A bank can die in two different ways. It can run out of cash: depositors want their money today, and the bank's money is tied up in loans and bonds that cannot be sold quickly at a fair price. Or it can run out of capital: its assets are simply worth less than what it owes. The first is a liquidity failure, the second a solvency failure, and the cure for each is different. A central bank can lend cash to an illiquid bank; it cannot lend a bank back into solvency. In real crises the two problems feed each other, often within hours. This part explains the mechanics with balance sheets, the economics of bank runs that won the 2022 Nobel Prize, how failures spread, the cases examiners ask about (from Global Trust Bank in 2004 to Silicon Valley Bank in 2023), and the tools India and other countries use to contain them, including the withdrawn FRDI Bill and the bail-in debate that the Supreme Court has just reopened.
The crisis episodes of 2018–2020 (IL&FS, PMC Bank, Yes Bank, Lakshmi Vilas Bank) and the RBI's financial-stability toolkit are covered in Part 15, and capital rules in Part 27. Liquidity ratios are in Part 28 and deposit insurance in Part 30. This part links to them rather than repeating them.
In the news
- 15 Oct 2026: Supreme Court upholds quashing of Yes Bank's AT1 write-off. The Court upheld the Bombay High Court's January 2023 order that set aside the March 2020 write-off of ₹8,415 crore of Yes Bank's Additional Tier 1 bonds; Yes Bank's shares fell 14.5% that day. Why it matters for the exam: it limits how far a bank rescue can impose losses on bondholders (a bail-in) without clear legal authority.
- 25 Sep 2026: Nano Banc fails in California. State regulators closed the Irvine bank (assets $736 million in June) and the FDIC sold its deposits to Sunwest Bank; the FDIC estimated the cost to its Deposit Insurance Fund at about $114 million. Why it matters for the exam: small US failures are routine and handled by purchase and assumption, with insured depositors losing nothing.
- 1 Apr 2026: higher LCR run-off rates for digital deposits take effect. Under rules announced on 21 April 2025, banks must assume an extra 2.5% run-off on retail and small-business deposits that can be moved by internet and mobile banking. Why it matters for the exam: it is India's regulatory answer to "digital bank runs" of the SVB kind.
- 1 Apr 2026: risk-based deposit insurance premium begins. The DICGC card rate stays 12 paise per ₹100 of assessable deposits, but sounder banks pay less; cover remains ₹5 lakh per depositor per bank. Why it matters for the exam: risk-based pricing reduces the moral hazard that flat-rate insurance creates.
- 13 Feb 2026: RBI cancels licence of Indian Mercantile Co-operative Bank, Lucknow. The bank had "inadequate earning prospects" and failed to meet Banking Regulation Act requirements; about 98.75% of its depositors are entitled to their full deposits from DICGC. Why it matters for the exam: small co-operative bank failures are the most common Indian bank failures, and deposit insurance pays out up to ₹5 lakh.
- 14 Oct 2025: Swiss court rules Credit Suisse AT1 write-down unlawful. The Federal Administrative Court, in a partial decision, found FINMA's March 2023 order writing off about CHF 16 billion of AT1 bonds lacked a sufficient legal basis; FINMA and UBS are appealing. Why it matters for the exam: together with the Yes Bank ruling, it shows that bail-ins must rest on clear law and contract terms.
Liquidity vs solvency
Two different tests
A bank's balance sheet has assets (cash, government securities, loans) on one side and liabilities (deposits, borrowings) plus equity capital on the other. Equity is what is left for shareholders after all liabilities are paid: equity = assets − liabilities.
- Solvency asks: are the assets worth more than the liabilities? If equity is positive (and above the regulatory minimum), the bank is solvent. Losses on loans and bonds reduce equity one-for-one.
- Liquidity asks: can the bank pay what it owes as it falls due? It needs cash, or assets it can turn into cash quickly without a big loss, at least equal to the withdrawals and repayments it faces.
A solvent bank can be illiquid if depositors all ask for money at once. An insolvent bank can stay liquid for years if depositors keep rolling over their money, which is how hidden losses at PMC Bank grew until 2019. Regulation therefore has two separate pillars: capital rules (CRAR, Basel III; see Part 27) for solvency, and liquidity rules (CRR and SLR in Part 5; LCR and NSFR in Part 28) for cash.
Why the line blurs in a crisis
In theory the central bank should lend to the illiquid-but-solvent bank and let the insolvent one be resolved. In practice the two collapse into one another:
- An illiquid bank must sell assets fast, at fire-sale prices. The discount is a real loss, which eats capital. Liquidity trouble becomes solvency trouble.
