Ride-Hailing Fleets Increase Vehicle Electrification and Reduce Emissions when Air Emissions Externalities are Internalized 

Abstract
In the US, the transportation sector is the largest source of greenhouse gas emissions and passenger cars make up its largest share. [1,2] Ride-hailing services from transportation network companies, such as Uber and Lyft, serve the fastest growing share of U.S. passenger travel demand. Depending on a given region's electricity generation mix, electrifying ride-hailing fleets may be a relatively low-cost option to reduce emissions, but few regions have incentive structures to induce such a shift and policy changes may be warranted. Nonetheless, the high use-intensity of ride-hailing vehicles may be economically attractive for electric vehicles, which typically have lower operating costs and higher capital costs than conventional vehicles. We optimize fleet technology composition (mix of conventional vehicles (CVs), hybrid electric vehicles (HEVs), and battery electric vehicles (BEVs)), vehicle routing, and battery charge scheduling to satisfy exogeneous trip demand at minimum cost. Private costs minimized include fuel and electricity costs, vehicle maintenance, and vehicle purchase costs minus the depreciated resale cash flow at each vehicle's end of fleet usage. We compare results across Austin, TX, Los Angeles, CA, and New York, NY, each with region-specific energy prices, air emissions from marginal electricity generation, and air emissions externalities. In all cases, the optimal fleet includes a mix of technologies. In present and future scenarios for each city, HEVs and BEVs make up the largest portion of vehicle distance traveled in optimized fleets, and CVs are used primarily for periods of peak demand (if at all). Separately, we consider how technology composition and fleet operations change when life cycle air pollution and greenhouse gas emission externality costs (from internal combustion, fuel refining, electric grid marginal emissions, and vehicle production and disposal) are internalized via a Pigovian tax. Across a wide range of scenarios for the three cities, a Pigovian tax leads to increased fleet electrification, a shift in charging toward periods when the grid is cleaner, and a reduction in emissions externalities. In the base case, externality reductions range from 10-22% across the three analysis cities. This equates to as much as $29M annually per city (in Los Angeles, the analysis city with the largest reductions). We also consider an array of alternative assumptions for model inputs including carbon price, air emissions externality valuation model, BEV battery capacity and price, private fleet discount rate, and labor costs. In all cases, the optimal fleet mix, its dispatch strategy, and resulting air emissions change substantively when air emissions externalities are internalized, suggesting a role for policy.