Abstract:Transportation electrification is viewed by many as a cornerstone for climate change mitigation, with the ultimate vision to phase out conventional vehicles entirely. In a world with only electric vehicles (EVs), transportation pollution would be primarily determined by the composition of the electricity grid. For the foreseeable future, however, environmental benefits of EVs must be measured relative to the (likely gasoline) car that would have been bought instead. This so-called counterfactual vehicle cannot be observed, but its fuel economy can be estimated. A quasi-experiment in California allows us to show that subsidized buyers of EVs would have, on average, purchased relatively fuel-efficient cars had they not gone electric. The actual incremental pollution abatement arising from EVs today is thus substantially smaller than one would predict using the fleet average as the counterfactual vehicle. We discuss implications for climate policy and how to accurately reflect EV choice in integrated assessment models.