Measuring Personality Traits and Predicting Loan Default with Experiments and Surveys
Last registered on April 12, 2017

Pre-Trial

Trial Information
General Information
Title
Measuring Personality Traits and Predicting Loan Default with Experiments and Surveys
RCT ID
AEARCTR-0001840
Initial registration date
April 12, 2017
Last updated
April 12, 2017 5:05 PM EDT
Location(s)
Region
Primary Investigator
Affiliation
Northwestern University
Other Primary Investigator(s)
PI Affiliation
Harvard University
PI Affiliation
Bureau of Labor Statistics
Additional Trial Information
Status
Completed
Start date
2005-01-01
End date
2009-01-01
Secondary IDs
Abstract
We use two natural field experiments and surveys to identify character elements, and test whether these traits can be used to predict the likelihood of loan default. In the first experiment we identify subjects with high psychosomatic moral costs by observing their reactions when a bank error is made in their favor. In the second experiment we identify subjects that were less naïve about their own ability to meet future commitments. We found that both individuals with higher moral costs and individuals who were the least naïve displayed lower default rates than other groups. We also explore the relationship between qualitative survey-based social capital measures and loan default. We find that survey-based social capital measures are not predictive of loan default for these individual loans, contrary to the results from a prior study -with group loans. Lastly, we examine whether more general personality index measures predict default, and we find that they do not. Overall, the lessons present evidence of moral hazard in microentrepreneurial credit markets to the extent that they reflect choice by the borrower about whether to repay. They also show the potential for adverse selection insofar as these personality measures are typically unobservable to the lender.
External Link(s)
Registration Citation
Citation
Karlan, Dean, Sendhill Mullainathan and Omar Robles. 2017. "Measuring Personality Traits and Predicting Loan Default with Experiments and Surveys." AEA RCT Registry. April 12. https://doi.org/10.1257/rct.1840-1.0.
Former Citation
Karlan, Dean et al. 2017. "Measuring Personality Traits and Predicting Loan Default with Experiments and Surveys." AEA RCT Registry. April 12. http://www.socialscienceregistry.org/trials/1840/history/16540.
Sponsors & Partners

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

Request Information
Experimental Details
Interventions
Intervention(s)
In this project, the researchers tested whether personal character traits can predict the probability of loan default among microfinance borrowers in the Philippines. They used two field experiments and two surveys to identify subsets of borrowers with certain character traits. Then, the researchers tested whether the default rate for these borrowers was significantly different from the default rate of the rest of the sample borrowers and whether these traits are predictive of default. In the first experiment, researchers identified sample borrowers with high psychosomatic moral costs. In the second, researchers identified sample borrowers that were less naïve about their ability to meet future commitments. The researchers then used surveys to measure sample borrowers' social capital and Big Five personality index score.

The researchers found that sample borrowers with high psychosomatic moral costs and borrowers that were less naïve about their ability to repay were less likely to default than other sample borrowers. Survey measures of social capital was not predictive of default, nor was a borrower's Big Five personality index score.
Intervention Start Date
2005-01-01
Intervention End Date
2006-12-31
Primary Outcomes
Primary Outcomes (end points)
Probability of Default; Fraction of loan defaulted on
Primary Outcomes (explanation)
Secondary Outcomes
Secondary Outcomes (end points)
Secondary Outcomes (explanation)
Experimental Design
Experimental Design
In the first experiment, sample borrowers were offered a cash reward for completing a baseline survey. While redeeming their reward, the bank cashiers intentionally gave the recipients too much cash. Those who left the bank with the excess cash and later returned it were hypothesized to have high moral standards, or a high psychosomatic cost of immoral behavior, and thus would be less likely to default. In the second experiment, borrowers were randomized into one of two groups: those asked to respond to a short survey in seven days via text message in order to win a cash reward, and those given a choice to either answer the survey in seven days via text message or to complete the survey the same day for a smaller cash reward. Those who chose to complete the survey the same day were hypothesized to be rationally self-aware and less likely to default, while those who chose the option to text in seven days but failed to do so were hypothesized to be naïvely over-optimistic and thus more likely to default. In the social capital survey, participants answered questions from the General Social Survey on trust, fairness, and helpfulness. Borrowers that indicated higher levels of social capital were hypothesized to have lower default rates than the remainder of the sample participants. In the Big Five personality survey, participants responded to the Ten-Item-Personality-Index.

Experimental Design Details
Randomization Method
For the second experiment, participants were randomly assigned at baseline to one of two experimental groups: those instructed to text survey responses after a week for a reward, and those given a choice to text responses a week later for the same reward or to receive a smaller reward while still in the bank
Randomization Unit
individual
Was the treatment clustered?
No
Experiment Characteristics
Sample size: planned number of clusters
725 individuals
Sample size: planned number of observations
725 individuals
Sample size (or number of clusters) by treatment arms
roughly half of the participants were assigned to the text-only group; the other roughly half were assigned to the choice group
Minimum detectable effect size for main outcomes (accounting for sample design and clustering)
IRB
INSTITUTIONAL REVIEW BOARDS (IRBs)
IRB Name
IRB Approval Date
IRB Approval Number
Post-Trial
Post Trial Information
Study Withdrawal
Intervention
Is the intervention completed?
Yes
Intervention Completion Date
December 31, 2006, 12:00 AM +00:00
Is data collection complete?
Yes
Data Collection Completion Date
January 01, 2007, 12:00 AM +00:00
Final Sample Size: Number of Clusters (Unit of Randomization)
725 participants
Was attrition correlated with treatment status?
Final Sample Size: Total Number of Observations
725 participants
Final Sample Size (or Number of Clusters) by Treatment Arms
roughly half of the participants were assigned to the text-only group; the other roughly half were assigned to the choice group
Data Publication
Data Publication
Is public data available?
No
Program Files
Program Files
Reports and Papers
Preliminary Reports
Relevant Papers