Page 62 - State-of-the-Industry-2013

State OF the induStry 2013

TexT BOx 19
airtime-baseD creDit scoring:
can it Drive innovative loan ProDucts for mobile moneY?*

With the vast majority of mobile subscribers in emerging markets using pre-paid SIMs, there is incredibly rich data available on
airtime purchase and usage behaviours. Couple that with the poor quality of information available at credit bureaus, and airtime-
based credit scoring is an attractive approach to targeting and risk profling for credit products.

As a result, customer patterns of airtime top-ups are being used to determine the credit-worthiness of a prospective borrower and
to approve/deny loans. Will this technique facilitate the development of innovative microloans via the mobile channel? Is the data
truly an adequate predictor of customers’ ability to repay? What’s the upside for the players involved?

AIRTIME-BASED CREDIT SCORING IN A NUTSHELL

Most emerging markets have little to no infrastructure that adequately collects customers’ credit history: credit bureaus either don’t
exist, or exist on a limited number of individuals and with very thin fnancial data. For individuals without credit history, the result is
stringent borrowing terms such as high collateral coverage, months of demonstrated savings, and/or individual or group guarantors.

The idea behind airtime-based credit scoring is to use an individual’s history of airtime top-up as a proxy indicator of what amount
they can aford to borrow and their credit-worthiness. The precise calculations and algorithms employed to do this is the “secret
sauce” of Experian MicroAnalytics and Cignif, two companies working in this space.

HOW TO MANAGE CREDIT RISK ON MOBILE LOANS?

Four distinct components are required to make mobile branchless loans work. These are:

1. An origination credit scoring system that utilizes the information available on the borrower at the time of application to predict
credit risk. The key predictors of risk are: airtime top-up patterns (for example, do you top-up large amounts once a month or
small amounts every other day?); voice and SMS usage; information gathered directly from the borrower (for example income,
marital status, etc.); and information available externally (for example, where available, from a credit bureau). When combining
this data it is possible to develop scorecards that discriminate well credit risk.
2. An automated customer management system to send alerts to borrowers to remind them a payment is due, to increase or
decrease dynamic exposure to good/bad borrowers, and to streamline the management of overdue payments.
3. A credit risk agent management system to dynamically rank agents by the quality of the clients they have introduced to the
bank and to calculate and disburse risk-adjusted commissions. In addition, the system alerts agents when their introduced
clients are late with a loan payment to trigger early collections action.
4. An enhanced mobile interface for end clients that allows them to manage their credit product and review, for example, when
their instalment is due, make anticipated payments, request additional credit lines, etc., all managed in an automated and
real-time fashion.

*This text box was adapted from blog posts by Yasmina McCarty, published on the MMU website on March 23, 2012 and December 6, 2012, and by Elio Vitucci, Managing Director of
Experian MicroAnalytics, published on August 7, 2012

56