Skip to main content

How do banks fail and what causes financial system crises?

Bank failures are the ultimate test of banking system health and central bank competence. When solvent-looking banks suddenly collapse, panic spreads globally, depositors withdraw funds from other institutions, and the financial cascade can trigger severe recessions or depressions. Bank failures reveal the leverage inherent in banking: banks operate with thin capital buffers (equity layers) backing large asset bases, meaning small changes in asset values wipe out equity and trigger insolvency. This article examines how banks fail through detailed case studies: the 2008 financial crisis (Lehman Brothers, Washington Mutual, over 500 regional banks), and the 2023 Silicon Valley Bank failure. Understanding how banks fail requires examining the anatomy of credit cycles (boom-and-bust patterns in lending), the role of leverage (how small capital cushions amplify losses when assets decline), the critical distinction between illiquidity (lacking liquid cash) and insolvency (liabilities exceeding assets), the cascading effects where one bank's failure triggers contagion at others, and the regulatory failures that precede crises. Bank failures are rarely surprising to economists studying leverage and credit cycles—they're predictable consequences of excess borrowing and relaxed lending standards during booms. However, preventing failures requires either restricting leverage (which slows growth), improving regulatory oversight (which is difficult and political), or accepting periodic failures as capitalism's cost and providing deposit insurance, emergency central bank lending, and crisis resolution mechanisms to contain damage.

Quick definition: A bank fails when liabilities exceed assets (insolvency) or when the bank cannot meet withdrawal demands with available liquid assets (illiquidity converted to insolvency). The 2008 crisis saw major banks fail due to mortgage losses; SVB 2023 failed from asset devaluation combined with deposit runs. Bank failures cascade through the financial system if confidence evaporates.

Key takeaways​

  • Banks fail when leverage becomes unsustainable: Small declines in asset value wipe out thin equity buffers, triggering insolvency
  • The 2008 financial crisis saw over 500 bank failures, including major institutions like Lehman Brothers ($619 billion assets) and Washington Mutual ($307 billion assets, the largest U.S. bank failure by asset size)
  • Lehman Brothers failed in September 2008 after losing over $50 billion when mortgage-backed securities deteriorated and wholesale funding evaporated
  • Washington Mutual failed in September 2008 with $77 billion in subprime mortgages that deteriorated, triggering deposit runs and insolvency
  • Silicon Valley Bank failed in March 2023 within 48 hours of deposit runs after its $100 billion bond portfolio lost ~$15 billion in value due to rising interest rates
  • Cascade effects are critical: Bank failures trigger uncertainty about other institutions, depositors panic and withdraw simultaneously, credit contracts, and recessions deepen
  • Regulatory failures precede crises: Regulators failed to detect or restrict the obvious risks in 2008 (subprime lending to unqualified borrowers) and 2023 (concentrated deposits and interest rate risk)

The 2008 Financial Crisis: How Leverage and Mortgage Losses Triggered Systemic Failure​

The Boom Years: Housing Inflation and Subprime Origination (2003–2007)

The U.S. housing market experienced extraordinary inflation from 2000 to 2006. Home prices doubled, from a national median of ~$160,000 (2000) to ~$320,000 (2006). Banks, seeing the profitable housing market and confident that prices would perpetually rise, originated unprecedented volumes of mortgages. Critically, they originated billions in subprime mortgages—loans to borrowers with poor credit, minimal down payments, and often no income verification.

Characteristics of Subprime Mortgages (2005–2007):

  • Loan-to-value (LTV) ratio: 95% or higher, meaning borrowers put down only 5% and financed 95% of home value
  • Borrower credit quality: FICO scores of 580–650 (poor credit; scores of 750+ are considered good)
  • Interest rates: Initial teaser rates of 2–3%, adjustable and resetting to 6–7%+ after 2 years
  • Income verification: Often absent—"NINJA" loans (No Income, No Job, No Assets)
  • Underwriting standards: Minimal; lenders assumed perpetual house price appreciation meant losses would be impossible

The implicit assumption: if borrowers couldn't repay, lenders would foreclose and sell homes at higher prices, covering losses. This assumption failed catastrophically.

Scale of Subprime Lending:

By 2007, lenders originated $1.2 trillion in subprime mortgages annually. The total subprime mortgage portfolio in U.S. financial system exceeded $2 trillion. This represented roughly 20% of all mortgages in America.

The Banking Connection and Incentive Misalignment:

Banks originated subprime mortgages but often sold them immediately to investors (investment banks, pension funds, government-sponsored enterprises like Fannie Mae and Freddie Mac). Lenders earned origination fees (1–2% of loan value) and had minimal skin in the game—they didn't bear the default risk.

This broke the fundamental incentive in lending: originators should bear some risk to ensure careful underwriting. Instead, originators earned fees for volume, not quality. The mortgage was sold, the originator was paid, the risk transferred to someone else.

