The RBI's reputation, set against most of its emerging market peers, is deservedly strong. It kept India out of a full currency crisis after 1991, it has largely resisted the political capture that has wrecked central banks elsewhere, and its inflation targeting framework has held up reasonably well since 2016. None of that makes the record clean. Walk through the actual episodes in order, from 1991 to 2020, and a specific, repeatable pattern shows up: nine separate incidents, three of them inside 2018 alone, and in almost every case the RBI saw the problem coming and still did not move until the damage was done.
Chapter 1How close did India actually come to default in 1991?
Closer than most retellings admit. By June 1991, India's foreign exchange reserves had fallen to somewhere between $1 and 1.2 billion, barely enough to cover two to three weeks of imports.
This did not happen overnight. It was the end point of years of the RBI and the government tolerating a widening current account deficit and heavy government borrowing through the 1980s, on top of an exchange rate allowed to drift out of line with reality. The final trigger was external. The Gulf War spiked oil prices just as remittances from Indian workers in the region dried up. But the vulnerability that made that shock nearly fatal was built at home over the preceding decade.
The response, when it finally came, was drastic. The RBI airlifted 47 tonnes of gold to the Bank of England and pledged 20 tonnes more to the Union Bank of Switzerland, raising roughly $600 million in emergency funds, all in secrecy in the middle of a general election. The RBI's own later assessment was blunt: the gold pledge eased immediate pressure but was, in its words, not sufficient to completely absolve the country of the crisis. The pledge itself was a rational emergency move given the alternative of defaulting on external payments. The mistake was everything that let the country arrive at a two week reserve cushion in the first place.
Chapter 2How did a stockbroker exploit gaps in bank regulation so easily in 1992?
Barely a year after the near default, a completely different kind of failure surfaced, this time inside the plumbing of the banking system itself.
Harshad Mehta, a Mumbai stockbroker, discovered that the interbank market for government securities was almost entirely paper based and effectively unsupervised in real time. Banks lent each other money against securities through instruments called Bank Receipts. Mehta used fabricated receipts to draw money out of the banking system, funnelled it into stocks, inflated share prices, then sold into the rally he had personally created. The amount diverted is estimated at well over Rs 1,000 crore, with some accounts of the fully unwound scam running towards Rs 5,000 crore, close to a billion US dollars at the time.
The scam was caught because the State Bank of India happened to notice that the collateral behind one of Mehta's transactions was fake. The RBI's own interbank monitoring flagged nothing. A court reviewing the case afterwards, in CBI versus Harshad Mehta, explicitly noted the RBI's inability to intervene and regulate the securities market in real time, because the tools did not yet exist. The scam led directly to SEBI's conversion from a toothless advisory body into a statutory regulator with real enforcement powers in 1992, which is itself the clearest evidence that the earlier architecture was inadequate. India's financial markets ran on manual, low transparency infrastructure that a single motivated actor exploited for the better part of two years before anyone in authority noticed.
Chapter 3Why did it take a full decade to admit the banking system had a bad loan problem?
Move to the 2010s and the pattern repeats at a much larger scale, hidden in plain sight inside bank balance sheets rather than in a stock market bubble.
After the 2008 global financial crisis, Indian banks were encouraged, implicitly and explicitly, to restructure loans to stressed companies rather than classify them as bad. The reasoning: give companies a bridge through a temporary global shock. Instead it became the default operating mode for years. Banks kept extending and restructuring loans to companies with no realistic path to repayment, a practice known as zombie lending, which keeps unproductive borrowers alive on subsidised credit while starving healthier companies of capital.
The RBI did not force a real reckoning until 2015, when its Asset Quality Review withdrew regulatory forbearance and made banks inspect their loan books honestly. What it found was ugly.
Close to 90% of these non performing assets sat in the weakest corners of the banking system. Delaying recognition for seven years meant the eventual cleanup, through the Insolvency and Bankruptcy Code and repeated public sector bank recapitalisations, cost the exchequer tens of billions of dollars. Confronted around 2010 instead of 2015, the bill could have been a fraction of that. The 2008 forbearance itself was not unusual; most central banks did something similar in that crisis. Extending an emergency measure into permanent policy for seven years is what turned a manageable problem into a systemic one.
Chapter 4Did demonetisation actually achieve what it set out to do?
Demonetisation in November 2016 was announced by the government, but the RBI had to execute it, print the replacement currency, and defend it publicly for years afterwards. That makes it fair to include here regardless of where the decision originated.