- An insolvent-looking bank loses depositors' trust, so they withdraw. Solvency trouble becomes liquidity trouble.
- In the middle of a run nobody, including the regulator, knows the true value of the loan book. A central bank deciding over a weekend whether to lend has to guess.
Worked example: how losses eat capital
Take a simple bank with assets of ₹100 crore:
| Assets | ₹ crore | Liabilities and equity | ₹ crore |
|---|---|---|---|
| Cash and reserves | 5 | Deposits | 88 |
| Government bonds | 25 | Borrowings | 4 |
| Loans | 70 | Equity | 8 |
| Total | 100 | Total | 100 |
Its leverage (equity ÷ assets) is 8%. Now two shocks hit.
Shock 1, credit loss. 10% of loans turn bad and the bank expects to recover only 40% of them.
- Bad loans = 10% × 70 = ₹7 crore
- Loss = 60% × 7 = ₹4.2 crore
- Equity falls from 8 to 8 − 4.2 = ₹3.8 crore
Shock 2, interest-rate loss. Market yields rise by 2 percentage points. The bonds have a modified duration of 6, so their price falls by about 6 × 2% = 12%.
- Bond loss = 12% × 25 = ₹3 crore
- Equity falls from 3.8 to ₹0.8 crore
Total assets are now 100 − 4.2 − 3 = ₹92.8 crore, so equity is 0.8 ÷ 92.8 = 0.86% of assets. Deposits and borrowings (₹92 crore) are unchanged: losses fall on equity first. One more loss of ₹1 crore and the bank's assets are worth less than what it owes; it is insolvent. Notice that the bank has not run short of cash at any point. It could keep paying depositors and look normal from the outside.
Now add a run. Suppose 20% of deposits (₹17.6 crore) are withdrawn. The bank has ₹5 crore of cash, so it must raise ₹12.6 crore by selling bonds now worth ₹22 crore. If panic in the bond market means it can sell only at a further 5% discount, it must sell bonds with a book value of 12.6 ÷ 0.95 = ₹13.26 crore to raise ₹12.6 crore, booking a further loss of about ₹0.66 crore. Equity falls to about ₹0.14 crore. The run has almost finished what the losses began. This is exactly the mechanism that killed SVB.
Maturity transformation: the reason banks are fragile
Banks borrow short and lend long. Savings and current account deposits can be withdrawn on demand; even fixed deposits can be broken early. Home loans run for 20 years and infrastructure loans for 15. This maturity transformation is the bank's job: it lets households keep money available while the economy gets long-term finance. It is also why banks earn a net interest margin (Part 24) and why they create money (Part 23).
The cost is a permanent mismatch. A bank never holds cash for more than a small fraction of its deposits. That works because withdrawals in normal times are predictable and largely offset by new deposits. It stops working the moment depositors expect other depositors to withdraw. Liquidity rules exist to stretch the time a bank can survive such a moment: the LCR asks for enough high-quality liquid assets to cover 30 days of stressed outflows; the NSFR asks for stable funding behind long-term assets.
Two variations on the same mismatch caused famous failures:
- Funding-market mismatch. Northern Rock (2007) funded fast-growing mortgages heavily from wholesale markets and securitisation rather than retail deposits. When those markets froze in August 2007, it could not roll over its funding.
- Interest-rate mismatch. SVB (2023) put short-term deposits into long-dated bonds bought when rates were near zero. When rates rose, the bonds lost value and deposits started leaving at the same time.
Bank runs: the Diamond–Dybvig model
The 2022 Nobel
On 10 October 2022 the Nobel Memorial Prize in Economic Sciences went to Ben Bernanke, Douglas Diamond and Philip Dybvig "for research on banks and financial crises". Bernanke's 1983 work showed that bank failures were a major reason the Great Depression was so deep and long. Diamond and Dybvig's 1983 model explained why banks are useful and why that same usefulness makes them prone to runs, and how deposit insurance and a lender of last resort can prevent runs.
The model in one example
Depositors do not know when they will need their money. A bank pools them, keeps some cash for those who need it early, and invests the rest in a long-term project that pays well only if left alone. The model has two equilibria, a good one and a run.
Set-up. 100 depositors each deposit ₹1. The bank promises ₹1 to anyone who withdraws at date 1. It keeps ₹20 in cash and puts ₹80 in a project that pays ₹1.5 per rupee at date 2 but only ₹0.5 per rupee if cut short at date 1.
