Distributional Approach to Risk Preferences

How to cite

Chemaya, Nir; Jabarian, Brain. 2026. "Distributional Approach to Risk Preferences." AEA RCT Registry. AEARCTR-0017728. https://doi.org/10.1257/rct.17728-1.0

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

Nir Chemaya* , Brain Jabarian

* Primary investigator

Registry ID
AEARCTR-0017728
DOI
https://doi.org/10.1257/rct.17728-1.0
First registered
Last updated
Status
Completed
Trial dates
2026-01-26 – 2026-01-27
Countries
Israel (), United States of America ()
Registry record
http://www.socialscienceregistry.org/trials/17728

Abstract

In this experiment, we asked subjects to make decisions in 20 risk-elicitation tasks, in which they chose between lotteries following the experimental design of Holzmeister and Stefan (2021). The experiment was incentives and subjects were paid based on their decisions. The experimental design is structured as follows: the 20 decisions are organized into five blocks. In each block, subjects make decisions on four different risk-elicitation tasks. The same four tasks are repeated across all five blocks. As a result, each subject makes 20 decisions in total, corresponding to four distinct tasks, with each task repeated five times (once in each block). At the end of the experiment, one of the 20 tasks is randomly selected for payment.

Our research question is whether subjects’ preferences become more consistent over time. We examine this using our Distributional Approach to estimate risk preferences and to evaluate the predictive power of our method.

Experimental design

In this experiment, we asked subjects to make decisions in 20 risk-elicitation tasks, in which they chose between lotteries following the experimental design of Holzmeister and Stefan (2021). The experiment was incentive-compatible, and subjects were paid based on their decisions. The experimental design is structured as follows: the 20 decisions are organized into five blocks. In each block, subjects make decisions on four different risk-elicitation tasks. The same four tasks are repeated across all five blocks. As a result, each subject makes 20 decisions in total, corresponding to four distinct tasks, with each task repeated five times (once in each block). At the end of the experiment, one of the 20 tasks is randomly selected for payment.

Design details

Randomization method
In our experiment, there is no treatment versus control group, since all subjects face the same risk elicitation tasks and the same experimental design. We use a randomization process in which each subject is assigned a different order of the risk elicitation tasks to control for order effects and potential history effects on their decisions. In addition, because payments are based on the outcomes of the lottery decisions participants make, the computer uses the corresponding lottery probability distributions to generate outcomes randomly and determine their payoffs. However, in our analysis we use the results of Holzmeister and Stefan (2021) as a benchmark. In their study, subjects completed each of the four risk-elicitation tasks only once (one decision per task). We use these results as a control to show that subjects in our experiment behave similarly in the first round—when they make decisions in each of the four tasks—relative to later rounds, in which they gain experience and potentially learn. Since we employ the same risk-elicitation tasks as Holzmeister and Stefan (2021), with comparable stakes (payoffs), and both experiments are conducted with similar student populations, this comparison is well suited for validation. If we nevertheless observe differences in the outcomes of the first four trials between our experiment and Holzmeister and Stefan (2021), we will run an additional control experiment in our laboratory that exactly replicates their design, in order to control for potential differences in lab environments or subject pools.
Randomization unit
NA
Clusters
N=99
Observations
N=99
Arms
NA
Primary outcomes
Our research question is whether subjects’ preferences become more consistent over time. We examine this using our Distributional Approach to estimate risk preferences and to evaluate the predictive power of our method.

Cited by

No citation data available for this registration yet.

Keywords