Mental Accounting and Risk Preferences

In this project, we study how mental accounting affects elicitations of risk preferences. We will study how the framing of a payment, i.e. the type of mental account into which the payment should be entered, affects risk-taking behavior.

We will study how the framing of a payment to go into different mental accounts affects risk-taking behavior with that payment. We will have three treatments:
i) Fun Use:
Subjects will be told they should use the money they receive from the experiment for a fun purpose.

ii) Serious Use:
Subjects will be told they should use the money they receive from the experiment for a serious purpose.

iii) Neutral Use:
Subjects will not be told they should use the money they receive from the experiment for a particular purpose.

Upon successful and promising completion of the experiment using an online modality (Prolific), we intend to complement it with an in-person lab experiment. Details of that experiment will be added to this PAP in a future appendix.

2024-06-28

2024-10-31

We collect the following primary outcome:
- Elicited Risk Preference: This is defined as the risk preferences elicited using the method from Gneezy & Potters (1997).

References
Gneezy, Uri, and Jan Potters. "An experiment on risk taking and evaluation periods." The quarterly journal of economics 112, no. 2 (1997): 631-645.

Below we outline how we will use our primary outcome and the key hypothesis.

Hypothesis 1: subjects will express greater risk-seeking when money is allocated to a “fun” mental account compared to a “serious” mental account.

Hypothesis 2: subjects will express an intermediate amount of risk-seeking in the neutral treatment compared to the “fun” and “serious” treatments (i.e. the elicited risk preferences in the neutral treatment should fall between the “fun” and “serious” treatments).

Hypothesis 3: subjects in the neutral treatment whose intended uses are more “fun” will express greater risk-seeking than when their intended uses are more “serious”.

Hypothesis 4: subjects in the “fun” treatment will experience more positive emotions (such as excitement) and subjects in the “serious” treatment will experience more negative emotions (such as anxiety).

We will add further hypotheses in the appendix with regard to the comparison between the online and in-person findings.

- Intended purchase: subjects will be asked to write one or two sentences outlining their intended purchase with the money they receive in the experiment.

- Current emotional state: subjects will be asked to rate to what extent they feel certain emotions.

- Anticipated emotional state: subjects will be asked to rate to what extent they would feel certain emotions in a particular situation.

- Intended Purchase Self-Rating: subjects will be asked to rate their intended purchase on:
- how fun vs. useful their intended purchase is
- how much their intended purchase is a want vs. need
- how much their full purchase would cost
- how their purchase cost relates to a fixed benchmark
- how the experimental payment affected their intention to buy the intended purchase

- Domain Specific Risk Preference Scale: subjects will be asked to respond to the domain specific risk preference scale (Blais and Weber, 2006)

We will add further outcomes as appropriate for the in-person sessions in a future appendix.

References
Blais, A.-R., & Weber, E. U. (2006) “A Domain-Specific Risk-Taking (DOSPERT) scale for adult populations.” Judgment and Decision Making, 1, 33-47.

The elicitations above are intended to:
1. Ensure that the intended purchases are being manipulated by the treatment
2. To capture the mental account of intended purchases in the neutral treatment
3. To capture the emotional response subjects have in the different treatments to the risk elicitation

Our design aims to measure the impact of mental accounting on elicited risk preferences.

The design consists of three stages.

In stage 1, subjects are told they will have a 1-in-20 chance to earn a bonus (in addition to their show-up payment) up to $100 as a part of the experiment and that we would like them to report to use their intended use of that money. In the “fun” treatment, we ask them to spend the bonus on something fun. In the “serious” treatment, we ask them to spend the bonus on something useful. In the neutral treatment we do not ask them to spend their bonus in any particular way, though we still ask them to tell us their intended use of their bonus.

In stage 2, we introduce the risk elicitation using the Gneezy & Potters (1997) methodology. We tell them that their bonus will be based their decision about how many tokens of an initial 100 (at a value of $0.40/token) they would like to swap for a token that has a 50% chance of the value being multiplied by 2.5 and 50% chance of losing all value. In the “fun” treatment, we remind them that their intended purchase is something fun. In the “serious” treatment, we remind them that their intended purchase is something useful. In all three treatments we remind them of their intended purchase. We do not use words related to risk, lottery, or investment – instead we say that they have tokens that will be swapped for other tokens that have a certain probability of different outcomes occurring and provide examples. We ask them to complete understanding check questions about this part.

In stage 3 we ask them a series of additional questions. This stage occurs prior to informing them of the outcome of the risk elicitation. They are asked the following questions:

• Their current emotional state

• Their anticipated emotional state if they had swapped the entire 100 tokens in the risk elicitation

• Their self-rating of how their intended purchase rates on:
o Being fun
o Being useful
o Being something they want vs. need

• The total cost of their intended purchase and how it relates to the $100 maximum they could receive as a bonus

• How receiving the bonus affects their intention to buy their indented purchase

• The Domain Specific Risk Preference Scale (Blais and Weber, 2006)

• Demographic questions


We will add further detail specific to the in-person lab sessions in a future appendix.

References
Blais, A.-R., & Weber, E. U. (2006) “A Domain-Specific Risk-Taking (DOSPERT) scale for adult populations.” Judgment and Decision Making, 1, 33-47.
Gneezy, Uri, and Jan Potters. "An experiment on risk taking and evaluation periods." The quarterly journal of economics 112, no. 2 (1997): 631-645.

Randomization between the three treatments will be done by computer.

The randomization unit is the subject.

No

We plan to have 1200 in our online sessions. We will add in a future appendix the intended number of in-person subjects.

The number of observations will be the number of subjects.

We plan to have 400 subjects in each treatment arm for Prolific. We will add the number of expected subjects per treatment arm for the in-person sessions in a future appendix.

Comparing three papers – Gneezy and Potters (1997), Menkhoff and Sakha (2017, and Hackethal et al., (2023) – we find fairly similar patterns of behavior. However, Hackethal et al., (2023) most closely resembles the risk structure used in our study, and so we use the results of that paper to structure our power calculation. In Hackethal et al (2023), the most similar elicitation (GP Students Incentives [Tables A1 + A2]) has mean bet= 3.70 (out of 6), SD=1.51, where the risk is “to decide which fraction to invest in a project that pays either 2.5 times the invested amount or €0, with equal probability”. Using Stata power twomeans and using a control mean as 62 (=3.7/6*100), SD of 25 (=1.51/6*100), alpha=0.05, and power=0.8, we find that we will have enough power to capture a 5 unit difference in the number of swapped tokens. We will add further power calculations for the in-person sessions in a future appendix. References Hackethal, Andreas, Michael Kirchler, Christine Laudenbach, Michael Razen, and Annika Weber. "On the role of monetary incentives in risk preference elicitation experiments." Journal of Risk and Uncertainty 66, no. 2 (2023): 189-213. Gneezy, Uri, and Jan Potters. "An experiment on risk taking and evaluation periods." The quarterly journal of economics 112, no. 2 (1997): 631-645. Menkhoff, Lukas, and Sahra Sakha. "Estimating risky behavior with multiple-item risk measures." Journal of Economic Psychology 59 (2017): 59-86.