Good equilibrium. Only the 20 depositors who really need money withdraw at date 1. Cash covers them exactly. The other 80 wait and share 80 × 1.5 = ₹120, or ₹1.50 each. Everyone is better off than holding cash.
Run equilibrium. Suppose everyone fears a run and all 100 come at date 1. The bank has 20 cash plus 80 × 0.5 = 40 from liquidating the project, ₹60 in total. Paying ₹1 each, it can serve only the first 60 in the queue (the "sequential service" rule). The last 40 get nothing.
So a patient depositor who expects others to run is right to run too: waiting gets him zero, running gets him ₹1 if he is early enough. The run is self-fulfilling. Nothing about the project changed; only beliefs did. A perfectly solvent bank (worth ₹1.50 per depositor if left alone) fails.
How the model's cures work
- Deposit insurance: if the government guarantees the ₹1, the patient depositor has no reason to run, so the run equilibrium disappears, usually at no cost, because the guarantee is never used.
- Lender of last resort: if the central bank lends against the project, the bank need not liquidate it at 0.5, so the "only 60 get paid" arithmetic never arises.
- Suspension of convertibility: if the bank can stop withdrawals once the 20 genuine needs are met, there is no point in running. A moratorium with withdrawal caps, as India imposed on Yes Bank and Lakshmi Vilas Bank, is a modern version, though it hurts depositors who genuinely need cash.
Digital-era runs
The Diamond–Dybvig run used to take days of queues outside branches. Northern Rock's run in September 2007 was famous for its pictures of queues. In 2023 it took hours and happened on phones.
- On 8 March 2023 SVB announced it had sold $21 billion of available-for-sale securities at a $1.8 billion after-tax loss and would try to raise $2.25 billion of capital.
- On 9 March depositors initiated withdrawals of $42 billion in a single day, according to the California regulator's closing order (the Federal Reserve's review says "over $40 billion"). The bank ended the day with a negative cash balance of about $958 million.
- Management expected over $100 billion more to leave on 10 March. Together, the actual and expected outflows were roughly 85% of its deposit base. The California regulator closed SVB on the morning of 10 March and appointed the FDIC as receiver.
For comparison, the Fed's review notes that Wachovia lost about $10 billion over 8 days in 2008 and Washington Mutual $19 billion over 16 days. Three features made SVB's run so fast: about 94% of its deposits were above the $250,000 insurance limit; its depositors were a close network of start-ups and venture capital investors who talked to each other, often on social media; and mobile banking let them move money in minutes.
Contagion and interconnectedness
One failure becomes a crisis when it spreads. Four channels matter:
- Direct exposures. Banks lend to each other, hold each other's bonds and trade derivatives. A failed bank's creditors take losses. Lehman Brothers' bankruptcy on 15 September 2008, with more than $600 billion of assets, still the largest in US history, hit thousands of counterparties at once.
- Fire-sale prices. When one bank dumps assets, prices fall for every bank holding the same assets, cutting their capital too.
- Information contagion. Depositors ask, "If that bank was hiding losses, which other bank is?" After SVB, deposits fled other mid-sized US banks with many uninsured depositors. Signature Bank was closed on 12 March 2023 and First Republic on 1 May 2023.
- Funding-market freezes. Lenders stop lending to an entire class of borrowers. After Lehman, the Reserve Primary Fund "broke the buck" on 16 September 2008 because of its Lehman paper, setting off a run on money market funds. In India, IL&FS's default in 2018 froze the commercial paper market for NBFCs (Part 15, Part 31).
Interconnectedness is why regulators treat some banks as systemically important. In India, SBI, HDFC Bank and ICICI Bank are D-SIBs that hold extra CET1 capital (Part 15). The flip side is the too-big-to-fail problem: if creditors expect a rescue, they lend cheaply and the bank takes more risk. On 16 September 2008, the day after letting Lehman fail, the Federal Reserve authorised a loan of up to $85 billion to the insurer AIG, because its failure would have hit too many counterparties.
Global cases
Northern Rock, 2007: the first modern British run
Northern Rock, a fast-growing British mortgage lender, relied on wholesale markets for much of its funding. When those markets seized up in August 2007, it asked the Bank of England for help. News of the Bank of England's emergency liquidity support came out on 14 September 2007, and queues formed at its branches that day; about £1 billion was withdrawn on the first day. The run stopped only when the Chancellor announced on 17 September that the government would guarantee all existing deposits. Northern Rock was taken into public ownership on 22 February 2008. Lessons: the UK's deposit insurance then had co-insurance (depositors bore part of the loss), which gave them every reason to run; and announcing liquidity support can itself trigger a run.
