There’s a new weapon in the fight to expand financial access.
The Entrepreneurial Finance Lab, founded by faculty and students from the Harvard Kennedy School and Harvard Business School, is pioneering new personality-assessment based tools to expand credit access. Survey-based measures of personality characteristics – such as ethics, character, intelligence, attitudes and beliefs – combined with measures of business skills turn out to be powerful predictors of loan repayment in real-world settings. The Entrepreneurial Finance Lab creates alternative credit scores based on these characteristics to expand credit access in partnership with banks and microfinance institutions from around the world.
The approach originates from research in both psychology and business administration showing that cognitive indicators are among the best selection criteria for job applicants and determinants of successful entrepreneurs. According to the Lab’s website, use of personality and cognitive assessments in employment screening is on the rise. They want to import these insights and the screening technologies developed from them to the frontier of finance in the developing world.
There are significant challenges in the practical implementation of this method, and this technical note describes some of their approaches in more detail. One challenge is how specifically to tailor the model to each country or bank. Models based on small amounts of data may not have enough predictive power. While aggregating information across banks and countries is one option, aggregation may be less than ideal if the effects of different measured characteristics vary across contexts. Another complication is how to make the most of the collected data, and the method developed by the Lab allows scoring to vary with each item in the survey, rather than with more aggregated indices, allowing for greater predictive capability.
Once developed, however, the model and the delivery strategy adopted by the Lab may be one of the most scalable interventions in the realm of financial access proposed and executed to date. The computer-based assessment can be delivered quickly on a non-networked tablet, PC, or mobile device. It can create a score that is strongly predictive of repayment behavior without additional financial information from the borrower. In fact, the Lab’s model is more predictive than traditional credit scores of repayment behavior, even in developed country credit markets.
This method has already been implemented in dozens of countries with considerable success, enabling loans to many thousands of entrepreneurs who would not have had access to formal finance otherwise. It is a smart and innovative use of “Big Data” techniques to lower costs of service provision and access in an important sector. We hope to see its scope grow, and we hope to see more breakthroughs like this going forward.