to 2010 tract boundaries using Census shapefiles. 11 8 The p...

No abstract

This paper estimates the impact of the first nationwide e-commerce expansion program on rural households. To do so, we combine a randomized control trial with new survey and administrative microdata. In contrast to existing case studies, we find little evidence for income gains to rural producers and workers. Instead, the gains are driven by a reduction in cost of living for a minority of rural households that tend to be younger, richer, and in more remote markets. These effects are mainly due to overcoming logistical barriers to e-commerce rather than additional investments to adapt e-commerce to the rural population. (JEL I31, L81, O12, O18, P25, P36)