Lehman Brothers, 2008: when the system is the problem
Lehman, a US investment bank, financed long-term mortgage assets with very short-term borrowing, much of it overnight repo. As mortgage losses mounted, lenders demanded more collateral and then stopped lending. No buyer was found, the US authorities declined to rescue it, and it filed for bankruptcy on 15 September 2008. Lehman was both illiquid and, on most estimates, insolvent. Its failure froze money markets worldwide and led to the AIG rescue, the money-market fund guarantee and, in October 2008, the US Congress's bank rescue fund. Lesson: there was no legal way to wind down a large, complex financial firm in an orderly way. Building one, the "resolution regime", became the central post-crisis reform.
SVB, Signature and Credit Suisse, 2023
- SVB (assets about $209 billion) grew from $71 billion to over $211 billion between 2019 and 2021 as tech deposits poured in, and invested much of the money in long-dated securities. By the end of 2022 its held-to-maturity bonds showed unrealised losses of about $15 billion, close to its total equity of about $16 billion. Because these losses were not counted in its regulatory capital, the bank looked solvent on paper. The run forced it to sell, and the losses became real. The California regulator's closing order called the bank insolvent.
- Signature Bank, New York (about $110 billion of assets), had a large base of uninsured deposits from crypto-asset firms. It was closed on 12 March 2023. That day the US authorities invoked a systemic risk exception to protect all deposits at SVB and Signature, including uninsured ones, and the Fed opened a Bank Term Funding Program, lending against government bonds at par value rather than market value.
- Credit Suisse, a global systemically important bank weakened by years of scandals and losses, faced heavy outflows. After emergency liquidity from the Swiss National Bank, it was taken over by UBS on 19 March 2023 for about CHF 3 billion. FINMA ordered about CHF 16 billion of AT1 bonds written off completely, while shareholders still received UBS shares. That reversal of the normal order (shareholders should lose before bondholders) shook the global AT1 market and is the subject of the October 2025 court ruling.
The common lesson: liquidity rules had assumed deposits leave over weeks; they can now leave in hours. Unrealised bond losses and the share of uninsured deposits are warning signs, not footnotes.
Indian cases
India has not let a scheduled commercial bank fail in a way that cost depositors money for decades. Its method has been to merge a failing bank into a stronger one, usually after a short moratorium. Co-operative banks are the exception: they do close, and their depositors depend on DICGC.
Madhavpura Mercantile Co-operative Bank, 2001
This Ahmedabad urban co-operative bank had lent heavily to stockbroking firms linked to Ketan Parekh. When his share positions collapsed in March 2001, rumours of the bank's exposure caused a run. A restructuring scheme made little progress over a decade; the RBI cancelled its licence with effect from 4 June 2012, saying it had ceased to be solvent. Lesson: concentrated lending to one group (here, to the stock market) and weak co-operative governance.
Global Trust Bank, 2004
GTB was a new-generation private bank from Hyderabad. It had large exposures to the stock market and its losses had been under-reported, which left it with a negative net worth; it could neither raise capital nor find a voluntary merger partner. The government placed it under a moratorium from 24 July 2004, and the RBI's scheme merged it with the public sector Oriental Bank of Commerce with effect from 14 August 2004. Its roughly nine lakh depositors were protected; shareholders lost. GTB became the template: moratorium, then amalgamation within weeks.
PMC, Yes Bank and Lakshmi Vilas Bank, 2019–2020
These three are covered in detail in Part 15. Read through the liquidity-solvency lens:
- PMC Bank (2019) was insolvent long before it was illiquid: about 73% of its loans had gone to one group, HDIL, hidden in dummy accounts. The RBI's directions of 23 September 2019 stopped a run by capping withdrawals.
- Yes Bank (2020) had a slow-motion run. Deposits fell from ₹2,27,610 crore (March 2019) to ₹1,05,364 crore (March 2020), a 54% fall, as doubts about its asset quality and capital grew. The moratorium of 5 March 2020 was followed by a reconstruction led by SBI, and by the AT1 write-off now struck down by the Supreme Court.
- Lakshmi Vilas Bank (2020) had years of capital erosion and no investor; it was put under moratorium on 17 November 2020 and merged into DBS Bank India on 27 November 2020.
