Electric Vehicle Charging at the Workplace: Experimental Evidence on Incentives and Environmental Nudges

How to cite

TEBBE, SEBASTIAN; Garg, Teevrat; Victor, David. 2024. "Electric Vehicle Charging at the Workplace: Experimental Evidence on Incentives and Environmental Nudges." AEA RCT Registry. AEARCTR-0013770. https://doi.org/10.1257/rct.13770-1.0

Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.

SEBASTIAN TEBBE* , Teevrat Garg , David Victor

* Primary investigator

Registry ID
AEARCTR-0013770
DOI
https://doi.org/10.1257/rct.13770-1.0
First registered
Last updated
Status
Completed
Trial dates
2023-06-06 – 2024-02-17
Countries
United States of America (La Jolla)
Registry record
http://www.socialscienceregistry.org/trials/13770

Abstract

To minimize the environmental costs of electric vehicles (EVs) and support decarbonizing electric grids, drivers must charge their EVs when renewable energy generation is abundant. To induce a shift in charging behavior toward daytime hours with ample solar energy, we conducted a field experiment (n=629) at a university campus to measure the influence of environmental nudges and financial incentives on the usage and timing of workplace charging. Environmental nudges led drivers to shift from early to later morning charging, whereas discounts to charge on campus increased total workplace charging and prompted a shift from daytime to early morning and overnight charging. We identify three clusters of mechanisms explaining these temporal shifts: the utilization and reliability of the charging network, concerns about charger scarcity, and driver demographics. Finally, we compute the societal effects of CO2 emissions and marginal electricity costs of these shifts in campus charging sessions.

Experimental design

The experiment consists of two interventions run in series: an informational treatment run over 19 days from October 5–23, followed by two phases of financial treatment run over 27 days from October 24 to November 19. Interventions were conducted within one academic quarter to maintain consistency in campus population and schedules.

In the informational intervention, half of the study participants were randomly assigned to treatment and half to control. Treatment consists of an email, delivered three times (once per week), stating the climate benefits of daytime charging compared to nighttime charging. In each email, benefits are reported as avoided CO2 emissions, equivalent unburned gasoline, and prevented global environmental damages.

In the financial intervention, drivers were given discounts for all Level-2 charging and randomly placed into treatment arms that varied discount size. The financial intervention consists of two phases. During the first phase (October 24 to November 5; 13 days), roughly one-third of participants receive a small discount ($.16/kWh), and two-thirds receive a large discount ($.23/kWh) - equivalent to 50% and 75% off the base campus rate of $.30/kWh, respectively. We set discounts so that the effective small-discount rate of $.14/kWh corresponds to the cheapest overnight home charging rate of the local electric utility, thus negating any economic advantage of overnight home charging. The large discount rate of $.07/kWh is equivalent to the mean LMP of wholesale electricity at UCSD, corresponding to the plausible lowest cost drivers would pay for charging.

During the second phase (November 6–19; 14 days), half of the large discount group continues with the large discount, while the other half moves to the small discount. To ensure parity in active campus days across both financial treatments, we schedule the second treatment to span 14 days, accounting for Veterans Day on November 10, when commuters are likely absent from campus. The second financial intervention thus has three treatment arms—LL (Large-Large), LS (Large-Small), and SS (Small-Small) discounts—given to three distinct groups. In this phase, we test for the presence of habit formation when financial discounts are reduced. If the charging behavior of participants on reduced discounts (LS) closely mirrors those who continue to receive the large discount (LL), our results are consistent with habit formation. In contrast, if the charging behavior of participants on reduced discounts (LS) reverts to those receiving the small-small sequence of discounts (SS), our results indicate the absence of habit formation during the first discount.

Four months after our series of informational and financial experiments, we ran a follow-up intervention over 13 days from February 5 to 17 to test for incentive-induced perceptions of scarcity of available chargers. In this intervention, we varied the discount notifications such that the messages to treated and control groups implied that different numbers of drivers would receive the discount. This follow-up experiment mimicked phase 1 of the financial experiment: the same methodology, same participants (i.e., new club enrollees were excluded), stratified block randomization into treatment arms that receive small or large discounts, notifications, and odometer surveys. In total, the experiment consisted of four treatment arms. Two arms received the large discount; two received the small. New to this experiment was that, within each discount regime, half of participants received a discount notification email that indicated that all drivers would receive the discount simultaneously, while the other half received a discount notification message that indicated that no more than 33% of drivers would receive the discount

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
Individual-randomization
Clusters
629 electric vehicle drivers
Observations
629 EV drivers.
Arms
Environmental prompt: 315 EV drivers; No environmental prompt: 314 EV drivers Large discount: 418 EV drivers; Small discount: 211 EV drivers Large-Large: 210 EV drivers; Large-small: 208 EV drivers; Small-Small: 211 EV drivers
Primary outcomes
We want to analyze the effect of our intervention on six outcomes of total campus charging activity and five periods of campus charging. The six total charging outcome variables include each driver’s share of charging done on campus, the number of sessions initiated, energy consumed, session duration, charging duration, and idle duration. A driver’s share of charging on campus is the total energy consumed from campus charging divided by the expected energy consumed from total driving, which we estimate from data on the driver’s daily vehicle miles driven, obtained through recurring odometer readings, and their vehicle’s energy efficiency. To measure the effect of interventions on the timing of charging (measured by the hour in which sessions are initiated), we analyze charging over five distinct periods: early morning (5:00–6:59), which sees the earliest morning commuters and has low utilization; morning (7:00–9:59), characterized by the arrival of most regular commuters and a rapid surge to near maximal levels of network utilization, along with rising solar production; midday (10:00–15:59), characterized by relatively constant high utilization and maximal solar generation; evening (16:00–20:59), characterized by departing commuters, arrival of nighttime workers, and rapidly waning solar generation; and overnight (21:00–4:59), characterized by low network utilization.

Cited by

No citation data available for this registration yet.

Keywords