LISTEN NOW: Alternative Data for MSME Credit Scoring


Author: Sarah Corley

This webinar explored the promise and potential of alternative credit scoring for MSMEs, looking at the gaps in the market, examples of products and the opportunities and challenges. It also raises some of the ethical issues, including bias, privacy and managing risk. Panelists for the webinar were: Islam Zekry, Chief Data Officer at Commercial International Bank; Alexandra (Kobishyn) Rizzi, Senior Research Director at Center for Financial Inclusion; Nicole Van Der Tuin, Co-Founder & CEO at First Access; and the webinar was moderated by Elizabeth Friend, Managing Director at The Small-Scale Sustainable Infrastructure Development Fund.



Why do MSME’s matter? Of the $8.9 trillion needed annually, only $3.7 trillion is available leaving a $5.2 trillion finance gap. These figures are underestimates as when you start to add in informal MSMEs the number and gap will increase. MSMEs can bring great innovation, increase financial inclusion and have an important role to play in the economy.

Traditional lending using past credit history and/or cash flow to create a credit score and determine whether credit can be offered. When these are not available or are not complete then challenges arise, which is the case for many MSMEs. Many MSMEs are financially excluded as the lack of data on them does not give the FSP the ability to trust them and give them credit. The challenge for FSPs is not with having the funds to loan but having appropriate trust in their customers to authorize credit responsibly.

This is where alternative data can help create a credit score using different types of data such as telecom, internet, financial transaction, social network, e-commerce, psychometric and mobile data. Alternative data is available and becomes increasingly available as we digitise – mobile phone use is increasing and leaves a digital footprint about the user. The challenge is how the data can be used in a way to create trust in a customer and allow the FSP to be comfortable with using the data and have a profitable product. It takes experimentation and pilots to be able to generate algorithms that work, which is time and resource consuming. There is a tension between risk and reward for FSPs, many are taking the steps, particularly as regulators begin to take steps to encourage use of alternative data through AI and machine learning to drive their financial inclusion mandates.

FSPs can consider partnering with organisations, such as in Latin America where banks are partnering with retailers to provide credit. Retailers can provide a stable history on the customer, who they are, how to contact them and patterns of spending. The bank can then offer a small credit product and can then begin to create a credit history with that customer.

Consumer protection and bias too need consideration. Algorithms will be proprietary but ensuring access to information on sources, assumptions, training, outcomes, models and adjustment processes can enable practices to be checked so they are not further excluding marginalised groups.