Moral preferences of Investors

How to cite

thesmar, david; Bonnefon, Jean-François; Landier, augustin; Sastry, Parinitha. 2022. "Moral preferences of Investors." AEA RCT Registry. AEARCTR-0008764. https://doi.org/10.1257/rct.8764-1.0

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

david thesmar* , Jean-François Bonnefon , augustin Landier , Parinitha Sastry

* Primary investigator

Registry ID
AEARCTR-0008764
DOI
https://doi.org/10.1257/rct.8764-1.0
First registered
Last updated
Status
Completed
Trial dates
2019-03-20 – 2020-07-15
Countries
United States of America ()
JEL codes
G11, G41, G51, M14
Registry record
http://www.socialscienceregistry.org/trials/8764

Abstract

These experiments seek to characterize the moral preferences of stock investors. We auction stocks though the BDM algorithm. Each participant first participates in a quiz designed to ensure good understanding of consequences of their actions. Then, participants bid for three stocks.
- Stock N pays a given dividend
- Stock E pays a given dividend, and makes a donation to a charity
- Stock U pays a given dividend, and takes away money from a charity fund
We use actual money in this experiment, and randomize dividends and charity values
There are 8 conditions:
- Baseline: the participant selects the charity a priori
- Baseline with donation: the participant has the option to donate to the charity directly
- Baseline with random charity: the participant does not choose the charity, it is picked randomly from one list
- Pivotal: the charity receives the transfer only if the bid is successful
- uncertainty: charity donation is random
- ambiguity: the stock takes money from one charity, gives to one other

Intervention

These experiments seek to characterize the moral preferences of stock investors. We auction stocks though the BDM algorithm. Each participant first participates in a quiz designed to ensure good understanding of consequences of their actions. Then, participants bid for three stocks.
- Stock N pays a given dividend
- Stock E pays a given dividend, and makes a donation to a charity
- Stock U pays a given dividend, and takes away money from a charity fund
We use actual money in this experiment, and randomize dividends and charity values
There are 8 conditions:
- Baseline: the participant selects the charity a priori
- Baseline with donation: the participant has the option to donate to the charity directly
- Baseline with random charity: the participant does not choose the charity, it is picked randomly from one list
- Pivotal: the charity receives the transfer only if the bid is successful
- uncertainty: charity donation is random
- ambiguity: the stock takes money from one charity, gives to one other