More recently, the RBI superseded the board of New India Co-operative Bank, Mumbai, in February 2025; it was merged into Saraswat Co-operative Bank from 4 August 2025.
The policy toolkit
Lender of last resort
The classic rule comes from Walter Bagehot's Lombard Street (1873): in a panic, a central bank should lend freely, at a penalty rate, against good collateral, to solvent institutions. Freely, so that the panic stops; at a penalty rate, so that banks do not use it in normal times; against good collateral, so that the central bank does not end up rescuing an insolvent bank. The RBI's routine liquidity windows (LAF and MSF, see Part 11) are the first line. It can also give targeted special liquidity, as it did with the ₹50,000 crore special liquidity facility for mutual funds in April 2020. The limit is the Bagehot test itself: lending to an insolvent bank only delays losses and shifts them to the public.
Deposit insurance
Insurance removes the small depositor's reason to run, as Diamond–Dybvig predicts. India's DICGC covers ₹5 lakh per depositor per bank (since 4 February 2020), must pay within 90 days of a bank being placed under directions or a moratorium, and has charged a risk-based premium since 1 April 2026 (details in Part 30). The US limit is $250,000. Two caveats: insurance can encourage moral hazard (banks take more risk because insured depositors do not monitor them), and it does little for uninsured depositors, who were the ones who ran at SVB.
Resolution regimes
A resolution regime lets an authority take over a failing bank and keep its critical functions (deposits, payments) running while losses are imposed on shareholders and creditors, without a long court bankruptcy. The global standard is the Financial Stability Board's Key Attributes of Effective Resolution Regimes (2011). Common tools:
- Purchase and assumption: a healthy bank takes over the deposits and some assets (the usual US method, as with Nano Banc in 2026 and First Republic, sold to JPMorgan Chase, in 2023).
- Bridge bank: a temporary bank run by the authority (Signature Bridge Bank, 2023).
- Bail-in: creditors' claims are written down or converted to equity.
- Liquidation, with insured deposits paid out.
India has no single resolution law for financial firms. Banks are handled under the Banking Regulation Act, 1949: under Section 45 the RBI can apply to the government for a moratorium and prepare a scheme of reconstruction or amalgamation (GTB, Yes Bank, LVB). The IBC does not apply to banks (Part 26).
The FRDI Bill, 2017: why it was withdrawn
The Financial Resolution and Deposit Insurance (FRDI) Bill was introduced in the Lok Sabha on 10 August 2017 and referred to a Joint Committee of Parliament. It would have created a Resolution Corporation to classify financial firms by risk, take over failing ones and resolve them through mergers, transfers, bail-in or liquidation, and it would have absorbed the DICGC.
The trouble was the bail-in clause. Critics read it as allowing deposits to be written down or converted to other instruments when a bank failed. The bill also removed the fixed ₹1 lakh cover written into the DICGC Act and let the Resolution Corporation set the insured amount. The government argued that deposits up to the insured limit would be protected and that bail-in was only one of many tools, but public fear, especially after the Cyprus bail-in of 2013, grew. The government withdrew the bill on 7 August 2018 "for further comprehensive examination and reconsideration". India still lacks the comprehensive resolution regime it was meant to create.
Bail-in vs bail-out
| Bail-out | Bail-in | |
|---|---|---|
| Who pays | Taxpayers, through the government | The bank's own shareholders and creditors |
| How | Capital injection, guarantees, purchase of bad assets | Write-down or conversion of bonds (and, in the extreme, uninsured deposits) into equity |
| Examples | US TARP (2008); recapitalisation of Indian PSBs | Cyprus 2013; Credit Suisse AT1 (2023); Yes Bank AT1 (2020) |
| Main risk | Moral hazard and fiscal cost | Panic among creditors; legal challenges |
In Cyprus (March 2013) uninsured deposits above €100,000 at Bank of Cyprus were converted into shares at a rate of 47.5%, and uninsured depositors at Laiki Bank, which was wound down, were largely wiped out. Since 2008 most countries have tried to move from bail-out to bail-in by making banks issue loss-absorbing bonds such as AT1. The Yes Bank and Credit Suisse court rulings show the condition attached: the write-down must follow the contract and the law exactly.
Early warning signs
Most failures were visible in the data before the run. Examiners and supervisors look for:
- Very fast balance-sheet growth: SVB tripled in three years; Yes Bank grew its loan book far faster than the system.
- Concentration: one borrower group (PMC and HDIL), one sector (Madhavpura and stockbrokers), or one type of depositor (SVB and start-ups).