However, some banks and investment banks retained substantial mortgage portfolios:

  • Washington Mutual: Held $77 billion in subprime mortgages (roughly 25% of their loan portfolio)
  • Countrywide Financial: Originated over $100 billion in subprime mortgages annually (later acquired by Bank of America to contain losses)
  • Lehman Brothers: Held massive mortgage-backed securities portfolios on its balance sheet
  • Bear Stearns: $350 billion in total assets, heavily laden with mortgage exposures

These institutions bore the default risk directly.

The Housing Crash (2006–2008):

Home prices peaked in 2006 and began declining in 2007. By 2009, the national median home price had fallen below $200,000—a 30%+ decline from peak.

Simultaneously, initial mortgage teaser rates reset to higher levels. Borrowers who took out 2% mortgages in 2004–2005 faced rates jumping to 6%+ in 2006–2007. Monthly payments doubled or tripled.

Borrowers found themselves underwater (owing more than homes were worth) and facing unaffordable payments. Many walked away, defaulting on mortgages.

Default Rate Escalation:

Lenders had expected 1–2% annual default rates. Actual default rates surged to 5%, 10%, and beyond. Loan losses were catastrophic.

Example calculation: A lender with $100 billion in mortgages expecting 1% default ($1 billion loss) but experiencing 8% default ($8 billion loss) faces a $7 billion loss exceeding expectations. If the lender's capital was only $5 billion, the institution becomes insolvent.

Lehman Brothers Collapse (September 2008):

Lehman Brothers, a 158-year-old investment bank with $619 billion in assets, faced massive mortgage losses as the housing crisis deepened:

Timeline of collapse:

  • 2007: Lehman reports profits; mortgage losses accumulate
  • Mid-2008: Mortgage losses accelerate. Lehman's reported equity declines from $28 billion to $15 billion
  • August 2008: Bear Stearns (similar size) is acquired by JPMorgan with Fed support. Market participants realize mega-institutions could fail
  • September 10, 2008: Lehman stock trades at $3.65 (down from $65 just one year prior). Management attempts to find buyers
  • September 15, 2008: No buyers emerge. Lehman files for bankruptcy—the largest U.S. corporate bankruptcy in history

Lehman's Asset Composition and Losses:

Lehman held approximately:

  • $50 billion in mortgage-backed securities (MBS)
  • $20 billion in mortgages directly
  • $70 billion in real estate and real estate-related investments

As housing prices collapsed, the value of these assets deteriorated rapidly. Mark-to-market accounting (valuing assets at current market prices) showed losses exceeding $50 billion. Lehman's equity buffer (capital) was overwhelmed.

Lehman's Funding Crisis:

As losses accumulated and credibility evaporated:

  1. Wholesale funding dried up—counterparties stopped rolling over short-term loans to Lehman
  2. Lehman couldn't refinance short-term debt (commercial paper, repo agreements)
  3. Without access to funding, Lehman couldn't pay obligations
  4. Lehman exhausted capital trying to stem losses
  5. Lehman was insolvent and illiquid—a failed institution

The Systemic Cascade:

Lehman's bankruptcy shocked the global financial system. If Lehman—a 158-year-old mega-institution—could fail, any bank could fail. Confidence evaporated globally.

Money market funds (which held Lehman commercial paper) suffered losses. The "money market crisis" nearly caused runs on money market funds as investors panicked.

Other investment banks faced funding freezes. Merrill Lynch was forced to sell to Bank of America at distressed prices. Goldman Sachs faced a funding crisis and had to convert to a bank holding company (gaining Fed access).

Credit markets froze. Banks wouldn't lend to each other; wholesale funding markets that had been liquid now traded at massive spreads. Banks that depended on overnight funding couldn't find lenders.

Washington Mutual Failure (September 2008):

Washington Mutual (WaMu), with $307 billion in assets, became the largest U.S. bank failure by asset size:

Background:

  • WaMu was a regional bank based in Seattle, specializing in mortgages
  • During the boom (2003–2007), WaMu expanded aggressively, originating record mortgage volumes
  • WaMu held $77 billion in subprime mortgages on its balance sheet—roughly 25% of assets

The Failure:

  • Early 2008: Mortgage losses mount. WaMu's capital declines
  • July 2008: Regulators rate WaMu as "troubled." WaMu's credit rating is downgraded
  • September 2008: Deposit runs intensify. Depositors withdraw $16.7 billion in a single week
  • September 25, 2008: Regulators close WaMu. The FDIC (Federal Deposit Insurance Corporation) takes control
  • Outcome: WaMu is acquired by JPMorgan at a fire-sale price ($1.88 billion). Uninsured depositors (above $250,000) face significant losses on their deposits

WaMu's Failure Cascade:

WaMu's failure created contagion:

  • Other regional banks holding mortgages faced similar problems
  • Depositors at similar banks panicked, triggering runs
  • Hundreds of regional and community banks failed over 2008–2010

The total was over 500 bank failures from 2008–2010—the worst banking crisis since the Great Depression.

