FinTech is a term which refers to a set of businesses which are leveraging advanced technology as a core ingredient in their business model to drive rapid evolution and transformation of financial services products and delivery. Machine Learning and AI on the other hand, happens to be at the frontier of today’s tech evolution landscape. Therefore, FinTechs are, not surprisingly, the most prolific users of ML and AI as a key ingredient in their strategy.
The universe of fintech consists of many different types of financial service solutions and platform providers to enable those services. Some of the most prominent ones include lending solutions, payment solutions, insurance solutions (often classified under InsureTech), wealth/asset management solutions, and so on.
FinTech businesses make use of AI/ML solutions across most, if not all, of their critical processes and functions. For example, for many fintech lenders, advanced ML models for credit scoring has become an integral part of their underwriting process – this enables highly efficient, fast application processing and leads to improved portfolio quality as powerful ML models predict delinquency risk quite accurately.
The advent of AI/ML has been extremely beneficial for the fintech players operating in the MSME lending space (more accurately, micro/small unorganized business lending space), especially the strata near the bottom of the pyramid, which have limited or no access to formal credit form banks and other lenders primarily active in the organized sectors.
The MSME sector has a rather wide range in terms of business sizes and degree of organization – at one end of the spectrum the businesses are large and formalized enough to be within the ambit of GST and have several years of detailed banking records, tax return filing information, etc. and on the other end, the businesses are much smaller and have no or inadequate credit history and banking records and rarely any tax filing or GST records. But the conventional credit evaluation and underwriting process followed by the banks and other similar mainstream lenders rely heavily of sufficient credit history, banking and accounting records, tax return filing information for several years, etc. The micro/small businesses lacking those data, therefore, cannot usually access credit from these institutions.
The fintech players intending to lend in this space addresses the challenge of inadequate credit history by devising their own innovative methods of creditworthiness evaluation. While there are certain heuristic approaches used for this purpose, appropriate AI/ML models are the most powerful tool in this aspect. Custom-built ML models deviate from the traditional data requirement constraints and enable the lenders to predict probability of delinquency/default using various kinds of alternative data quite accurately and hence evaluate credit risk and underwrite accordingly. For example, some fintech lenders lending to these micro and small businesses use localized sectoral economic trends, business image data (e.g., stock of goods, store space, store frontage and location, etc.), permitted mobile scrape data (transactions SMS data, for example), along with any limited banking data available, informal accounting data from mobile apps, etc. to build such AI and machine learning models. In cases where the business owners have some credit bureau data, albeit thin files with very little history, those are also incorporated in computing their risk scores by the models built for this specific purpose and market segment. Thus, fintech lenders are leveraging AI/ML in expanding financial inclusion to underserved sections of the MSME sector while maintaining a strong portfolio performance through smart, quantitatively informed underwriting.
These fintech firms which use AI/ML in their business processes, apply the tool in functions other than underwriting as well. For instance, they often use custom models to manage EMI bounce, optimize collection efforts, target cross-sell and upsell of products, etc. Geolocation-based analysis for acquisition and/or collection planning and optimization, multilingual AI chatbots for customer support, etc. are also some of the common practices in this respect.
In conclusion, AI/ML is proving to be a valuable tool in helping Fintechs not only to drive efficiency and profitability, but also to fulfill a philanthropic duty of enabling financial inclusion of a sector that has suffered decades of discrimination and exclusion from formal lending channels in the country.
Views expressed above are the author’s own.