Home / Industry / Banking /  SBI to amp up early warning system to spot stress, fraud

MUMBAI : State Bank of India (SBI), the country’s largest lender, is looking to upgrade its early warning system to preempt loan defaults, fund diversions and frauds by tapping external data like rating revisions, stark share price movements and even media reports, showed documents reviewed by Mint.

The bank believes that, while diverting funds through related parties, trusts, and foreign subsidiaries is among the chief causes of asset quality challenges and frauds, lenders get conclusive information only after a forensic audit. It wants a data service provider to figure out ways to flag fraudulent transactions and potential defaults even before detailed audits show a public tender document.

The data provider, SBI said, has to provide alerts on breaches of a set of 199 early warning signals from external sources. These include alerts on changes in ratings or outlook; share prices hitting a 52-week-low during any particular week; promoter holding falling below 26%; decline in the number of employees on a monthly basis as per data from the Employees’ Provident Fund Organization (EPFO); negative news of raids by income tax, sales tax, central excise duty officials and imposition of penalties by government agencies.

In 2021-22, SBI reported 4,192 frauds worth 7,101 crore, setting aside an equal sum of money as provisions against it. In 2020-21, it reported 5,724 frauds of 10,086 crore, its FY22 annual report showed.

“(The) bank possesses data like Central Repository of Information on Large Credits (Crilc) and credit information company (CIC), which cannot be shared with the service providers in terms of current regulatory guidelines. How can the bank leverage this data to strengthen the early warning signal system in collaboration with the service provider?," asked SBI in one of the documents cited above.

The Crilc database is used by banks and other financial institutions to share, among themselves and with RBI, the classification status of borrowers. Concerned over the rising incidence of large borrowers attempting to defraud banks, RBI, in 2015, released a framework for the prevention and early detection of such cases. “The early detection of fraud and the necessary corrective action is important to reduce the quantum of loss which the continuance of the fraud may entail. The government is separately looking into the issue of more timely and coordinated action by the law enforcement agencies," it said in May 2015.

Ashwini Kumar Tewari, managing director in charge of risk, compliance, and stressed assets at SBI, had said in July that the bank would use analytics to track emerging stress across sectors.

“Say there are problems in the steel sector in a particular state; one way to know is to track media reports, but is our database showing impending stress from that region? After it becomes NPA, it will have to be dealt with, but if there are early indicators of stress, then this can be handled proactively," Tewari had said.

Experts said that banks could do a lot more even with internal transaction data they have access to, which other institutions like credit rating agencies cannot use. According to an analyst, not only can banks use the Crilc database, which points to defaults by borrowers with any bank, but they can also access borrower transaction data to access possible siphoning of funds.


Shayan Ghosh

Shayan Ghosh is a national writer at Mint reporting on traditional banks and shadow banks. He has over a decade of experience in financial journalism. Based in Mint’s Mumbai bureau since 2018, he tracks interest rate movements and its impact on companies and the broader economy. His interests also include the distressed debt market, especially as India’s bankruptcy law attempts recoveries of billions worth of toxic assets.
Catch all the Industry News, Banking News and Updates on Live Mint. Download The Mint News App to get Daily Market Updates.
More Less

Recommended For You

Trending Stocks

Get alerts on WhatsApp
Set Preferences My ReadsWatchlistFeedbackRedeem a Gift CardLogout