- Unstable funding: a high share of uninsured or wholesale deposits, or reliance on short-term market funding (Northern Rock, Lehman).
- Hidden or unrealised losses: divergence between NPAs reported by the bank and assessed by the RBI; large unrealised losses on bond holdings.
- Thin capital: CRAR or CET1 close to the regulatory floor; repeated failure to raise capital (GTB, LVB).
- Governance failures: frequent changes of auditors or top management, regulatory penalties, related-party lending.
- Market signals: a falling share price, rising bond yields or rating downgrades; deposit outflows.
The RBI's formal early-intervention tools are the Prompt Corrective Action framework (triggered by CRAR, CET1, net NPAs and leverage) and supervisory action under the Banking Regulation Act (Part 15, Part 18). The risk-management side, inside the bank, is covered in Part 40.
Common exam traps
- Illiquid is not insolvent. An illiquid bank can be solvent; the lender of last resort is meant for it. An insolvent bank needs resolution, not loans.
- Losses hit equity first, not deposits. Depositors lose only after equity (and, in a bail-in, subordinated and AT1 bonds) is exhausted, and then only above the insured limit.
- Diamond–Dybvig is about self-fulfilling runs. The bank in the model is solvent; the run happens because of beliefs. Bernanke's prize was for the Great Depression research, not the run model.
- The 2022 Nobel went to three people: Bernanke, Diamond and Dybvig.
- SVB's $42 billion left in a single day (9 March 2023); the over $100 billion was expected for 10 March, when it was closed.
- The FRDI Bill was withdrawn (7 August 2018). It was never passed. Do not confuse it with the FSDC (2010) or the IBC (2016).
- GTB merged with Oriental Bank of Commerce (2004), not with a private bank. LVB went to DBS Bank India, PMC to Unity SFB.
- The IBC does not apply to banks. Indian banks are resolved under the Banking Regulation Act.
- Bagehot's rule is "lend freely, at a penalty rate, against good collateral", not "lend cheaply to anyone".
Key facts at a glance
| Liquidity | Ability to meet payments as they fall due |
| Solvency | Assets worth more than liabilities (positive equity) |
| Maturity transformation | Borrowing short (deposits), lending long (loans) |
| Diamond–Dybvig model | 1983; self-fulfilling bank runs; cures: deposit insurance, LOLR |
| Nobel 2022 | Bernanke, Diamond, Dybvig, "for research on banks and financial crises" |
| Bagehot's dictum | Lombard Street (1873): lend freely, at a penalty rate, against good collateral |
| Northern Rock | Run from 14 Sept 2007; nationalised 22 Feb 2008 |
| Lehman Brothers | Bankruptcy 15 Sept 2008; over $600 bn of assets |
| SVB | $42 bn withdrawals on 9 March 2023; closed 10 March 2023; about 94% uninsured deposits |
| Signature Bank | Closed 12 March 2023 |
| Credit Suisse | Taken over by UBS, 19 March 2023, about CHF 3 bn; about CHF 16 bn of AT1 written off |
| Global Trust Bank | Moratorium 24 July 2004; merged with OBC 14 Aug 2004 |
| Madhavpura Mercantile Co-op Bank | Run March 2001 (Ketan Parekh); licence cancelled 4 June 2012 |
| Yes Bank | Moratorium 5 March 2020; AT1 write-off of ₹8,415 crore struck down (SC, 15 Oct 2026) |
| FRDI Bill | Introduced 10 Aug 2017; withdrawn 7 Aug 2018 (bail-in fears) |
| Deposit insurance | ₹5 lakh per depositor per bank (DICGC); US $250,000 |
| Digital deposits in LCR | Extra 2.5% run-off from 1 April 2026 |
Exam corner: what to remember
- Liquidity = cash to pay today; solvency = assets above liabilities. The LOLR cures the first, resolution handles the second.
- Losses come out of equity; leverage magnifies them. With 8% equity, an 8% fall in asset values wipes the bank out.
- Maturity transformation makes banks useful and fragile at the same time.
- Diamond–Dybvig (1983): two equilibria; runs are self-fulfilling; deposit insurance and LOLR remove the run equilibrium. Nobel 2022 with Bernanke.
- Contagion channels: direct exposures, fire sales, information, funding freezes.
- SVB: 8 March loss announcement; $42 bn out on 9 March; closed 10 March 2023; ~94% uninsured; ~$15 bn unrealised HTM losses.
