Evaluating the Impact of AI-Driven Training on Microenterprises: Evidence from Pakistan

How to cite

Shaukat, Sarah; Rehman, Attique. 2023. "Evaluating the Impact of AI-Driven Training on Microenterprises: Evidence from Pakistan." AEA RCT Registry. AEARCTR-0012371. https://doi.org/10.1257/rct.12371-1.0

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

Sarah Shaukat* , Attique Rehman

* Primary investigator

Registry ID
AEARCTR-0012371
DOI
https://doi.org/10.1257/rct.12371-1.0
First registered
Last updated
Status
In development
Trial dates
2024-02-01 – 2025-05-31
Countries
Private
Registry record
http://www.socialscienceregistry.org/trials/12371

Abstract

A standard business training program is classroom-based and teaches a range of better business practices. Evaluations of these programs show mixed evidence, particularly in developing countries, and highlight that selection matters. Additionally, there are concerns around the inability to adapt content to the specific needs of the businesses and the high costs of training. It is thus important to find ways to share customized knowledge, reduce the cost of training, and identify policies to help the most dynamic entrepreneurs grow. To deal with these concerns, this project focuses on teaching high-growth-potential microenterprise owners how to use generative AI, specifically ChatGPT, for their business activities.

We aim to use a randomized control trial to study how training business owners in using ChatGPT to then in turn train themselves affects their productivity and entrepreneurial success. We aim to explore this question in the context of Pakistan, a lower-middle-income country in South Asia. Given the varying entrepreneurial challenges and opportunities that men and women encounter in a developing country like Pakistan, we also plan to conduct a heterogeneity analysis to dissect the impact of this AI-based training intervention through the gender lens.

Experimental design

Our approach will involve a randomized controlled trial, comprising of two treatment arms and a control group, with each group consisting of around 400 small firms, stratified by gender (power calculations will determine the exact sample size). Firms will either receive generative AI training or training on how to use Google for their business activities in the two treatment arms. The control group will receive no training.

Design details

Randomization method
randomization in Stata
Randomization unit
individual
Clusters
NA
Observations
1200
Arms
400 in each
Primary outcomes
business knowledge, profits, sales, number of customers

Cited by

No citation data available for this registration yet.

Keywords