RCT in FinTech Lending

How to cite

Xiao, Steven; Xia, Han; Tang, Xiaoxiao. 2020. "RCT in FinTech Lending." AEA RCT Registry. AEARCTR-0005489. https://doi.org/10.1257/rct.5489-1.0

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

Steven Xiao* , Han Xia , Xiaoxiao Tang

* Primary investigator

Registry ID
AEARCTR-0005489
DOI
https://doi.org/10.1257/rct.5489-1.0
First registered
Last updated
Status
In development
Trial dates
2020-03-01 – 2021-12-31
Countries
Private
JEL codes
G20, G50
Registry record
http://www.socialscienceregistry.org/trials/5489

Abstract

We collaborate with a leading Chinese FinTech lending company, to conduct a randomized controlled trial (RCT) to test for information asymmetry in the FinTech consumer credit market. In the experiment, we plan to manipulate three contract terms: interest rate, loan amount, and time to maturity, and offer randomized contracts to loan applicants. We then observe the impact of various treatments (contracts) on: (1) contract acceptance rate by borrowers, (2) loan default rate among the executed loans, (3) gross and net IRR of the loan. We hypothesize that among the pool of borrowers that share the same level of credit score, those who accept less favorable contract terms (e.g. higher interest rate, lower loan amount, and shorter time to maturity) tend to have higher unobservable credit risk, and have higher default rate that cannot be fully explained by credit score. Further, we hypothesize that there may exist a group of borrowers for whom none of the offered contracts can result in profitable lending (on average) because of the adverse selection problem.

Intervention

The company will manipulate three contract terms: interest rate, loan amount, and time to maturity.

Experimental design

The company will randomly manipulate three contract terms it offers to a loan application: interest rate, loan amount, and time to maturity.

Design details

Randomization method
Randomization done by computer.
Randomization unit
Individual
Clusters
3 to 6 credit score groups.
Observations
4000-5000 customers.
Arms
All customers are treated with different contracts.
Primary outcomes
(1) contract acceptance rate by borrowers, (2) loan default rate among the executed loans, (3) gross and net IRR of the loans.

Cited by

No citation data available for this registration yet.

Keywords