Pedestrians walk past a Punjab National Bank (PNB) branch in Mumbai, India. In February 2018, the bank accused jeweler Nirav Modi of involvement in a multi-billion dollar Letters of Understanding fraud worth Rs 14,000 crore. Photographer: Dhiraj Singh/Bloomberg

Patchwork Approach To ‘Early Warning Signals’ May Do Little To Prevent Bank Frauds

The persistent rise in bank frauds has had Indian authorities worried for some time now. At last count, in 2017-18, banks lost over Rs 41,000 crore to frauds, shows data from the Reserve Bank of India. Ninety percent of these frauds took place at public sector banks.

To better prepare banks to deal with the rising instances of fraud, the RBI, in 2015, had prescribed a framework for putting in place ‘early warning signals’. As a start, the regulator prescribed a list of 45 early warning signals that banks need to map by putting in place the requisite technology and systems.

“Banks tend to report an account as fraud only when they exhaust the chances of further recovery,” the RBI had noted in its 2015 circular while going on to prescribe ways in which fraud detection and prevention can be speeded up.

Where Things Stand

Three years later, it looks like most banks are taking a patchwork approach to improving their readiness in dealing with frauds.

A few have put out tenders for implementing new solutions while others have chosen to upgrade existing systems, said consultants and bankers that BloombergQuint spoke to. This is particularly true for public sector banks, where the largest number of frauds have been reported.

Of the 21 public sector banks, only six have issued Request for Proposals (RFPs) between 2015 and 2018 to implement a new system for early warning signals and red flags, according to the websites of these banks which are required to display any tenders put out. Emails sent to each of the banks seeking an update on the implementation of early warning signals did not elicit responses.

However, what many banks seem to be doing is upgrading existing systems.

“In order to develop a solution for early warning signals, some banks have opened tenders while others have asked their existing vendors to rope in a solution within the current technology platform,” said Dhruv Phophalia, managing director at Alvarez and Marsal.

The regulator has not said that there needs to be an entirely new system, explained K V Karthik, partner at Deloitte India. As such banks are creating a technology module “based on multiple existing systems” or they “look at implementing a comprehensive Fraud Risk Management solution incorporating this framework and early warning signals needed to generate this information,” he said.

The list of signals suggested by the RBI include: delay observed in payment of outstanding dues, frequent invocation of bank guarantees, bouncing of high value cheques, concealment of certain vital documents, frequent ad hoc sanctions, high value payments to unrelated parties or heavy cash withdrawals, among others.

According to a senior banker, who spoke on condition of anonymity, the list provided by RBI is an indicative one. Banks are free to expand this list. There was also no specific timeline over which this system was to be implemented, this banker said. Some of the early warning signals have been built in to existing systems while others are being added via new modules, he told BloombergQuint.

Once an early warning signal is triggered, the account is then alerted or ‘red-flagged’ to the concerned credit monitoring department at the bank, the banker said.

Implementing a system of early warning signals depends on the digital maturity of an organisation. The key lies not in just integrating the systems but to generate the early warning signal based on the triggers defined and conveying it to the concerned person and monitoring the action that has been taken.
Mrityunjay Mahapatra, Chief Executive Officer, Syndicate Bank

The second two aspects are more important than just integrating software to throw up the signals, he said.

Beyond Technology

Most agree that no IT system alone can reduce the instances of fraud. For that, awareness and response-time across the banks needs to change.

Take, for instance, the number of alerts that banks would get from any early warning system. These alerts would be large in number and any bank would need to act promptly to weed out transactions that may be genuine and others which may throw-up red flags. To do this, a bank needs access to a lot more background data, said KV Karthik.

The difficulty presently is that there could be many alerts generated based on the 45 early warning signals and to map these scenarios there needs to be a lot of data available. The number of false positives the early warning system might throw up, result in the bank needing to investigate a significant number of cases.
KV Kartik, Partner, Deloitte India

Another issue that banks face in effective fraud management, particularly in the case of loan frauds, is that of inadequate data sharing between banks. Most companies deal with multiple banks and unless information is seamlessly available, banks would find it difficult to spot any red flags at the right stage.

To be sure, the establishment of the Central Fraud Registry or Central Repository of Information on Large Credits would help in making information available across the system.

Also, where banks have given loans in a consortium, some information sharing is mandated, said Mahapatra of Syndicate Bank. But there are always loopholes. Companies will maintain a good track-record with larger banks who have more efficient early warning systems, but their accounts with smaller lenders will start to show deterioration. “When the borrower is in trouble, it usually manifests itself in smaller remits,” he said.

Phophalia noted that a truly efficient early warning system needs to bring together technology and the underlying institutional knowledge of bankers. “Even at a branch level, bank managers generally have a good assessment of the risk accounts, but that intelligence is not necessarily collated in an early warnings signal system,” said Phophalia. Artificial-intelligence and big-data led technology solution can only be a “25-40 percent contributor to an early warning signals program, he added.