Experimental design

Here is the baseline condition in detail, and the variants in later sections. In the baseline, we start the experiment by asking respondents to agree to a consent form that includes a one sentence description of the experiment, a ball park estimate of payments, and the experiment’s expected duration. In the first page of the interface per se, participants are given a short description of the game. They are told they will begin the experiment with a “virtual wallet” containing $2.00. Separately, $1.00 is placed in a fund that will be donated to a charity, which we refer to as the “charity wallet.” Participants are then told that they will make investment decisions that affect how much money is added or subtracted from both their wallets and the charity wallet. At the end of the experiment, participants receive a base payment of $2.00 and a bonus equal to the amount in their virtual wallet. The charity receives the final content of the charity wallet.
In the baseline condition, we make it clear to the participant that both successful and failed bids lead to the same changes in the charity wallet. Thus, purely impact-driven investors should not be willing to pay more than the selfish value (the expected cash-flows). Put differently, even if they are altruistic, consequentialist investors are not expected to bid at a premium for charity-giving stocks (or at a discount for charity-taking stocks), since their actions have no consequence.
At the end of this first page, participants are asked to select the charity that will receive the content of the charity wallet. Participants choose from the following six options: the American Civil Liberties Union, the World Wildlife Fund, Food for the Poor, the American Cancer Society, Save the Children, and the Environmental Defense Fund. The charities are well-respected nation- ally and span a range of environmental, social, and governance issues. We provide a screenshot of the first page in Appendix Figure A.1.
Participants then proceed to the practice quiz, a key step designed to ensure that participants fully understand how the bidding game works and the consequences of their choices on their wallet and the charity wallet. Participants are first shown a detailed example of the “neutral” firm that does not modify the charity wallet. They are forced to click line-by-line through the example to ensure slow digestion of information. They do not make any decisions and are not asked any questions during the example. Afterwards, participants are quizzed on both a hypothetical “ethical” and an “unethical” firm, which respectively add money to, and subtract money from, the charity wallet. Participants are given three opportunities to obtain a perfect score on the whole practice quiz. Only those who pass may continue to the main experiment. This is done to ensure that participants understand the consequences of their actions on flows of money.
The details of the practice quiz work as follows. The individual is shown, in random order, two hypothetical situations, which vary according to two dimensions. The first dimension is about the company’s prosocial actions: one is ethical and the other one unethical. Specifically, the ethical company gives $0.4 of its $1.5 profit to charity, and the unethical company earns $0.7 in profits and takes $0.4 from the charity. The second dimension is about the success or failure of the hypothetical bid. In the “succeed” scenario, the random bid is set at $0.50, while the hypothetical bid is at $1.1, so that the company share is actually purchased (for $0.50). In the “fail” scenario, the random bid is set at $2, above the hypothetical bid value of $1.1, so that the company share is not purchased. Participants see two scenarios drawn at random: The first firm is ethical with probability 1/2 (in which case the second firm is unethical), and purchased with probability 1/2 (in which case the second firm is not purchased). After presenting each hypothetical, we quiz participants on whether the firm’s share is purchased, how much they would hypothetically receive in dividends, and how much the charity would hypothetically receive/lose under the given parameters.

Once participants have fully mastered the quiz, they progress to the main experiment. In the main experiment, participants submit bids for a share in each of the three hypothetical companies: Neutral, Ethical and Unethical. Each company randomly draws a profit from {0.5, 0.6, 0.7, 0.8, 0.9, 1}. The neutral company gives the entirety of this profit to the shareholder. In contrast, the ethical company gives a random portion of this profit to the charity, drawn from {0.1,0.2,0.3,0.4,0.5}. The unethical company gives the shareholder the entirety of its profit along with money which it takes away from the charity wallet. The amount subtracted from the charity is also drawn from {0.1, 0.2, 0.3, 0.4, 0.5}. We randomly vary the order in which firm types (neutral, ethical and unethical) are presented in the main experiment (as we did in the quiz). At the end of the paper, we test if our results are affected by the order of presentation – and find that they are not.
After bidding on all three companies, participants are shown the amount in both their personal wallet and the charity wallet. Finally, we ask participants to answer a short survey designed to provide data on socio-demographics (education, age, gender, financial literacy).

Design details

Randomization method
Charity values and dividend amounts are randomized
Randomization unit
There are four waves (one in summer 2019, the second one in summer 2020)
Clusters
There are 8 conditions: - Baseline: the participant selects the charity a priori - Baseline with donation: the participant has the option to donate to the charity directly - Baseline with random charity: the participant does not choose the charity, it is picked randomly from one list - Pivotal: the charity receives the transfer only if the bid is successful - uncertainty: charity donation is random - ambiguity: the stock takes money from one charity, gives to one other In each treatment arm, participants do 3 (or 6) rounds of bidding each.
Observations
1,550 participants, recruited on Mturk over the summers of 2019 and 2020.
Arms
about 120 participants per treatment arm
Primary outcomes
we seek to characterize moral preferences of stock investors. This is to inform the debate on responsible finance, which has two views: value alignment (investor buy the stock they think act morally) and social impact (investors seek to change companies' footprints by investing and be involved in governance)
Secondary outcomes
We will seek to measure the structure of moral preferences: For instance, are preferences linear in corporate externalities, are they additive? Or are preferences more digital, in the sense that investors refuse to buy badly behaving firms at all, whatever the amount of harm that they cause.

Cited by

No citation data available for this registration yet.

Keywords