RBI wants NBFCs to use AI to spot borrower stress
Reserve Bank of India Deputy Governor S C Murmu urged non-banking financial companies (NBFCs) to leverage artificial intelligence and machine learning to identify early indicators of borrower stress during his address at the 7th CII NBFC and HFC Summit. Murmu emphasized that NBFCs and housing finance companies (HFCs) have transformed from alternative lenders into specialized, technology-driven institutions that support banks in serving underserved populations and remote areas.
AI and machine learning tools can be utilized to enhance early warning systems, dynamic provisioning, and stress testing, according to Murmu. He cautioned against sacrificing underwriting standards for rapid growth expansion. NBFC credit currently constitutes 16.7% of nominal GDP, up from 15.9% the previous year, and accounts for roughly 27% of credit extended by scheduled commercial banks, up from 26% a year ago.
The Deputy Governor also highlighted the increasing role of digital public infrastructure, such as the Unified Lending Interface (ULI) and Account Aggregator framework, in enabling NBFCs to reduce dependence on physical collateral and extend formal credit to small and medium-sized enterprises (MSMEs) and microfinance borrowers. Murmu emphasized the necessity of robust governance and liquidity management as NBFCs expand, advocating for diversified funding sources and a more developed corporate bond market.
He stressed that securitization should progress from being merely a liquidity instrument to a method for genuine risk transfer and capital release. The Deputy Governor emphasized that rapid digitalization must not compromise customer protection, stating that conduct regulation, grievance redressal, responsible lending, and cyber resilience will continue to be crucial priorities.
Murmu concluded by reiterating that the sector's growth prospects are promising, driven by technology and trust.
Written by urgent.news from The Economic Times's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.