Behavioral biases in annuity choice

How to cite

Taubinsky, Dmitry; Oliveira, Priscila de; Luttmer, Erzo. 2019. "Behavioral biases in annuity choice." AEA RCT Registry. AEARCTR-0004226. https://doi.org/10.1257/rct.4226-1.0

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

Dmitry Taubinsky* , Priscila de Oliveira , Erzo Luttmer

* Primary investigator

Registry ID
AEARCTR-0004226
DOI
https://doi.org/10.1257/rct.4226-1.0
First registered
Last updated
Status
On going
Trial dates
2019-05-10 – 2020-06-30
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/4226

Abstract

We will run an MTurk experiment to examine several basic biases that we think are present in annuity choice decisions. The study will use simple and abstract decision scenarios that consist of several "accounts" and several states that determine payoffs. The goal of this study will be to test for two biases. The first is the "insurance heuristic"--an aversion to moving currency to a state of the world in which payoffs are already high, even if the marginal utility from currency is also high in that state of the world. The second is status quo bias, which we conjecture to be stronger when the options are more complex.

Intervention

The study will use simple and abstract decision scenarios that consist of two "accounts" and two states that determine payoffs in account. The goal of this study will be to test for two biases. The first is the "insurance heuristic"--an aversion to moving currency to a state of the world in which payoffs are already high, even if the marginal utility from currency is also high in that state of the world. The second is status quo bias, which we conjecture to be stronger when the options are more complex, and will depend on the frame.

We will randomize four things, both between and within participants: 1) Which option is the status quo; 2) The expected difference in earnings between the two options; 3) The extent to which the "good" state in account 2 is better than the other state; and 4) whether the two options differ in all states in the second account or just in one of the states. We will also randomize how the status quo is framed. The "basic" frame just endows participants with one option and asks them if they want to exchange it for another option. The "enhanced" frame asks participants if they want to move currency from account 1 to account a state in account 2.

Experimental design

The study will use simple and abstract decision scenarios that consist of two "accounts" and two states that determine payoffs in account. The goal of this study will be to test for two biases. The first is a the "insurance heuristic"--an aversion to moving currency to a state of the world in which payoffs are already high, even if the marginal utility from currency is also high in that state of the world. The second is status quo bias, which we conjecture to be stronger when the options are more complex.The second is status quo bias, which we conjecture to be stronger when the options are more complex, and will depend on the frame.

We will randomize four things, both between and within participants: 1) Which option is the status quo; 2) The expected difference in earnings between the two options; 3) The extent to which the "good" state in account 2 is better than the other state; and 4) whether the two options differ in all states in the second account or just in one of the states. We will also randomize how the status quo is framed. The "basic" frame just endows participants with one option and asks them if they want to exchange it for another option. The "enhanced" frame asks participants if they want to move currency from account 1 to account a state in account 2.

The currency will be points, with each point translated to a 1% chance of winning a $5 prize.

Design details

Randomization method
Randomization done in office by a computer
Randomization unit
Individual
Clusters
1500 At this point, we ran a pilot with 326 individuals (clusters), which are going to be included in the final analysis.
Observations
9000
Arms
N/A: Randomization is both within and between participants.
Primary outcomes
The likelihood of choosing option A.

Cited by

No citation data available for this registration yet.

Keywords