LOAN FRAUDS AND BAD BOY BILLIONAIRES: A NEW APPROACH OF LOAN FRAUD PREVENTION USING NATURAL LANGUAGE PROCESSING (NLP)

Paper Number :WP/04/2021
Publication Date :Nov. 1, 2021

Credit frauds, where a loan turns into bad debt on account of fraudulent activities damage the core business of the banking industry and dent its reputation. While transactionsbased activities like credit card frauds or cybercrimes are captured at a point in time, credit frauds perpetuate over time. We present a methodology to analyse frauds happening over time through NLP tools. The methodology can use linguistic information on customers as well as documents to identify causes of frauds. We use the methodology to analyse 653 known cases on fraud from India, which has seen a growing number of credit frauds, to identify the Early Warning Signals. We develop a ranking of the EWS and further use an ordered logit model to analyse the most important EWS impacting high value frauds.