Do Teams Alleviate or Exacerbate the Extrapolation Bias in the Stock Market?

How to cite

Pezone, Vincenzo; Cassella, Stefano; Barahona, Ricardo; Jansen, Kristy. 2024. "Do Teams Alleviate or Exacerbate the Extrapolation Bias in the Stock Market?." AEA RCT Registry. AEARCTR-0013710. https://doi.org/10.1257/rct.13710-1.0

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

Vincenzo Pezone* , Stefano Cassella , Ricardo Barahona , Kristy Jansen

* Primary investigator

Registry ID
AEARCTR-0013710
DOI
https://doi.org/10.1257/rct.13710-1.0
First registered
Last updated
Status
In development
Trial dates
2024-05-29 – 2024-06-19
Countries
Private
JEL codes
G41, G10
Registry record
http://www.socialscienceregistry.org/trials/13710

Abstract

Investors have been found to attribute an excessive role to more recent return realizations when making predictions, even though at low to medium horizons stock returns exhibit little autocorrelation. This tendency is known as “extrapolation bias.” Using an online survey, we test whether team interaction mitigates or amplifies such a bias.

Intervention

Respondents are presented with a series displaying the past realizations of a hypothetical stock return. They are then asked to predict the next realization in the series. Once they make the prediction, the actual realization is displayed, and they are asked to predict the next realization. The task is repeated twenty times.
There are three treatment arms. In the first arm, participants make predictions individually. In the second arm, participants are randomly paired and communicate via a chat box to agree on a joint decision. In the third arm, participants first make individual predictions and then make joint decisions in two-person teams (it is thus a combination of the first two arms). They are compensated based on the accuracy of their predictions.
We will test how the tendency to rely on the most recent realization to make predictions varies between the different treatments.
Participants are then presented with a survey and asked to respond some demographic questions.

Experimental design

Respondents are presented with a series displaying the past realizations of a hypothetical stock return. They are then asked to predict the next realization in the series. Once they make the prediction, the actual realization is displayed, and they are asked to predict the next realization. The task is repeated twenty times.
There are three treatment arms. In the first arm, participants make predictions individually. In the second arm, participants are randomly paired and communicate via a chat box to agree on a joint decision. In the third arm, participants first make individual predictions and then make joint decisions in two-person teams (it is thus a combination of the first two arms). They are compensated based on the accuracy of their predictions.
Participants are then presented with a survey and asked to respond some demographic questions.
The purpose of this experiment is to test how the tendency to rely on the most recent realization to make predictions varies between the different treatments.

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
Unit
Clusters
1250
Observations
1250
Arms
250 for the first arm 500 for the second arm 500 for the third arm The sample size for arms 2 and 3 is double the sample size in arm 1 to account for the fact that the choices of the participants paired in teams are going to be correlated.
Primary outcomes
Stock return forecasts

Cited by

No citation data available for this registration yet.

Keywords