The stated goal was to flush out black money: cash hoarded outside the formal banking system and therefore untaxed. The mechanism was banning the two largest denomination notes overnight and requiring everyone to deposit old notes by a deadline, on the theory that hoarders would be too afraid to deposit illegal cash and would be left holding worthless paper.
That is not what happened. The RBI's own 2017-18 annual report revealed that 99.3% of the roughly Rs 15.44 trillion in demonetised notes came back into the banking system. Only about Rs 0.13 trillion, less than 1%, never returned.
A return rate this high means one of two things. Either most cash in circulation was legitimate savings rather than hoarded black money, or a meaningful share of illegal cash laundered itself back into the formal system before the deadline, defeating the entire premise. Real secondary benefits followed: an expanded tax filer base and a durable shift towards digital payments. But those were not the stated goal. Hitting a side benefit while missing the primary target, and disrupting cash dependent small businesses for months along the way, is not the same as the policy working as designed.
Chapter 5How did two bank employees run a Rs 14,000 crore fraud for seven straight years?
In February 2018, Punjab National Bank disclosed it had been defrauded of roughly Rs 14,357 crore, close to $2 billion at the time and still one of the largest banking frauds in Indian history. The mechanism exposes a specific, fixable, long ignored gap in bank technology.
Two employees at PNB's Brady House branch in Mumbai issued fraudulent Letters of Undertaking, effectively bank guarantees, to jeweller Nirav Modi's firms. They used the SWIFT international messaging system to instruct overseas branches of other Indian banks to extend credit against these guarantees. The fraud worked because PNB's SWIFT system was never integrated with its core banking system, so the transactions never appeared in the bank's internal books at all. The arrangement ran for roughly seven years, from 2011 to early 2018. It was caught by accident, when one of the two employees retired and his replacement, unfamiliar with the informal arrangement, asked Nirav Modi's firm for collateral it had never been required to post.
The CBI went further than blaming two rogue employees. It named PNB's former CEO and two executive directors, alleging they had failed to implement RBI circulars and cautionary notices specifically about reconciling SWIFT with core banking systems, notices that predated the scandal. Afterwards, the RBI discovered the same SWIFT gap at several other banks. The vulnerability behind a $2 billion fraud had been sitting unaddressed across a chunk of the banking system, not one branch. The RBI's fix, banning Letters of Undertaking for trade credit and mandating SWIFT integration, was the right one. It arrived after seven years and Rs 14,000 crore, not before.
Chapter 6Did the RBI actually see IL&FS coming, and did it say so?
This episode needs no hindsight reconstruction, because the RBI's own published research called the shot in writing, two weeks before it happened.
IL&FS, a large infrastructure financing conglomerate registered with the RBI as a systemically important Core Investment Company, had built a debt pile of roughly Rs 91,000 crore against a parent company equity base of just Rs 9.83 crore, leverage sustained by a complex web of subsidiaries and inter group guarantees. In September 2018, the RBI published its 18th Financial Stability Report. Buried in its network analysis section was an explicit warning: non banking financial companies had grown so interconnected with the banking system that the failure of the largest NBFC could trigger contagion losses comparable to the failure of a major bank.
Two weeks after that report was published, IL&FS began defaulting. The liquidity crunch spread through the entire NBFC sector, froze mutual fund redemptions, and tightened credit conditions across the real economy for the following two years. The RBI had flagged the exact mechanism, NBFC interconnectedness as contagion risk, in a public document weeks before it materialised, and the system, including the RBI's own supervisory arm, did not act on the warning in time. Publishing an accurate warning and then not acting on it is arguably worse than missing the risk altogether. The analytical capacity existed. The response mechanism did not.
Chapter 7How did a bank hide a loan fraud for years right under the RBI's nose?
Punjab and Maharashtra Cooperative Bank collapsed in September 2019, a year after IL&FS. Its failure exposed a structural blind spot in how India regulates cooperative banks that had existed for decades.
PMC Bank's management colluded with the promoters of real estate company HDIL to hide the fact that over Rs 2,500 crore, more than 70% of the bank's entire loan book, was concentrated in loans to a single defaulting borrower. This is not a subtle accounting trick. It is a level of single borrower concentration that violates basic regulatory norms so obviously that the real question is how it went undetected for years.
