The Electric Vehicle Rebound Effect

Electric vehicles (EVs) are often considered a promising technology for the decarbonisation of transportations, especially now that battery storage technologies are becoming more energy-dense and affordable and policy efforts are in place to make electricity generation cleaner. This has triggered intense policy action around the planned phase-out of internal combustion engine (ICE) vehicles and large volumes of investments by automakers as they convert their production. Yet, the degree to which EVs and ICEs are substitutable in the eyes of consumers is an empirical question. EVs are different from ICEs on a number of dimensions, most notably they generally cost much less to drive but more upfront, and charging stations are not yet as pervasive as gas stations so that users often complain about range anxiety. Other differences include noise level and vehicle weight. Progress in the range of newer EVs and greater presence of charging stations in the last few years should allow EVs to be more substitutable for ICE alternatives, but the authors still observe that most households with EVs also have at least one ICE vehicle in their portfolio. As EVs are substantially cheaper to drive per mile than ICEs, a rebound effect may occur after a household purchase a EV if the savings are then reallocated to increase total VMT and the miles traveled using ICEs. Yet, applying estimates of the rebound effect obtained from studies that look at purchase of more fuel efficient vehicle would be inappropriate as the difference in costs per miles is substantially more pronounced. Moreover, policies that make charging cheaper (such as TOU rates, managed charging) may again increase total driving (a second type rebound effect), potentially reducing the promised environmental benefits of EVs. At the same time, policies that increase gasoline prices (such as carbon/gasoline tax) should shift VMT away from ICE vehicles and towards EVs. Understanding the substitutability between these vehicles is therefore crucial for the development of a pricing policy that internalize externalities of gasoline consumption. The aim of the paper presented in this ZEW Research Seminar is to look at how households adjust their total VMT and the allocation of VMT between vehicles in response to the purchase of an EV (the authors focus on battery electric vehicles, or BEVs), as well as in response to changes in gasoline prices and electricity rates. These results will also help estimate to what extent EVs can offset the use of existing ICE vehicles. To answer these questions, the authors use detailed information on the composition of households’ vehicle portfolio in Massachusetts (from registration data) and the miles driven by each vehicle over time (from inspection data) to perform an event-study and difference-in-difference analysis. Purchasing a new vehicle, no matter the fuel type, may be a response to an increased demand for driving and may also result in an increase in driving linked to the novelty effect. The authors therefore parse this effect by comparing driving patterns before and after the purchase of an ICE and a BEV vehicle. The authors further distinguish between households who purchase a vehicle as an addition to their vehicle portfolio and households who purchase a vehicle to replace one. Coefficients for the change in VMT after purchase are estimated using a difference-in-difference event study, combined with propensity score matching for the likelihood of purchasing an EV. To estimate the rebound effect linked to changes in electricity prices, the authors use a regression discontinuity design (RDD) around the borders of utility catchment areas, to capture the variation in electricity rates and in the time of their adjustments between utilities. The authors further estimate the elasticity of VMT to changes in gasoline prices for households with and without BEVs, using a fixed effect model and instrumenting for the price of gasoline using the price of crude oil in the international markets as instruments.