The latest Financial Stability Report of the RBI has been curiously silent about the banking frauds. However, as the previous edition of the bi-annual Financial Stability Report revealed that the Indian banking system faced 6500 instances of fraudulent transactions amounting to over INR 30,000 Crore in the last fiscal. The number of frauds is alarmingly large, and there seems to be a repetition in the pattern of frauds. Further, the majority of the frauds have been happening in the lending space.
From the bank’s perspective, approving the loan is merely half the job done. However, during the time while the borrower is servicing the loan, there can be numerous changes in his / her credit profile. Such changes may occur due to changes in the industry trends or the economic environment impacting the ability of the borrower to repay the loan. Therefore, despite the sound credit assessment of the borrower, the financial institutions still run a credit risk due to changes in the business environment.
How can the banks ensure that this credit risk is kept under control? Well, Technology has a solution!
Before a healthy person turns ill, we can see some symptoms of sickness. Diagnosing the symptoms help us reach the root cause of illness and treating/avoiding it. Similarly, before the borrower’s credit rating deteriorates, certain signs signify the borrower’s deteriorating position. Early Warning System (EWS) of banks detect such warning signals that may impact the borrower’s ability or willingness to pay back the loan. Before going forward to see how EWS + Technology can solve the credit risk problem for the bank, it is imperative to understand what an EWS is.
What is an EWS?
In 2015, RBI prescribed certain illustrative red flags or early warning signals that the banks can detect to identify frauds or treat rising NPAs. Early Warning Systems are a set of guided processes for identifying risks/signals at an early stage. A well-designed protocol of an EWS system can help the bank officials and the senior management forecast an impending event that may negatively affect the loan book of the bank. EWS can, therefore, strengthen the oversight of bank assets. As per a Mckinsey report, effective monitoring using EWS of the loan-book can reduce the loan-loss contingency by 10%-20%. The EWS has traditionally been a manual process involving monitoring on a post-facto basis, primarily based on past events and selective financial projections.
However, in current times, banking procedures and systems are digitized. Therefore, the traditional Early Warning Signals reflected in the borrower’s ledgers, stock statements, MSOD, QIS, financial statements, site inspection reports, minutes of the loan consortium meetings, et al have become redundant. The signals provided by them post facto are no longer relevant to the current times’ given crime sophistication. Therefore, with digitized banking operations, even the EWS has to be modernized.
The modernized form of EWS involves AI and ML-based EWS, which raises red flags ahead of time in the digitized banking system. Such modernized EWS has been devised to replace the traditional system that involves manual, biased, and subjective decision-making. The modernized EWS systems make use of technologies like analytics, data modeling, AI, and ML. Such a system is alert 24*7 and picks up events indicating distress signs on the borrowers and behavioral changes indicating a willingness to repay loans. The data from these events are utilized by AI/ML-based analytics to issue warning signals to enable the borrower to be classified as a Red Flagged Account. The modernized EWS is a development from a traditional compliance-driven post facto system to a preventive control-based digital system to raise early warnings.

Category A –Banks with traditional EWS
Category B – Banks with Modern EWS with AI
The above diagram is a testament to the fact that proactive controls help bank restrict their exposure to non-performing assets. The chart postulates that Category B has restricted non-performing assets in a much better manner than Category A banks.
Detailed difference between traditional and modern EWS?
Why is the modern technology-based EWS important?
1. Reduces the cost and increases the coverage
One of the major challenges for banks is the effort and time required to monitor loan-book for risks. Also, only monitoring traditional and financial data points is not relevant as it does not provide 360 degrees view of the borrower. AI-based EWS can manage volumes of data from traditional and non-traditional sources like news, social media, and alternate data sources (used for sentiment analysis). This not only reduced the cost but increases the coverage multi-folds.
2. Reducing the error rate with continuous learning
With time the ML algorithms keep getting better as more data is fed into it. As the algorithm becomes better, the warning systems get more accurate. Increased accuracy reduces the overall error rate of the system and reducing the cost. AI and ML systems can also track new patterns created by money laundering or new activities.
