Digital lenders should use AI to reduce default rates

Digital lending in Kenya is steadily growing with 35 per cent of mobile phone owners, approximately 6.1 million, being digital borrowers with active loans at 45 per cent from Hustler Fund, 28 per cent from mobile banking platforms like Mshwari and KCB M-Pesa and 23.7 per cent from chamas or groups, according to FinAccess Digital Credit Tracker Survey 2023.

Consequently, one in eight borrowers have defaulted in their loans and 47 percent do late payment. Digital lending firms operate in a volatile and unregulated environment, therefore credit risk assessment is inevitable.

Statistics from Central Bank of Kenya (CBK) indicates that from the approved 51 service providers, there is an increase in non-performance digital loans portfolio from the entire bank loan book of 11.4 percent in comparison to NPL of 9.7 percent in all bank loans, hence the need for a credit scoring models by digital lenders.

Conventional credit scoring has proved unsatisfactory since most digital loans are issued on daily basis, therefore requiring daily credit score ratings as well.

Loan recovery

Many households and businesses have suffered immensely from the austere measures taken by digital lenders on loan recovery. The worst case scenario is when borrowers’ information is listed with Credit Reference Bureau (CRB), therefore negatively affecting their credibility. CBK has mandated CRBs to embrace other sophisticated credit scoring models with risk-based credit pricing for quality customer reporting as well as digitising on Credit Information Sharing (CIS) framework for proper rating scores prior to credit approvals.

Structural credit risk models in AI have been proposed in machine learning tools in credit risk prediction aimed at obtaining borrowers’ probability of default. Digital loans in Kenya are easily available through mobile apps at high interest rates ranging from six to 10 percent monthly for a one month’s loan, which is expensive compared to conventional formal loans. With frothy background checks and inadequate credit history of borrowers, many digital lenders have encountered huge NPLs, that are common in KCB Mshwari, Fuliza and Hustler Fund loans issued without checking on the credit risk and probability of default.

However, as a way of adhering to consumer protection laws, several complaints from the public are under investigations. They are linked to exorbitant interest rates and cases where borrowers’ details are exposed by contacting close family members whenever there are cases of defaulting. This is proving to be punitive and derogative.

Whilst several red flags have been sighted ,digital loans have significantly contributed to the massive number of defaulters reported to CRB, a total of 2.7 million Kenyans — an average of 400,000 individuals with loans of less than Sh200.

Credit bureaus should not rely on data from their peers but instead embrace AI tools in credit risk modelling and default prediction. Interest rates should also be moderated across all digital lenders.

The writer is a Nairobi-based member of Institute Certified Public Accountant (ICPAK), [email protected]