Back to History

Fields Changed

Registration

Field Before After
Trial Title The Onset of Offsets: The Limits of Social Signaling in Eco-Friendliness The Onset of Offsets: Social Signaling and Eco-Friendliness
Abstract Citizens in the USA spend a large amount of time, effort, and money engaging in voluntary actions to improve environmental quality. The motivation behind these actions is not clearly understood. They could be driven by private desire for improved environmental quality or by a social desire to signal eco-friendliness. To determine if these actions are driven by private or social considerations, we run an experiment that manipulates the publicity of carbon offset purchases, posting some consumers' names online and giving them share-able certificates that confirm their purchase. On average, pilot data shows that the public treatment does very little to increase uptake. However, the public treatment more than doubles uptake (14% to 31%) among those that have small second-order beliefs -- those that believe that everyone else thinks that carbon offsets are rare. This points to consumers receiving social utility from being first-movers on an eco-friendly action, though this social utility erodes quickly as consumers grow to believe that others will not see them as first-movers. This finding confirms the existence of an "honor region" as described in (Benabou and Tirole, 2011). We use this finding to discuss broader implications for tax and subsidy policy regarding policy on climate and other environmental externalities. Despite being a top concern for global welfare, policymakers have struggled to implement taxes to address climate change. This has led to an increasing reliance on voluntary actions taken by firms and consumers to combat carbon emissions. In this paper, I experimentally test a non-standard policy that publicizes voluntary consumer carbon mitigation, leveraging social rewards to increase uptake. Specifically, I show that posting names of carbon offset purchasers online is an effective tool to encourage voluntary carbon mitigation, as confirmed by experimentally estimated demand curves. Further, I show that social rewards vary heavily by perceived market penetration of offsets. Uptake increases vastly among those with the lowest perceptions of carbon offset market penetration but is only slightly impacted among those with moderate perceptions. I then estimate a structural model of demand for prosocial actions in the face of social rewards to understand the implications of my findings on optimal tax policy. To avoid crowding out social incentives, I show that optimal subsidies for consumer carbon mitigation technologies should start out small at low participation rates and ramp up as these technologies become more common.
Trial Start Date June 10, 2024 November 04, 2024
Trial End Date June 13, 2024 November 08, 2024
JEL Code(s) D62, D64, D91, Q54
Last Published December 06, 2023 08:25 AM September 17, 2024 10:36 AM
Intervention (Public) There are two elements of randomization. The first is whether or not decisions to buy the carbon offsets are public. In the public group, those who buy carbon offsets will be allowed to provide their name to be posted online on a website of our design. This website is promoted on an Instagram page with thousands of followers, as participants are informed. The group who is not public will not have any of these options, they will just decide whether to purchase or not under status quo conditions. The additional layer of randomization involves altering consumer beliefs about the market penetration of carbon offsets. To do this, each participant is sent a signal which varies perceptions about market penetrations before any decisions about buying offsets are made. This creates exogenous variation in beliefs which we leverage to determine how social rewards vary by these beliefs of market penetration.
Intervention Start Date June 11, 2024 November 04, 2024
Intervention End Date June 12, 2024 November 08, 2024
Experimental Design (Public) This is a survey experiment. Participants are recruited online and provide us with demographic information. We then inform them about carbon offsets, in case they are unfamiliar. From here, first- and second-order beliefs on market share of carbon offsets are elicited in an incentivized manner. We provide participants with high and low signals that shift second-order beliefs and serve as an instrument. Finally, we give participants the choice between receiving a bonus payment or buying an offset, where some participants are told their action will yield their name being posted online while others are not. Some participants are told that name-posting is mandatory and others are told it is voluntary, which tests for privacy concerns. This is a survey experiment. Participants are recruited online and provide us with demographic information. We then inform them about carbon offsets, in case they are unfamiliar. From here, first- and second-order beliefs on market share of carbon offsets are elicited in an incentivized manner. We provide participants with random signals that shift second-order beliefs and serve as an instrument. Finally, we give participants the choice between receiving a bonus payment or buying an offset, where some participants are told their action will yield their name being posted online while others are not. From here, we ask several questions to determine which mechanisms drive the impact of name-posting as well as to determine endline beliefs about carbon offsets and the survey more specifically.
Planned Number of Clusters 4200 individuals 3,500 individuals
Planned Number of Observations 4200 individuals 3,500 individuals
Sample size (or number of clusters) by treatment arms Treatments are public required/public voluntary/private, price variation ($1.50, $1.10, $0.75), offset tonnage variation (150 lbs., 190lbs., 225lbs.), and signal for second-order beliefs (low, medium, high). This yields 81 total treatment groups. The 4200 individuals will be roughly evenly divided across these 81 groups (~52 each) Treatments are public/private, price variation ($1.40, $0.70, $0.10), and signal for second order beliefs (14, 56, 128, 217, 315, or 438). This yields 36 total treatment groups. The 3,500 individuals will be roughly evenly divided across these 36 groups (~97 each). In practice though, the treatment on second order beliefs is to yield an instrumental variable rather than to make discrete across-group comparisons. Neglecting that variation, there are then 6 treatment groups, for 583 people per group.
Additional Keyword(s) Social Signaling, Eco-friendliness, Prosociality
Intervention (Hidden) There are two elements of randomization. The first is whether or not decisions to buy the carbon offsets are public. In the public group, those who buy carbon offsets will be allowed to provide their name to be posted online on a website of our design. The website is available at https://www.onset-of-offsets.com/. This website is promoted on an Instagram page with thousands of followers, as participants are informed. The Instagram page is available at https://www.instagram.com/onsetofoffsets/. The group who is not public will not have any of these options, they will just decide whether to purchase or not under status quo conditions. The additional layer of randomization involves altering consumer beliefs about the market penetration of carbon offsets. To do this, each participant is sent a signal which varies perceptions about market penetrations before any decisions about buying offsets are made. This creates exogenous variation in beliefs which we leverage to determine how social rewards vary by these beliefs of market penetration. Belief elicitation and alteration require a bit more discussion. To elicit beliefs, we first ask participants to guess how many people taking the survey self-reported having purchased a carbon offset before. This is incentivized for correctness. Since this measures how common each person thinks carbon offsets are, we refer to them henceforth as "first order beliefs." After this, we then ask people to guess what the average first order belief was, again incentivized for correctness. As these measure beliefs about beliefs, we refer to them as "second order beliefs." Second order beliefs are the beliefs which may impact social rewards, and are thus the beliefs which are of most pertinence to our study. We experimentally manipulate these beliefs by sending random signals to consumers and asking them what their posterior second order belief is after seeing the signal. These signals read "We previously surveyed 1,000 United States adults. In this previous survey, we found an average guess of [PROP] out of 1,000 people buying a carbon offset." Here, "[PROP]" randomly varies to be 14, 56, 128, 217, 315, or 438. These values come from averaging guesses in previous pilot studies we ran.
Back to top

Partners

Field Before After
Partner Name Climate Vault
Partner Type ngo
Partner Website (URL) https://climatevault.org/
Public Yes
Back to top