The microcredit literature has shown that access to credit has modest impacts on microenterprises, but it is possible that this is because the loans are too small to make a significant difference in firm outcomes. We are working with the a large micro-finance institution (MFI) in Egypt which has over 400,000 borrowers, called the Alexandria Business Association. Less than 1% of those borrowers have loans over $1000. We have worked with the lender to identify borrowers who are "high potential" and to provide them loans that are much larger than their current loan through a randomised experiment.
Loan officers reached out to existing borrowers that they thought would benefit from a larger loan. Borrowers were invited to apply and were told that if approved they would either receive a control loan (2x size of prior loan), or a treatment loan (at least 4x size of prior loan). 1004 businesses were approved for the loans and entered our sample and 87.5% took out the loan. We collected two rounds of follow up data from this sample of borrowers.
In addition to estimating the impacts of expanding access to credit for these businesses, we are also interested in improving our ability to identify ex-ante which businesses and entrepreneurs are most likely to benefit from this intervention and which are best able to grow in the future. We will do this using novel data and new machine learning methods.
We collected baseline data on the individuals in the sample before we randomised and then the MFI provided the loans. In the baseline we collected a large variety of data on firm and household characteristics, like age, education and experience of the owner, business revenues, expenditure and profits, and information on suppliers, credit access, marketing etc. We also collected a battery of psychometric questions which have borrowers report on a 1-5 scale how they feel about different statements like “I have always believe I am going to be successful” and “Success is never down to luck”.
With a novel experiment, new types of data, and cutting edge machine learning methods we’re excited to contribute to the literature on entrepreneurship and enterprise development.