A large-scale RCT informing gasoline car owners on the private and social benefits of transitioning to electric vehicles

The experiments involve four information treatments including a control group, each group composed of 8000 clients of a leading financial institution, totaling 32,000 participants. The information to be shared with participants include the private benefits of EV ownership relative to gasoline vehicles (including charging infrastructure), the social benefits of EV ownership, and salient information about financial incentives of EV adoption that the financial institution already provides to the general public. Participants will be sent information at least once between October 2022 and January 2023 ["information campaign"], after which they will be surveyed on whether they purchased a vehicle in the past quarter ["endline survey"].

2022-10-05

2023-01-06

Which type of car was purchased including none in the last 3 months (even if undelivered)

[All data are anonymized]

Participants, all of whom are clients of the FI, not only finance their car purchases through the FI but also through competing financial institutions and other firms such as car dealers and OEMs. Alternatively, some buyers purchase a car without a loan. For this reason, at the end of the information campaign we will survey all participants on their car purchases rather than rely on the FI's loan book.

We know that the endline survey response rate will be significantly below 100%. Importantly, we aim to achieve a similar survey response rate across the four treatment groups. If there are disparities in the survey response rate between groups, to compare participants accounting for similar proportions across the groups we may consider the earlier responses in those groups for which we obtain higher response rates.

The FI has a team responsible for surveying its clients. Based on their advice and pending the response rate in the days after launching the endline survey, we may consider incentivizing clients to increase the survey participation rate (e.g., with a voucher reward or a "lucky draw").

In addition to the endline survey results, we will also examine car purchases in the FI's loan book across the four treatment groups.

We will observe anonymous visits to the four information websites. A different website is accessible to each treatment group, to complement the email communications. These websites were prepared for the purpose of the experiment and were not publicized elsewhere. We will also observe interest in Emissions Reduction Certificates (see below).

We will also observe the anonymous participants' spending on gasoline and on public transport (sum across credit and debit cards issued by the FI we are working with) in the months after the information campaign. This complements the analysis of demographic information by anonymous participant, which includes spending on gasoline and public transport during a baseline pre-period, when choosing the experimental sample (see below).

Participants in two of the treatment groups, should they transition to an EV, will be offered the opportunity to sign up to receiving an "Emissions Reduction Certificate (ERC) [which they] can share on social media [to] encourage others to make the switch. [The] ERC will include an updated carbon calculator figure."

The sample is composed of clients of the FI who have purchased car finance in the last decade (overwhelmingly for gasoline cars) and/or spent above a certain amount on gasoline in the past quarter. Participants will be randomized across the four information treatments.

The time-stamped templates for email communications by treatment group including text, tables, and infographics, co-developed between the academic and industry partners, and the underlying time-stamped calculators including the assumptions, will be published along with the findings.

Randomization done in office by a computer. Prior to launching the experimental campaign we will confirm that demographic variables of subjects (age, gender, education, income, residential area, past car financing, past purchases of gasoline) are balanced across the treatment groups.

Individual randomization

No

32,000 individuals

32,000 individuals

8000 individuals control, 8000 individuals for information on private benefits, 8000 individuals for information on private and social benefits, 8000 individuals for information on private and social benefits coupled with salient information on more attractive financing that the FI already provides to the general public.

Given a sample size of 32,000 (if all respond to the endline survey) and a base probability of buying a new car of 2.5% [1], we can detect an increase in probability of about 0.66 percentage points. This is about a 25% increase in the probability. [1] Based on an information campaign period of 3 months out of 120 months of Singapore's Certificate of Entitlement (a car license) eligibility.