The 2023 Silicon Valley Bank Failure: A Modern Bank Run in the Digital Age​

Background: Growth and Deposits (2019–2022)

Silicon Valley Bank (SVB), based in San Jose, California, specialized in lending to venture capital-backed startups. During the tech boom (2019–2021), venture capital investment surged. Startups raised record funding rounds. Deposits flowed into SVB from venture firms and their portfolio companies.

Deposit Growth:

  • 2019: SVB deposits ~$50 billion
  • 2021: SVB deposits ~$175 billion
  • 2022: SVB deposits ~$191 billion

This 3-year deposit growth of 280% was extraordinary and unsustainable.

Asset Deployment: Interest Rate Assumptions

With massive deposit inflows, SVB needed to deploy capital into investments. SVB invested heavily in securities:

  • U.S. Treasury bonds (government debt)
  • Mortgage-backed securities (mortgage pools)
  • Average yield: 1–2% (appropriate for the near-zero rate environment of 2019–2021)

SVB's portfolio by 2022:

  • ~$100 billion in marketable securities
  • ~$90 billion in loans (mortgages to startups, real estate)

All assumptions were based on near-zero interest rate persistence. If rates remained at 0.25%, the portfolio was fine. If rates rose, losses would emerge.

The Interest Rate Pivot (March 2022 Onward):

The Federal Reserve began raising rates aggressively in March 2022 to fight inflation:

  • March 2022: Fed raises fed funds rate from 0% to 0.25–0.50%
  • June 2022: Rate reaches 1.5–1.75%
  • December 2022: Rate reaches 4.25–4.50%
  • July 2023: Rate reaches 5.25–5.50%

As rates rose, the market value of SVB's bond portfolio plummeted. Bonds yielding 1.5% (purchased in 2020–2021) became worth substantially less when new bonds were yielding 5%.

Mark-to-Market Losses:

Using mark-to-market accounting (valuing securities at current market prices), SVB's bond portfolio had deteriorated significantly:

Example: SVB purchased a 10-year Treasury bond in 2020 for $100 (yielding 1%). Interest rates rise to 5%. A new 10-year Treasury would yield 5%. SVB's bond (yielding 1%) is now worth only $70 (investors would only pay $70 for a bond yielding 1% when they could buy a 5% bond for $100).

The $100 million portfolio SVB purchased in 2019–2021 was worth only ~$85 million by 2023—a $15 billion loss on ~$100 billion in securities.

Regulatory Forbearance:

Critically, SVB's regulators (the Federal Reserve and California Department of Financial Protection and Innovation) allowed SVB to use "hold-to-maturity" accounting. Rather than mark-to-market (which would show losses immediately), SVB could value bonds at cost. As long as SVB didn't sell, losses weren't recognized.

This accounting forbearance allowed SVB to hide the deterioration.

The Bank Run (March 10–12, 2023):

On March 10, SVB disclosed its bond portfolio losses in a regulatory filing. The filing revealed:

  • Bond portfolio value: ~$91 billion book value (cost)
  • Market value: ~$76 billion
  • Unrealized loss: ~$15 billion
  • SVB's equity (capital): ~$10 billion
  • Conclusion: Unrealized losses exceeded capital

Venture capital firms, reading the filing, realized SVB faced problems. If losses continued or interest rates rose further, SVB could become insolvent.

Deposit Run:

  • March 10, 2023: Venture capital firms and startups begin withdrawing deposits. SVB receives $42 billion in withdrawal requests on a single day (March 10)—roughly 25% of total deposits
  • March 11–12, 2023: Withdrawals continue. SVB cannot meet the demand
  • March 12, 2023: Regulators close SVB. The FDIC takes control

The Speed of Modern Bank Runs:

Historically, bank runs took days or weeks—depositors would queue at banks, slowly withdrawing cash. In the digital age, runs happen in hours:

  • Depositors call their banks (or use apps) requesting withdrawals
  • Money is transferred electronically
  • Within hours, a billion dollars can flee
  • Within a day, tens of billions can flee

SVB faced a multi-billion-dollar run in a single business day.

SVB's Insolvency:

Could SVB meet the withdrawals? No. SVB held:

  • $100 billion in bonds (market value ~$85 billion)
  • $90 billion in loans (difficult to liquidate quickly)
  • ~$15 billion in cash and cash equivalents

To meet $42 billion in withdrawals, SVB would need to sell bonds at market prices (~$85 billion value), crystallizing the $15 billion loss. This would eliminate SVB's $10 billion equity buffer and render the bank insolvent.

