We employ an online randomized lab experiment to estimate the impact of a company's bankruptcy status on consumer demand for that company's products and services. We identify each participant's willingness to pay for a product of a firm. We measure how that willingness to pay changes when we exogenously vary information about the firm's bankruptcy status. We also exogenously vary information about the purpose of the survey to study experimenter demand.
Intervention
Our interventions involve randomly providing participants with different news articles about a firm, some of which include information about the firm's bankruptcy status. We also randomize information provided about the purpose of the survey: some participants are told the survey is about bankruptcy, others are told that it is about studying consumer purchase behavior (both are true).
Experimental design
We identify each participant's willingness to pay for a product of a bankrupt firm. We measure how that willingness to pay changes when we exogenously introduce information about the bankrupt firm. Our analysis will look similar to our earlier experiment related to the Hertz bankruptcy. A key novel feature is that we also randomly inform participants that the purpose of the survey is to study reactions to bankruptcy. We study how our baseline treatment effect (how randomly learning about bankruptcy impacts willingness to pay) varies based on how the participant perceives the purpose of the survey.
Design details
Randomization method
We implement our experiment using a Qualtrics survey. Qualtrics offers a computerized randomization procedure, which we use.
Randomization unit
We randomize at the participant level.
Clusters
We plan on having 1600 participants after excluding participants that (i) fail our two initial comprehension checks or (ii) have nonmonotonic price-list responses (i.e., preferring $50 over the good at firm B and preferring the good at firm B over $55). We run our survey in batches so we will not end with precisely 1600.
Observations
We plan on having 1600 participants after excluding participants that (i) fail our two initial comprehension checks or (ii) have nonmonotonic price-list responses (i.e., preferring $50 over the good at firm B and preferring the good at firm B over $55). We run our survey in batches so we will not end with precisely 1600.
Arms
We randomize participants into two "purpose" groups. One group is told the survey is about bankruptcy. The other is told the survey is about consumer purchase behavior. We then independently randomize participants into two information groups. One group learns the firm is bankrupt, the other group does not. We have approximately 25% of each bin: two purpose bins by two information bins. Since we expect roughly 1600 participants, we expect 400 participants per bin.
Primary outcomes
Our key outcome variable is the willingness to pay for a product of a firm that is in Chapter 11 bankruptcy. We measure how this depends on an endogenous variable: awareness that the firm is in Chapter 11 bankruptcy. Our experimental design introduces exogenous variation in this awareness. We also introduce exogenous variation in the extent to which the participant believes this is a survey about bankruptcy. We measure how the relationship between our awareness treatment and willingness to pay varies based on how the participant views the purpose of the survey.
Secondary outcomes
We also measure how many participants were already aware that the firm was in bankruptcy by asking participants to note which firms in a list are currently in bankruptcy. We also measure what the participant believes the purpose of the survey is.
Cited by
No citation data available for this registration yet.