- Credit Suisse: UBS, 19 March 2023, CHF 3 bn; AT1 write-off ruled unlawful by a Swiss court (October 2025, under appeal).
- India's method: moratorium plus amalgamation (GTB 2004 with OBC; LVB 2020 with DBS); reconstruction (Yes Bank 2020).
- FRDI Bill: introduced 10 August 2017, withdrawn 7 August 2018 over the bail-in clause.
- Yes Bank AT1: ₹8,415 crore write-off; Supreme Court upheld its quashing on 15 October 2026.
- Early warnings: fast growth, concentration, unstable funding, hidden losses, thin capital, poor governance, market signals.
Test yourself
A bank whose assets are worth more than its liabilities but which cannot meet a sudden surge of withdrawals is:
Show answer
Answer: (b) Positive net worth means it is solvent; lacking cash to meet withdrawals means it is illiquid, the case a lender of last resort is meant for.
The 2022 Nobel Memorial Prize in Economic Sciences, for research on banks and financial crises, was awarded to:
Show answer
Answer: (a) Ben Bernanke, Douglas Diamond and Philip Dybvig shared the 2022 prize.
A bank has assets of ₹500 crore and equity of ₹40 crore. It suffers loan losses of ₹25 crore. Its equity-to-assets ratio after the loss is about:
Show answer
Answer: (c) Equity = 40 − 25 = ₹15 crore; assets = 500 − 25 = ₹475 crore; 15 ÷ 475 = 3.16%, about 3.2%.
In the Diamond–Dybvig model, a bank run is best described as:
Show answer
Answer: (b) The bank in the model is sound; depositors run because they expect others to run, and the expectation makes itself true.
In March 2023, Silicon Valley Bank faced deposit withdrawals of about $42 billion on a single day. That day was:
Show answer
Answer: (c) Withdrawals of $42 billion came on 9 March; the bank was closed on 10 March, and Signature Bank on 12 March.
The Financial Resolution and Deposit Insurance (FRDI) Bill, 2017 was withdrawn mainly because of concern over its provision on:
Show answer
Answer: (b) Fears that the bail-in clause could be used against deposits led to its withdrawal on 7 August 2018.
Global Trust Bank was amalgamated in August 2004 with:
Show answer
Answer: (c) After a moratorium from 24 July 2004, GTB was merged into Oriental Bank of Commerce with effect from 14 August 2004.
A bank holds ₹40 crore of bonds with a modified duration of 5. If yields rise by 1.5 percentage points, the approximate loss on the bonds is:
Show answer
Answer: (d) Price fall ≈ 5 × 1.5% = 7.5%; 7.5% × ₹40 crore = ₹3.0 crore.
In October 2026, the Supreme Court upheld the quashing of the write-off of Additional Tier 1 bonds worth ₹8,415 crore issued by:
Show answer
Answer: (d) On 15 October 2026 the Court upheld the Bombay High Court's 2023 order setting aside Yes Bank's March 2020 AT1 write-off.
From 1 April 2026, the RBI's Liquidity Coverage Ratio rules require banks to assume an additional run-off on internet and mobile banking-enabled retail deposits of:
Show answer
Answer: (a) The final rules of 21 April 2025 set an extra 2.5% run-off, half the 5% proposed in the July 2024 draft.
Further reading
- Federal Reserve, Review of the Federal Reserve's Supervision and Regulation of Silicon Valley Bank (April 2023): the clearest official account of a modern digital bank run, with the balance-sheet data.
- Royal Swedish Academy of Sciences, Scientific background to the 2022 economics prize: a readable explanation of the Diamond–Dybvig model and Bernanke's Depression research.
- Financial Stability Board, Key Attributes of Effective Resolution Regimes for Financial Institutions: the global standard that the FRDI Bill tried to bring to India.
- RBI, Financial Stability Report (latest and archive): the liquidity and solvency stress tests that the RBI runs on Indian banks every six months.
- RBI, Basel III liquidity standards: revised LCR instructions, 21 April 2025: how India adjusted its liquidity rules for digital deposits.
- National Audit Office, HM Treasury: The nationalisation of Northern Rock (2009): a step-by-step account of a run, a guarantee and a nationalisation.
- Walter Bagehot, Lombard Street: A Description of the Money Market (1873): the original statement of the lender-of-last-resort rule, still quoted by every central bank.
