On the Microeconomic Foundations of the Role of Immigration on Growth: Experimental Evidence from Galgo Credits
How to cite
Undurraga, Raimundo; Montoya, Ana Maria; Gertler, Paul. 2025. "On the Microeconomic Foundations of the Role of Immigration on Growth: Experimental Evidence from Galgo Credits." AEA RCT Registry. AEARCTR-0016224. https://doi.org/10.1257/rct.16224-1.0
Download: BibTeX · RIS — for reference managers without a browser translator. Zotero and most others pick up this page's citation metadata directly.
* Primary investigator
Abstract
This project examines the challenges that international migrants face when integrating into local labor and credit markets. Galgo employs innovative scoring methods and diverse data sources to provide capital credits in the form of motorbikes to vulnerable entrepreneurs in Colombia, which are often overlooked by traditional banking institutions. Specifically, Galgo's loan portfolio targets both native and immigrant borrowers, with a significant number of loans granted to recent Venezuelan migrants. Through a randomized experiment, the project aims to assess the impact of the motorbike loans on capital productivity for both migrant and native entrepreneurs, considering loan repayment rates, sales, revenues, and labor market indicators as outcomes. In addition, the study will analyze potential credit discrimination against migrants and its impact on capital productivity among entrepreneurs.
Intervention
Galgo, a Fintech company operating in Colombia, Mexico, Peru, and Chile, offers motorbike loans through an online platform as well as face-to-face at point of sale (POS) locations, enabling clients to access credit seamlessly through their preferred channel. What distinguishes Galgo from other Fintechs is its unique business model, which was specifically designed to address the challenges migrants face in accessing credit. Initially focused exclusively on serving migrant clients, the company has since expanded its reach to include a mixed portfolio of both native and migrant customers in each country.
In order to minimize credit risk, Galgo employs a novel risk-scoring procedure that integrates diverse financial data sources from both origin and destination (host) countries to thoroughly evaluate each loan application. This process goes beyond traditional credit assessments by incorporating a broad spectrum of variables. In addition to basic demographics (such as age and gender) and the specific attributes of the motorbike the client aims to acquire, Galgo leverages detailed information about the applicant's debt history, traffic fines, and driver license characteristics—sourced from both their origin and host countries. These factors are carefully weighted within a dynamic and adaptive risk-scoring model that continuously refines its predictions based on evolving data patterns. By adopting this comprehensive and data-driven approach, Galgo ensures that motorbike loans are extended with fair and reasonable credit conditions, making financing accessible to immigrant and vulnerable populations who often face restrictive terms or outright exclusion from traditional banking institutions.
Experimental design
We randomize Galgo's loans through a rolling randomization procedure, where individual applicants are assigned to treatment and control groups on a case-by-case basis. To ensure balance across key characteristics, randomization is stratified by groups of interest. Specifically, applicants are stratified based on their native or migrant status and their classification as having a high or low risk score, with randomization occurring within these strata.
The dynamics of the demand for Galgo loans is short-term: clients submit their application in any point in time, and a fast response is critical to secure participation and take-up. Hence, it is not feasible to pool the full experimental sample of applicants and randomize at once. With that in mind, we designed an online and automated operation system that randomizes in real time, case by case, as loan applications come in. In addition to make randomization possible by reducing the costs of waiting for a large batch of cases, the platform used for the operation allows to supervise and give immediate feedback to both ends of the team (Galgo and research team).
Importantly, not all loan applications participate in the experiment, but only a random sample of them. While the whole potential clients of Galgo pass through the platform, some of them are randomly selected to participate in the experiment, which are then randomly assigned to treatment or control status.
Design details
Cited by
No citation data available for this registration yet.