Visual Nudges in Lending

How to cite

Yoon, Joon Sang; Wittenberg-Moerman, Regina; Zhang, Tianyu; Huang, Kanyuan; Tang, Mengwei. 2024. "Visual Nudges in Lending." AEA RCT Registry. AEARCTR-0014536. https://doi.org/10.1257/rct.14536-2.0

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

Joon Sang Yoon* , Regina Wittenberg-Moerman , Tianyu Zhang , Kanyuan Huang , Mengwei Tang

* Primary investigator

Registry ID
AEARCTR-0014536
DOI
https://doi.org/10.1257/rct.14536-2.0
First registered
Last updated
Status
In development
Trial dates
2025-09-15 – 2025-12-15
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/14536

Abstract

In this study, we examine whether providing a simple visual nudge by labeling the loan product can change borrowers’ loan choices, potentially helping them make better decisions. We work with a government-backed online lending platform in China. The platform allows borrowers to directly borrow money from banks. Banks will list their loan products on the platform. Any individual with a valid Chinese ID can register with the platform and apply for loans. We randomly assign borrowers into one of the following four groups: 1) a control group with no label; 2) a treatment group where a label is placed on the loan with the lowest interest rate; 3) a treatment group where a label is placed on the loan with the fastest processing time; and 4) a treatment group where a label is placed on the loan with the highest approval rate. By comparing borrowers’ loan choices between the control and treatment groups, we estimate the causal impact of visually salient labels on borrower decision-making in consumer lending.

Intervention

We work with a government-sponsored lending platform in China. When borrowers apply for loans on the platform, the platform will prompt a survey asking the borrower to fill in basic background information, such as education and income range, as well as their requested loan amount maturity, and the ability to provide collateral.
Based on the borrower’s loan application, the platform will recommend five loan products to the borrower. The borrower can apply for one or more of those recommended loans. Once the borrower applies, the bank is notified and will start to process the loan application.
Treatment: Once the borrower completes the survey, she will be randomly assigned to one of the following groups:
a. Control group: no label.
b. Treatment 1: Lowest rate label – we place a label on the loan with the lowest interest rate.
c. Treatment 2: Fastest processing time label - we place a label on the loan with the fastest processing time
d. Treatment 3: Highest success rate label - we place a label on the loan with the highest approval rate.
We observe whether the borrower applies for a loan and which loan product the borrower picks.

Experimental design

We work with a government-sponsored lending platform in China. When borrowers apply for loans on the platform, the platform will prompt a survey asking the borrower to fill in basic background information, such as education and income range, as well as their requested loan amount maturity, and the ability to provide collateral.
Based on the borrower’s loan application, the platform will recommend five loan products to the borrower. The borrower can apply for one or more of those recommended loans. Once the borrower applies, the bank is notified and will start to process the loan application.
Treatment: Once the borrower completes the survey, she will be randomly assigned to one of the following groups:
a. Control group: no label.
b. Treatment 1: Lowest rate label – we place a label on the loan with the lowest interest rate.
c. Treatment 2: Fastest processing time label - we place a label on the loan with the fastest processing time
d. Treatment 3: Highest success rate label - we place a label on the loan with the highest approval rate.
We observe whether the borrower applies for a loan and which loan product the borrower picks.

Design details

Randomization method
We work with the lending platform to implement a dual-randomization design—across both treatment assignment and loan order, which enable us to examine the causal effect of visually salient labels on borrower loan choices. Treatment assignment randomization: we randomly assign a borrower to the control group or one of the three treatment groups. The randomization is done via a random number generator that generates a number between 1 to 4. The probability of entering each group is 25%. Loan order randomization: we also randomize the order in which loan products are displayed. This ensures that the labeled loan is equally likely to appear in any position on the loan product list (i.e., first, second, third, etc.)
Randomization unit
Treatment assignment is randomized at the individual borrower level; loan order randomization is implemented at the loan level.
Clusters
No
Observations
About 1,500 borrowers.
Arms
Approximately 1,500 borrowers divided into four groups, with about 375 in the control group, and 375 in each of the treatment arms.
Primary outcomes
1. Whether the borrower applies for a loan. 2. The loan products that the borrower selects.

Cited by

No citation data available for this registration yet.

Keywords