A fictional investigative series inspired by patterns observed in publicly documented corporate governance failures.
Mumbai, 9 in the morning.
The bank branch had opened like any other day. Customers were walking in; cheques were being submitted; accounts were being checked; and employees were settling into their usual work.
In one corner, a newly-transferred officer was reviewing documents relating to a diamond-trading business. The representatives of the company were waiting confidently across the desk.
“We've received this facility before,” one of them said.
“Do we have the records?” she asked.
A junior officer searched the bank's system. Nothing came up. He checked again.
“Nothing, madam.”
“But they say the bank gave them guarantees earlier,” he said.
She looked at the papers they had brought with them. There they were — letters of undertaking (LoU), through which the bank had promised other banks that it would honour certain obligations, allowing the beneficiary to obtain credit against those assurances.
The documents existed. Foreign banks had acted on them and had extended real money. But inside the bank's own core banking system, the corresponding exposure was not properly recorded.
That was the first thing that did not make sense.
Banks use SWIFT, an international messaging network, to communicate financial instructions to other banks. SWIFT itself was not the problem. The problem was that messages sent through SWIFT were not being properly reconciled with the bank's own core banking system.
One system carried the message. The other was supposed to carry the corresponding financial record. The two were not reliably talking to each other.
That meant a guarantee could be communicated to another bank without the corresponding liability being properly reflected in the issuing bank's own records.
And once that gap existed, it could be used again.
One guarantee became another, and then another. Foreign banks continued extending credit on the strength of promises that the issuing bank's own records did not properly show.
When an earlier obligation came due, a fresh LoU was issued to close it out, and the cycle continued. Over the years, the numbers grew and so did the exposure.
A transaction appeared in one place, while the liability that should have accompanied it did not appear where it was supposed to. The reconciliation that should have brought the two together was not being carried out effectively.
A question should have been asked much earlier. It wasn't.
And when a control failure continues for long enough, people can stop treating it as a failure and start treating it as part of the normal process.
Back at the branch, the officer continued searching through the records.
“No such approval is showing here,” she told the representatives.
The room went quiet.
If the bank had not recorded the guarantee, who had issued it? If it had been issued, who had authorised it? How had the message reached a foreign bank when nothing corresponding to it appeared in the bank's own records?
Then came the question that mattered most.
“How many more are there?”
One answer led to another transaction. That led to another guarantee and then another.
What had first looked like a small discrepancy was beginning to look like a pattern that had been building quietly for years.
Eventually, the scale became clear. Unauthorised guarantees running into thousands of crores had created substantial obligations.
A liability that had not been properly reflected in the bank's own records was suddenly visible to the wider financial system because other banks had extended credit relying on those guarantees.
To them, the promises had been real.
Trust had moved money across institutions long before anyone checked whether the numbers actually matched.
When the news became public, the stock market reacted and the bank's share price fell.
That evening, a small investor opened his demat account. He had bought the shares believing they were a relatively stable holding. Now he watched the price falling on the screen.
“Is the bank going bankrupt?” his wife asked.
“Not that,” he said, and paused. “But nobody knew how much risk was sitting behind those guarantees. Not even, it seems, the bank itself.”
The shares in his account were not just a number. They represented his children's school fees, money he had put aside for retirement, and years of careful saving. What had looked like a stable investment was suddenly uncertain.
The decisions of a handful of people working inside a control system that had not been watched closely enough had now become a loss in thousands of ordinary investment accounts.
The people responsible were nowhere near the investor's desk. The consequences were.
What This Pattern Teaches
A bank's size does not tell you how strong its controls are. Its reputation does not tell you whether its systems actually reconcile with each other. And its reported profit does not tell you how much contingent exposure may be sitting outside the obvious loan book.
No retail investor can see a bank's internal SWIFT messages or know which employee authorised a particular transaction. But there are still questions worth asking before investing.
→ How large are the bank's guarantees and contingent liabilities?
→ How significant are its off-balance-sheet exposures?
→ Do different parts of its financial disclosures reconcile with each other?
→ What do its auditors say about internal controls and material weaknesses?
→ Are risks concentrated in a particular business line, borrower, branch or relationship?
→ When two of the bank's own disclosures don't quite reconcile, is that really just a technical difference, or is it something worth questioning?
Here is the uncomfortable part: an investor can read every published number correctly and still miss a risk that the bank's own systems failed to capture properly.
That is why governance cannot end with accurate reporting. It has to begin earlier, with the systems that generate the reporting in the first place.
A control matters only when it works in practice. A reconciliation matters only when someone is responsible for performing it and following up when the numbers do not agree. No system should depend entirely on the integrity of one person. That is precisely why segregation of duties exists.
Two systems can each contain information and appear perfectly normal when examined separately. The organisation can still be operating blind if nobody is responsible for comparing what one system knows with what the other knows.
A guarantee may remain outside the numbers investors normally see. A liability may remain hidden in the gap between two systems.
But when the obligation finally comes due, the financial consequences cannot remain hidden.
Case Status: Closed.
Evidence fades. Patterns don't.
(SVVN Babji Rao is a retired deputy commissioner (prohibition and excise), Govt of Andhra Pradesh, with 39 years of experience in investigation, enforcement and administration. A law graduate and corporate governance practitioner, he writes on governance, institutional behaviour and investor protection. His Case Diary series examines the warning patterns behind corporate failures.)
