Narratives for Credit

How to cite

Neshat, Navid; Kim, Yongseok. 2025. "Narratives for Credit." AEA RCT Registry. AEARCTR-0016729. https://doi.org/10.1257/rct.16729-1.0

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

Navid Neshat* , Yongseok Kim

* Primary investigator

Registry ID
AEARCTR-0016729
DOI
https://doi.org/10.1257/rct.16729-1.0
First registered
Last updated
Status
In development
Trial dates
2025-12-01 – 2026-01-31
Countries
Afghanistan ()
Secondary IDs
G21
Registry record
http://www.socialscienceregistry.org/trials/16729

Abstract

This study investigates the impact of loan applicants’ narratives on loan approval decisions.

Intervention

We use a set of loan applications submitted to a bank in Afghanistan. These applications include standard financial information such as account balances, as well as a non-conventional section where applicants describe in their own words why the bank should trust their credibility and commitment to repayment. We refer to this as the narrative section. Our study examines how loan officers evaluate these narratives by recruiting loan officers to review applications and assign a score indicating how likely they believe the applicant is to repay the loan at maturity. To identify the effect of narratives, each application is evaluated by multiple officers, with randomization occurring across officers: some see the full application, while others see the same application with the narrative section removed.

Experimental design

This study investigates how loan officers evaluate borrower narratives when making lending decisions. We use a set of loan applications submitted to a bank in Afghanistan. These applications include standard financial information such as account balances, as well as a non-conventional section where applicants describe in their own words why the bank should trust their credibility and commitment to repayment. We refer to this as the narrative section. To examine the effect of narratives, we recruit loan officers toto review applications and assign a score indicating how likely they believe the applicant is to repay the loan at maturity. Randomization occurs at the application level, where the narrative section is either shown or hidden.

Design details

Randomization method
Randomization is done by a computer.
Randomization unit
Loan officer–application pair
Clusters
The number of clusters is equal to the number of loan officer–application pairs included in the experiment, expected to be around 2,500 evaluations (25 officers × 100 applications each).
Observations
Approximately 2,500 loan officer–application evaluations (about 25 officers reviewing 100 applications each).
Arms
Narrative shown (full application): ~50 percent of observations (≈1,250). Narrative hidden (full application without narrative section): ~50 percent of observations (≈1,250)
Primary outcomes
The primary outcome is the loan officer’s evaluation score, measured on a scale from 1 to 10, reflecting how likely they believe the applicant is to repay the loan by the maturity date, where 1 indicates “very unlikely to repay.”

Cited by

No citation data available for this registration yet.

Keywords