Personal Carbon Account for Promoting Public Transit

How to cite

Zhang, Da; Salvo, Alberto; Yu, Runxin; Zhou, Yanyu. 2025. "Personal Carbon Account for Promoting Public Transit." AEA RCT Registry. AEARCTR-0016387. https://doi.org/10.1257/rct.16387-1.0

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

Da Zhang* , Alberto Salvo , Runxin Yu , Yanyu Zhou

* Primary investigator

Registry ID
AEARCTR-0016387
DOI
https://doi.org/10.1257/rct.16387-1.0
First registered
Last updated
Status
In development
Trial dates
2025-07-17 – 2026-09-30
Countries
Private
JEL codes
Q58 Q51 Q54 C93
Registry record
http://www.socialscienceregistry.org/trials/16387

Abstract

This project evaluates the efficacy of a Personal Carbon Account (PCA) system as a behavioral intervention to promote low-carbon public transportation in Beijing, China. In collaboration with the Beijing Municipal Transportation Commission and a public transport enterprise, we implement a randomized controlled trial (RCT), engaging over one million residents. Participants are randomly assigned to a carbon credit group (earning carbon credits based on public transport usage, which are redeemable for goods), a direct subsidy group (receiving cash-equivalent fare reductions for public transit), or a control group. The intervention design for the carbon credit group aims to: (1) assess whether introducing a carbon credit redemption mechanism drives sustained changes in low-carbon travel behavior over time, and explore how varying credit redemption rates—reflecting different financial incentive intensities—impact the magnitude of behavioral shifts. (2) examine the influence of distinct invitation framings—highlighting economic benefits versus environmental advantages—on participants’ motivation to engage in the campaign (3) investigate how different types of information nudges, including periodic feedback on non-price factors, price-related information, and community participation, shape green travel behavior (4) compare the behavioral effects of short-term versus long-term exposure to the intervention, testing for immediate compliance versus delayed habit formation, as well as potential declines in nudge efficacy. This study further compares the carbon credit mechanism against direct subsidies to quantify their relative efficacy in driving behavioral change and long-term sustainability. To assess shifts in environmental attitudes and awareness, two rounds of structured surveys will be administered through our partnering enterprises. By combining administrative travel data with behavioral and attitudinal measures, the experiment aims to provide empirical evidence on the design and scalability of PCA schemes and offer policy insights for climate and public transport strategy in China and other urban settings.

Experimental design

This study designs a large-scale randomized controlled trial (RCT) with over 1 million digitally active public transit users. The study randomly assigns eligible participants to one of three groups: a carbon credit group (receiving indirect monetary incentives via carbon credits), a direct subsidy group (receiving cash-equivalent transit discounts), and a control group.

1. Carbon credit group (main treatment group)
Users in this group receive indirect monetary incentives in the form of carbon credits, earned by using low-carbon transportation (bus/subway). Participants can redeem credits for rewards. This group varies along three dimensions:
(1) Carbon credit redemption rate: Users are randomly assigned to one of three subgroups offering different effective carbon prices (¥100/ton, ¥200/ton, ¥500/ton). Redemption rates are revealed only after user registration.
(2) Invitation message framing: Prior to registration, users receive SMS/in-app invitations randomly framed to highlight either financial benefits or environmental/social impact.
(3) Information intervention: After registering, users receive periodic reports with either (a) non-price information, (b) non-price information and price information, or (c) non-price information and social comparison (ranking).

2. Direct subsidy group
This group receives direct fare subsidies in the form of 150 instant-use vouchers equivalent in value to the ¥100/ton carbon group. Vouchers are applied automatically to transit rides upon campaign registration and are valid for three months. This group enables a comparison between cash-equivalent vs. credit-based incentives.

3. Control group
Users in the control group receive no intervention during the main experimental window.

4. Treatment duration variation
To study persistence of behavioral change, a subset of carbon credit users is randomly assigned to a 9-month treatment, while the rest receive the standard 3-month treatment. This allows comparison between short- and long-term effects.

5. Surveys
Two waves of surveys are conducted to evaluate the behavioral and attitudinal impact of the intervention. The first survey, available for the first two months to participants who confirm their participation, collects baseline data including demographics, household characteristics, environmental awareness, and willingness to adopt low-carbon travel. The second survey, distributed in the third month to intervention participants, tracks life changes and reassesses awareness and willingness related to carbon inclusion, along with demographics.

Design details

Randomization method
Randomization done in office by a computer.
Randomization unit
The unit of randomization is the individual user. All users are randomly assigned at the individual level to one of the experimental arms: direct subsidy group, carbon credit group, or control group. Within the carbon credit group, users are further randomly assigned (also at the individual level) to different: encouragement invitation types (price incentive vs. non-price incentive), information interventions (non-price message, price message, or social comparison message), carbon price levels (¥100/ton, ¥200/ton, ¥500/ton), and continued vs. discontinued treatment arms (for extended follow-up analysis). There is no cluster-level randomization in the design; all assignments are conducted individually within stratified blocks.
Clusters
N/A
Observations
The study will collect weekly panel data for approximately 1.03 million users over a 51-week period, resulting in roughly 52.5 million total user-week observations.
Arms
The experiment includes approximately 500,000 users in the carbon credit group, 30,000 users in the direct subsidy group, and 500,000 users as the control group, yielding a total sample size of around 1.03 million users.
Primary outcomes
User-level weekly travel frequency, measured via detailed records of individual bus and subway usage; Participation rate among invited individuals, used to evaluate the heterogeneous effects of the encouragement invitation.

Cited by

No citation data available for this registration yet.

Keywords