What could possibly go wrong? Predictable Misallocation in Simple Debt Repayment Experiments

Last registered on September 08, 2020
View Trial History

Pre-Trial

Trial Information

General Information

Title
What could possibly go wrong? Predictable Misallocation in Simple Debt Repayment Experiments
RCT ID
AEARCTR-0006400
Initial registration date
September 05, 2020

Initial registration date is when the trial was registered.

It corresponds to when the registration was submitted to the Registry to be reviewed for publication.

First published
September 08, 2020, 9:36 AM EDT

First published corresponds to when the trial was first made public on the Registry after being reviewed.

Locations

Country
Germany
Region
Country
United States of America
Region

Primary Investigator

Name
Florian Gärtner
Affiliation
Justus-Liebig-Universität Giessen
Contact Primary Investigator

Other Primary Investigator(s)

PI Name
Darwin Semmler
PI Affiliation
Justus-Liebig-University Giessen
Contact Investigator
PI Name
Christina Bannier
PI Affiliation
Justus-Liebig-University Giessen
Contact Investigator

Additional Trial Information

Status
Completed
Start date
2018-08-01
End date
2020-09-01
Keywords
Finance & Microfinance, Lab
Additional Keywords
MTurk
JEL code(s)
Secondary IDs
Abstract
How do people repay debt? In a simple debt repayment experiment we provide subjects with two credit cards with different interest rates and levels of debt that are to be repaid. From a rational choice perspective, this is arguably one of the simplest financial decisions. Nevertheless, we observe severe deviations from optimal, i.e. debt minimizing, repayment decisions with one particularly persistent type of misallocation that has not been found before. In consecutive experiments we show that this and further fallacies are predictable so that behavior can be steered towards more efficient repayment decisions.
External Link(s)

Registration Citation

Citation
Bannier, Christina, Florian Gärtner and Darwin Semmler. 2020. "What could possibly go wrong? Predictable Misallocation in Simple Debt Repayment Experiments." AEA RCT Registry. September 08. https://doi.org/10.1257/rct.6400-1.0
Sponsors & Partners

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information
Experimental Details

Interventions

Intervention(s)
We test in an online and lab context wether people repay debts optimally, i.e. in an interest-minimizing way. We also try to predict patterns in deviations from optimal behavior, and try to increase optimal behavior.
Intervention Start Date
2018-08-01
Intervention End Date
2019-07-31

Primary Outcomes

Primary Outcomes (end points)
Misallocation of income
Primary Outcomes (explanation)
Misallocation is defined as the fraction of income not used to repay the high interest card.

Secondary Outcomes

Secondary Outcomes (end points)
To test if financial literacy leads to more optimal behavior
Secondary Outcomes (explanation)

Experimental Design

Experimental Design
We use an incentivized debt repayment experiment to test how our participants use income to repay debts. We endow our subjects with two virtual credit card accounts, which both have 2200$ virtual debts and charge 3% or 5% interest rate per round, respectively. In each of the ten rounds subjects get a virtual income of 250$ that they can use to repay these debts. We meassure how much money is NOT used to repay the 5% card.
In experiment #1, we test if the basic paradigm works on MTurk and in the lab.
In experiment #2, we test if we can use the way we present information to increase or decrease optimal (=interest minimizing) behavior. Same experimental design as described, but some variation in the way we present information.
In experiment #3, we test if we can predict seven non-optimal repayment strategies, using a modified one shot game version of our experiment. We design seven pairs of scenarios ("scenario" being defined as a single combination of starting debts, interest rates and income) in which we either try to provoke a certain repayment heuristic, or suppress it. Every scenario pair differs in exactly one value, except for one, where two values differ.
Experimental Design Details
Randomization Method
In experiment #1, we replicate MTurk results in the lab to show that both subject pools behave comperable. Randomizing here is not necessary.
In experiment #2, we post sessions of all three treatments on MTurk at the same time using the same wording to advertize them. Technically this is self selection which could be seen as participants self-selecting themselves into a lab experiment where treatment 1 is conducted in one session and treatment 2 in another. Since participants cannot differentiate the treatments, there is no variable to self select for.
Experiment #3 has only one treatment.
Randomization Unit
Session
Was the treatment clustered?
No

Experiment Characteristics

Sample size: planned number of clusters
Overall planned: 820
Sample size: planned number of observations
Overall planned: 820
Sample size (or number of clusters) by treatment arms
Experiment #1: 130 on MTurk, 100 in the lab
Experiment #2: 130 per treatment. Experiment #1 is a replication of the control group of experiment #2, we do not double count th MTurk treatment.
Experiment #3: only one treatment, 330.
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
Supporting Documents and Materials
IRB

Institutional Review Boards (IRBs)

IRB Name
IRB Approval Date
IRB Approval Number
Analysis Plan

Post-Trial

Post Trial Information

Study Withdrawal

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Intervention

Is the intervention completed?
Yes
Intervention Completion Date
July 31, 2019, 12:00 +00:00
Data Collection Complete
Yes
Data Collection Completion Date
July 31, 2019, 12:00 +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
835
Was attrition correlated with treatment status?
No
Final Sample Size: Total Number of Observations
835
Final Sample Size (or Number of Clusters) by Treatment Arms
Experiment #1: 131 on MTurk, 96 in the lab; Experiment #2: 131, 135 and 138 (Experiment #1 is a replication of the control group of Experiment #2, we do not double count the MTurk treatment); Experiment #3: only one treatment, 335.
Data Publication

Data Publication

Is public data available?
No

There is information in this trial unavailable to the public. Use the button below to request access.

Request Information

Program Files

Program Files
Reports, Papers & Other Materials

Relevant Paper(s)

Reports & Other Materials