Consuming Values: Estimating Consumer Demand for Corporate Social Responsibility

Download data and study materials from OSF

Principal investigators:

Jacob Conway

University of Chicago Booth School of Business

Email: Jacob.Conway@chicagobooth.edu

Homepage: https://jacob-conway.com/

Levi Boxell

Independent researcher

Homepage: https://sites.google.com/site/leviboxell/


Sample size: 2586

Field period: 07/01/2021-10/17/2022

Abstract
Our project investigates the extent to which individuals' consumption decisions are influenced by their value fit with firms on three corporate social responsibility dimensions (political, environmental, diversity). Through a randomized information provision experiment, we shift respondents' beliefs about the corporate social responsibility (CSR) behaviors associated with a given firm. We then measure the effect of these belief changes on consumption preferences (in $) using incentivized choices in a gift card lottery. We explore heterogeneity in preferences across consumers and CSR dimensions, and test for potential mechanisms using cross-randomized strategic incentive and social image concern treatments. We find that value fit impacts (unincentivized) favorability towards the firm for all three CSR dimensions. However, we find strong evidence of willingness-to-pay for value fit with firms only on environmental issues.
Hypotheses

We have in mind three primary possible hypotheses on the link between value fit and gift card preferences:
1. Aligned Buycotts: If a respondent learns that a firm’s values are closer to her own values than she thought previously, she will want to consume at that firm more and so will increase her valuation of that firm’s gift card.
2. Opposed Boycotts: If a respondent learns that a firm’s values are further from her own values than she thought previously, she will want to consume at that firm less and so will decrease her valuation of that firm’s gift card.
3. Cheap Talk: If a respondent learns that a firm’s values are closer or further from her own values than she thought previously, this will not change her desire to consume at that firm nor her valuation of that firm’s gift card. This would suggest that when individuals express a preference for value fit in unincentivized surveys, these responses may be “cheap talk” in that they are not matched by actual consumption responses.

In our instrumental variables estimation, a negative coefficient would be consistent with either an aligned buycott or an opposed boycott response. In contrast, a coefficient of zero would be consistent with the cheap talk hypothesis. We can also split apart and separately test for asymmetry (i.e., aligned buycotts vs. opposed boycotts).

Experimental Manipulations

We test in our survey experiment how the randomized provision of true information regarding a firm's CSR activities influences a respondent’s perception of her value fit with the firm as well as her valuation of spending at that firm. Our experiment analyzes three dimensions of CSR activity: environmental activity is quantified by the firm’s percentile rank in an environmental impact rating relative to other firms in its industry group (source: Refinitiv); gender diversity by the share of women among a firm’s corporate board (source: 50/50 Women on Boards); and political activity is quantified by the share of a firm’s political donations (made either by the firm’s employees or by its political action committee) which went to Republicans in the 2020 election cycle (source: OpenSecrets). Some individuals are also provided with placebo information regarding a firm’s equity ratio as a percentage of firm assets (source: Fortune).

Randomly selected individuals receive the true values for all firms J_i along dimension k (some receive environmental information for all firms, some receive placebo information for all firms, etc.). We also cross-randomize two secondary treatments designed to test mechanisms: (a) a strategic incentive information treatment in which randomly selection individuals receive information on the share of individuals who have reported boycotting any firm before, by party, and (b) a social image concern treatment in which respondents are incentivized to share one of their choices on social media.

A working copy of the survey instrument can be completed here: https://stanforduniversity.qualtrics.com/jfe/form/SV_3efjNTHEOlItQY6

Outcomes
- Willingess-to-Pay: Our primary outcome is the change (posterior - prior) in an individual’s willingness-to-pay for a firm’s $50 gift card. We elicit willingness-to-pay using the Becker– DeGroot–Marschak method (BDM), and we do so both before and after the experimental information provision in our experiment.
- Favorability: Our secondary outcome is the change in an individual’s favorability towards a firm. Favorability is measured on a 0-100 feeling thermometer, as in the American National Election Studies.
- Perceived Value Fit: An intermediate/first-stage outcome in our instrumental variables design is respondent i's perceived value fit with a firm f on dimension j. This defined as |d_{ifj} - d^{*}_{ij}| where d_{ifj} denotes the individual's perceived CSR value of the firm, and d^{*}_{ij} denotes i's ideal point.
Summary of Results

When analyzing the impact of value fit on favorability toward the firm, we find impacts that are strongly statistically significant for environmental fit (IV coefficient=−0.16; pval=0.019), similar in magnitude but statistically insignificant for gender diversity fit (IV coefficient=−0.13; pval=0.241), and similar in magnitude but marginally statistically significant for political fit (IV coefficient=−0.14; pval=0.052).

When analyzing impacts on willingness-to-pay, we find that only environmental value fit has a significant impact (IV coefficient=−0.24; pval=0.011). This point estimate suggests that a typical respondent who positively updates her beliefs about a firm's environmental impact from the 25th to the 50th percentile would then be willing to pay about $6 more ((-25)*(−0.24)=6) for a $50 gift card at that firm. Our estimates are much smaller in magnitude and statistically insignificant for the impact of gender diversity fit (IV coefficient=−0.01; pval=0.894) and political fit (IV coefficient=0.05; pval=0.532) on WTP.

Across demographic subgroups, our estimates of willingness-to-pay for environmental value fit are stronger among men and are increasing in respondent’s income. They appear to be somewhat weaker for individuals aged 30-44 relative to other age groups.

Analyzing heterogeneity in consumer demand for value fit across our cross-randomized treatments, we do not find significant differences in our estimates between respondents who did or did not receive an incentive to share their choices on Facebook, or among respondents who learned that they had over- vs. under-estimated the frequency with which others boycott. This might indicate that mechanisms other than strategic considerations and social image concerns drive demand for value fit, or might be due to a weak treatment.