Use Efficiency and Pricing in Public EV Charging Infrastructure
Abstract: The electrification of road transport depends on the provision of charging infrastructure. Whereas the number of installed public charging stations grows, how intensively they are used remains poorly understood. Using transaction-level data from a major Swiss provider of charging stations from 2019 to early 2022, I show that vehicles routinely remain connected long after charging has ended. Idle occupancy accounts for 52% of connection time at AC chargers and 33\% at DC chargers. However, the large physical inefficiency does not translate into an externality cost of the same magnitude. Chargers are rarely scarce with only 1.5% of station-time blocks being fully occupied, such that idle time that hinders an arriving driver is worth about a tenth of a simplified accounting estimate, or around CHF 15000 per year scaled to the national charger stock. The prevailing tariff structure is very heterogeneous. Some operators charge only per kWh fees, whereas others levy per-minute fees once a session passes sixty minutes. These per-minute threshold-based fees are a manifold of the externality cost of the blocked capacity, and charging at AC stations responds to them. A bunching estimator exploiting the sixty-minute kink yields duration elasticities of 0.17--0.79. In contrast, no similar bunching response is found at DC chargers since sessions are usually shorter than one hour. The idle occupancy cost is small in my data, but the sample period spans an early stage of EV adoption in Switzerland; as adoption and utilisation rise, idle occupancy is likely to become more costly.Congestion Management for Electric Vehicle Charging Stations (with Jing Li)
Abstract: Congestion at electric vehicle (EV) charging stations has been a concern for drivers since the early days of the EV industry, and the concerns and issues grow as new EV sales continue to expand rapidly in many countries around the world. Using data from the charging industries in Germany and Switzerland from 2022 and 2023, we document two policy-relevant patterns in the data: (i) Congestion occurs throughout the EV charging networks in Germany and Switzerland. (ii) The distribution of charging session duration has a long right tail (right-skewed). (iii) Demand spills over from a congested charging station to neighboring ones. To evaluate the effectiveness of three congestion-management policies, we specify a model where drivers choose trip departure times, where to charge, and how long to charge: expanding the number of connectors, upgrading the electrical power, improving service quality (i.e reducing outages), and congestion pricing.