UC Berkeley

The transportation sector is the largest contributor to anthropogenic U.S. greenhouse gas(GHG) emissions. If the U.S. is to address climate change, a key mechanism will be through a transition to a decarbonized transportation sector. Given environmental externalities, it is clear that the private market will produce too little low-carbon transportation infrastructure. Policy-makers are attempting to overcome this scale problem by encouraging the build out of alternative transportation infrastructure. In this dissertation, I examine existing clean infrastructure policies — the spatial efficiency of infrastructure installed to-date, their effect on adoption of the new transportation mode, and their effect on other transportation modes.

In the first chapter of this dissertation, I study the location of electric vehicle chargingstations. Absent policy, it is well understood that EV charging network size may be inefficient because of pollution and network externalities. This paper argues that there will also be spatial inefficiencies in the location of electric vehicle fast charging stations. I empirically test for the presence of a spatial inefficiency in the free market by comparing the location decisions of a vertically integrated firm that sells vehicles and provides a charging network to those of a non-vertically integrated charging network. I define a metric, enabled e-miles, that captures whether a charging station mitigates consumers’ range anxiety. I then combine estimates from a spatial demand model and a simulation of charging behavior to compare charging demand and enabled e-miles at charging stations in California. I find strong evidence in favor of a spatial inefficiency in the location decisions of non-vertically integrated charging providers. These findings show that EV policy should not be location-neutral, but should consider spatial inefficiencies as well as network size.

In the second chapter of this dissertation, I examine whether the availability of home- andworkplace-charging infrastructure for renters has a detectable effect on EV adoption. Policies supporting charging infrastructure in workplaces and multi-unit dwellings (MUDs) are nascent. I aim to address two questions: (i) are such policies effective in increasing investment in charging stations, and (ii) whether they are effective in promoting EV adoption. I provide the first causal evidence on whether charging infrastructure represents a barrier to adoption for MUD occupants. I find that the subsidies increased the number of charging stations in a census block group (CBG) by .01. Against the sample average of 0.053, this represents a 20% increase. I did not detect any effect of charging stations on EV adoption. In the third chapter of this dissertation, I explore the installation of a bike sharing system in New York City and its interaction with existing vehicle infrastructure. Bike sharing is one form of shared mobility service. These increasingly ubiquitous services provide shared use of a vehicle, bicycle, scooter, or other mode of transportation. They can simultaneously complement and conflict with existing transportation infrastructure. New York City’s Citibike is the largest bike-sharing system in the country and among its goals is the relief of traffic congestion. I estimate the causal effect of Citibike on historical street speeds on Manhattan avenues. I employ Google Maps data to chart the routes between Citibike docks. I match these routes to rider counts and to novel data on traffic speeds at a 10-meter spatial resolution. This allows us to exploit variation in treatment intensity along and across avenues in fine resolution. I control for bike lanes, and other changes in street conditions over time. I find that the Citibike system decreased speeds on avenues in Manhattan. Overall, I estimate that the system decreased speed by 2.3% on average. At the maximum observed system utilization, travel time increased by 9.6%.

Together, these chapters employ travel data, spatial variation and tools, causal identification,and a combination of data and economic reasoning to draw policy relevant insights about how to decarbonize the transportation sector. A theme throughout is that the efficient deployment of infrastructure supporting low carbon transportation should take into account the location, scale, and utilization of existing infrastructure, be it the installed base of charging stations, or alternative transportation options.