Sources
- Federal Reserve, Review of SVB supervision, 28 April 2023: 8 March announcement ($21 bn sale, $1.8 bn loss, $2.25 bn raise); over $40 bn outflow on 9 March and over $100 bn expected on 10 March; 85% of deposits; Wachovia and WaMu comparisons; ~94% uninsured; assets $209 bn (bank) and growth from $71 bn to over $211 bn, 2019–2021; closure by the California regulator (accessed 24 October 2026)
- Federal Reserve, Barr testimony, 28 March 2023: systemic risk exception for SVB and Signature on 12 March; BTFP lending at par (accessed 24 October 2026)
- California DFPI order, 10 March 2023 and Fortune, 11 March 2023: $42 bn initiated withdrawals on 9 March; negative cash balance of about $958 million; bank insolvent (accessed 24 October 2026)
- SEC, SVB Financial Group Form 10-K for 2022 and ICAEW, March 2023: unrealised HTM losses of about $15 bn against equity of about $16 bn (accessed 24 October 2026)
- FDIC press release, 12 March 2023 and Boston Globe, 12 March 2023: Signature Bank closed and transferred to a bridge bank; about $110 bn of assets (accessed 24 October 2026)
- PBS NewsHour (AP), 1 May 2023: First Republic seized and sold to JPMorgan Chase (accessed 24 October 2026)
- CNBC, 23 March 2023 and Investment Week, March 2023: UBS takeover of Credit Suisse for about CHF 3 bn; AT1 write-down of about CHF 16 bn (accessed 24 October 2026)
- Swissinfo, October 2025 and Jurist, October 2025: Federal Administrative Court partial decision, published 14 October 2025; FINMA appeal (accessed 24 October 2026)
- Business Today, 15 October 2026 and Outlook Business, 15 October 2026: Supreme Court upheld the quashing of the ₹8,415 crore AT1 write-off; shares fell 14.5% (accessed 24 October 2026)
- Morung Express (IANS), 7 May 2020: Yes Bank deposits of ₹2,27,610 crore (March 2019) and ₹1,05,364 crore (March 2020) (accessed 24 October 2026)
- ABA Banking Journal, 26 September 2026: Nano Banc closure, assets $736 million, Sunwest Bank, DIF cost about $114 million (accessed 24 October 2026)
- RBI press release, 21 April 2025 and Business Standard, 21 April 2025: extra 2.5% run-off on IMB-enabled deposits from 1 April 2026; 5% in the July 2024 draft (accessed 24 October 2026)
- RBI press release, 6 February 2026: risk-based deposit insurance premium from 1 April 2026 (accessed 24 October 2026)
- RBI press release, 13 February 2026: licence of Indian Mercantile Co-operative Bank, Lucknow cancelled; 98.75% of depositors fully covered (accessed 24 October 2026)
- CNBC, 10 October 2022 and NBER, 2022: 2022 Nobel to Bernanke, Diamond and Dybvig (accessed 24 October 2026)
- Wikipedia, Nationalisation of Northern Rock and National Audit Office, 2009: BoE support on 14 September 2007; £1 bn withdrawn on the first day; guarantee of 17 September; public ownership from 22 February 2008 (accessed 24 October 2026)
- Wikipedia, Bankruptcy of Lehman Brothers: filing on 15 September 2008; over $600 bn of assets; largest US bankruptcy (accessed 24 October 2026)
- Federal Reserve press release, 16 September 2008 and ICI, money market funds in 2008: AIG loan of up to $85 bn; Reserve Primary Fund broke the buck (accessed 24 October 2026)
- RBI press release on Global Trust Bank, 2004 and RBI notice, scheme of amalgamation: moratorium from 24 July 2004; merger with OBC from 14 August 2004; nine lakh depositors (accessed 24 October 2026)
- RBI press release, 2012 and Moneylife, June 2012: Madhavpura licence cancelled from 4 June 2012; run of March 2001; Ketan Parekh exposure (accessed 24 October 2026)
- Drishti IAS, 19 July 2018 and Udayavani, August 2018: FRDI Bill introduced 10 August 2017, referred to a JPC, withdrawn 7 August 2018; bail-in and deposit insurance concerns (accessed 24 October 2026)
- CEPR VoxEU, Banking crisis, bail-ins and money holdings and Cyprus Mail archive: Bank of Cyprus 47.5% conversion of uninsured deposits; Laiki uninsured depositors (accessed 24 October 2026)
- Business Standard, August 2025 and Moneylife, 2025: New India Co-operative Bank board superseded February 2025; merged with Saraswat Bank from 4 August 2025 (accessed 24 October 2026)