3. No human bias involved
The traditional early warning system involved human bias in sample selection for performing tests. Human bias or error in human judgment may cause financial losses for banks. With the AI-powered EWS system, human biases are eliminated and losses are reduced.
4. Structure unstructured data by training NLP models
Banks sit on loads of unstructured data which is useless until it is decoded. The traditional systems find it very time-consuming to structure and analyze the data. However, the ML can cleanse the data, restructure the data, and convert it into meaningful information. Natural Learning Processing uncovers meaningful insights/information from such under-used data. Early warnings signals are generated from such information to identify potential defaulters early on.
This is how technology can act as an enabler to solve credit risk and rising NPAs. Now that we have seen the features of modernized EWS, it is also important to know what AI-ML-based EWS systems are capable of doing practically.
Practical cases where EWS could have averted crises?
Banks like Punjab National Bank could have also unearthed the INR 6800 crore fraud of Nirav Modi timely had there been an AI-based EWS that could detect the scrupulous transactions timely. It is not a surprise that Punjab National Bank is looking to deploy an AI-based early warning system that will accumulate and assimilate data of the borrowers and their social media interactions.
In the case of Punjab and Maharashtra Co-operative Bank (PMC), the EWS that mine and understand credit data (which includes both transactional and external touchpoints) could have detected and flagged red flags ahead of time. The crisis faced by PMC bank would have been avoided through an EWS red-flagging the transactions with the HDIL group. Almost 73% of the total loan book of PMC worth INR 11,800 Crore given to HDIL Group turned into NPA. With the installation of an EWS, the sanctioning of a loan that constituted such a large component of the total loan book would be categorized as a red flag.
The adoption of AI-based EWS could have not just red-flagged the transactions that led to the fraud but it could have also saved the priceless goodwill of the banks before the occurrence of such transactions. There is an increased fear amongst the people to have PMC as their banker after the crisis unfolded.
Who needs the technology-based EWS the most?
Due to stronger corporate governance mechanisms, the majority of private banks in India are already using advanced EWS. Some of the public sector banks have also followed the private banks in digitizing the EWS system.
However, co-operative banks somewhere lag behind in terms of overall technology adoption.
Why do we feel it is the need of the hour for the Co-operative banks to start adopting these systems?
1. Structure of co-operative banks making them risk-prone
As per RBI’s recent report, the urban co-operative banks have reported nearly 1000 cases of fraud worth more than INR 220 crores. The sheer nature of co-operative banks is prone to risks as a group of individuals pooling capital to form a bank. They do so while availing loans from the bank they have formed.
In such arrangements, there is a higher possibility of well-connected persons forming co-operative banks and siphoning off funds. Such incidents have already led to a loss of goodwill for co-operative banks. This has caused the movement of customers from co-operative banks to large private Indian banks or SBI.
2. Rising NPAs in the co-operative banks
As per a recent statement of the RBI, the proportion of gross NPAs of Urban co-operative banks saw a rise to 10.8% as against 7.8% in the previous year. Also, the borrowers being categorized under the lowest D category with the co-operative banks increased during the last fiscal year. The rising NPAs of the Urban Co-operative banks are a major source of stress, and crises like the PMC bank can be attributed to the increasing NPAs.
Drop a comment or write us an email at info@pirimidtech.com for any feedback about this article. Do check out our portfolio in building Robo advisory, Large Scale Trading Systems, Algo Trading, Stock Sentiments, Price Trends Forecasting, Backtesting frameworks, Credit Model, Open Banking, etc. on our website. Connect with us to discover how our Fintech expertise can help you build cutting-edge solutions powered by AI/ML.
This is a guest post by Yash Surana, who is a CA and an MBA (Finance) from SPJIMR, Mumbai, and loves writing about finance, strategy, startups & Personal Finance.
References:
https://www.livemint.com/companies/news/pnb-turns-to-ai-to-curtail-loan-frauds-11603903929444.html
https://www.livemint.com/industry/banking/a-year-after-pmc-bank-scam-rbi-says-npas-rise-in-urban-co-operative-banks-11609258429826.html
https://www.entrepreneur.com/article/356245