The answer is structural. Cooperative banks sat under a dual regulatory framework, supervised partly by the RBI and partly by state Registrars of Cooperative Societies, a split that diluted accountability on both sides. The RBI only had authority to inspect most cooperative banks' books once a year, nowhere near the continuous supervision applied to scheduled commercial banks, even though cooperative banks held deposits from millions of small and often financially unsophisticated savers. When the fraud surfaced, the RBI capped withdrawals at Rs 1,000 per account, later raised to Rs 40,000, triggering panic among depositors who had no idea their savings had been sitting inside a bank running on doctored books. The RBI's response afterwards, pushing the Banking Regulation Amendment Bill in 2020 to bring cooperative banks under commercial bank audit and governance standards, is essentially an admission that the supervisory architecture had been inadequate all along.
Chapter 8Why did it take a liquidity run for the RBI to act on YES Bank?
YES Bank's collapse in March 2020 followed the PMC pattern in one important respect: the warning signs were visible for years before the RBI moved decisively.
YES Bank had spent the mid 2010s aggressively expanding its loan book with high risk loans, particularly to real estate, under governance that multiple independent directors publicly flagged as inadequate before resigning in protest. The stock, around Rs 400 in early 2018, had collapsed to around Rs 50 by late 2019, a signal visible to anyone watching, let alone the regulator. Deposits fell from Rs 2.1 trillion to Rs 1.7 trillion between September and December 2019 alone, a nearly 19% outflow in three months. A bank run, playing out in slow motion, in full public view.
The RBI eventually superseded the board and engineered a rescue built on a Rs 10,000 crore capital infusion led by State Bank of India. The rescue was executed competently and prevented a full depositor panic. But competent crisis management after the fact does not erase years of visible governance deterioration, during which supervisory intervention clearly was not enough to change the bank's trajectory before depositors started running on their own.
Chapter 9Why did the Supreme Court have to strike down the RBI's crypto ban?
This one differs in character from the rest. It is not a story of the RBI missing a problem. It is a story of the RBI overreaching and being overruled by the judiciary, a distinct kind of institutional failure.
In April 2018, the RBI issued a circular ordering every bank and payment processor it regulated to cut off all services to cryptocurrency exchanges and traders, citing consumer protection, money laundering and market integrity concerns. The effect was severe. Trading volumes on some Indian crypto platforms fell by more than 99%, an outright ban implemented through the banking system rather than through legislation, since the RBI has no direct statutory power to ban a legal asset class.
The Internet and Mobile Association of India challenged the circular, and in March 2020 a three judge Supreme Court bench unanimously struck it down. The Court found the RBI could not point to any actual damage suffered by regulated banks from crypto activity, that a blanket ban was disproportionate to whatever risk existed, and that the RBI had not considered less intrusive alternatives, such as restricting only anonymous transactions, before reaching for total prohibition. The consequences ran beyond the crypto industry. For two years, a legal, taxable, employment generating sector operated under an existential threat imposed by regulatory circular rather than law, only for that circular to be declared unconstitutional. The RBI has since discouraged bank involvement with crypto through advisories rather than binding circulars, a tacit acknowledgment that the direct approach failed a basic legal test the first time it was tried.
Chapter 10What is the pattern across all nine of these?
Line up 1991, 1992, the 2008 to 2015 forbearance years, 2016, the PNB scam, IL&FS, PMC Bank, YES Bank and the crypto ban, and a single mechanism repeats far more often than bad luck would predict.
In seven of the nine episodes, the failure was not a wrong judgment call made in the moment. It was a visible, sometimes self documented problem that sat unaddressed for years before it forced a response. The 1991 crisis was years of ignored current account deterioration. The 1992 scam ran for roughly two years on unsupervised interbank paper. The NPA crisis was seven years of extended forbearance instead of early recognition. The PNB fraud ran for seven years through a technology gap the RBI had already circulated warnings about. IL&FS is the starkest case: a public RBI document naming the exact risk two weeks before it happened. PMC Bank was a decade of light touch supervision on cooperative banks. YES Bank was a stock price and deposit base collapsing for over a year before the board was superseded. Only the crypto ban breaks the pattern, and it breaks it in the opposite direction: moving too fast and too bluntly, without building a legal case that could survive a court challenge.
The honest conclusion is that the RBI's institutional strength lies disproportionately in crisis response rather than crisis prevention. Every episode here ended with real, often effective action: a gold pledge, a statutory SEBI, an Asset Quality Review, board supersessions, capital infusions, a cooperative banking overhaul. What is far harder to find across eight decades of this record is the RBI acting early enough that the crisis response was never needed at all.