Without selling bonds, SVB had only ~$15 billion in liquid assets against $42 billion in withdrawal requests. The bank couldn't meet demand.

Resolution:

The FDIC took control of SVB and ultimately allowed it to be acquired by First Citizens BancShares. The FDIC guaranteed all deposits (insured and uninsured) to prevent contagion.

Lessons:

  1. Illiquidity converts to insolvency: SVB was technically insolvent on a marked-to-market basis (unrealized losses exceeded equity), but the crisis was triggered by illiquidity (inability to meet immediate withdrawals)
  2. Digital age accelerates runs: Runs can happen in hours, not days. This limits central bank response time
  3. Interest rate risk: Rising rates expose poor asset-liability management. SVB matched long-term assets (bonds, mortgages) with short-term liabilities (deposits), creating duration risk
  4. Regulatory forbearance: Allowing hold-to-maturity accounting masked the problem until it was too late
  5. Contagion risk: SVB's failure triggered concerns at similar regional banks, nearly causing systemic instability (contained by Fed emergency lending)

Why Regulatory Failures Precede Bank Failures​

Both the 2008 and 2023 crises involved regulatory failures that preceded the financial collapses:

2008 Regulatory Failures:

  • Subprime origination standards: Regulators allowed lending to borrowers with FICO scores of 580 (poor credit), 95% LTV ratios, and no income verification. The riskiness was obvious but unaddressed
  • Inadequate capital requirements: Investment banks held only 3% capital relative to assets, meaning 1% losses would eliminate equity entirely. This excessive leverage was known but accepted
  • Risk model failures: Banks used risk models assuming housing prices couldn't fall nationally (they never had in modern history). The models assigned near-zero probability to the housing collapse scenario that occurred
  • Systemic risk ignored: Regulators didn't restrict the interconnectedness of major institutions, allowing the failure of one to threaten all

Regulators either didn't see the problems or were unwilling to restrict activity (politically and economically difficult) that would slow growth.

2023 Regulatory Failures:

  • Interest rate risk oversight: SVB was examined by regulators but interest rate risk in its bond portfolio wasn't prioritized
  • Accounting forbearance: Regulators allowed SVB to use hold-to-maturity accounting, masking deterioration until it was critical
  • Concentration risk: SVB had concentrated deposits (venture capital firms) and concentrated assets (Treasury bonds). A shock to the venture industry (which occurred in 2022–2023 as VC funding dried up) threatened both funding and the customer base
  • Elevated risk ignored: The U.S. banking system's interest rate risk (many banks held long-duration bonds at low yields) was discussed but downplayed as manageable

Can Bank Failures Be Prevented Entirely?​

Theoretically, yes—with sufficiently restrictive regulation:

Ultra-Conservative Approach:

  • Require 100% capital ratios (no leverage)
  • Restrict lending to only the safest borrowers
  • Require daily mark-to-market of assets
  • Restrict asset concentration

However, this approach would eliminate banking's credit creation function. Growth would slow sharply. Most economies accept some risk of failure as the cost of capitalism.

The Accepted Middle-Ground Approach (Post-2008):

Instead of preventing failures entirely, modern economies:

  • Provide deposit insurance (depositors protected up to $250,000)
  • Offer emergency central bank lending (solvent banks can borrow at penalty rates)
  • Deploy quick resolution (regulators take control of failing banks quickly, resolving them within days)
  • Implement higher capital requirements (banks must hold 10%+ capital)
  • Require stress testing (banks must survive simulated severe crises)

This approach allows banks to fail but prevents systemic collapse:

  • Depositors are protected (don't panic and withdraw)
  • Solvent institutions can borrow from the Fed if needed
  • Failing institutions are resolved quickly (removed from the system before contagion spreads)

The 2023 SVB resolution followed this approach: the FDIC took control, guaranteed deposits, and arranged acquisition within days.

Real-World Examples: Bank Failure Cascades and Responses​

The Great Depression (1929–1933): Uncontrolled Cascade

Bank failures cascaded without government intervention. Over 9,000 banks failed. No deposit insurance, no central bank emergency lending.

Contagion was severe: depositors at Bank A heard that Bank B failed, so they withdrew from Bank A (fear of contagion). Bank A, facing runs, failed. This validated fears, triggering runs at Bank C, D, E. The cascade continued until the banking system was largely destroyed.

The 2008 Crisis: Emergency Intervention

The Federal Reserve and Treasury deployed emergency measures:

  • Fed lowered the discount rate and lent to banks through the discount window
  • Fed created the "Term Auction Facility" (TAF), allowing banks to borrow at good rates
  • Fed purchased commercial paper (short-term corporate borrowing), stabilizing funding markets
  • Treasury injected capital into banks (TARP, Troubled Asset Relief Program)
  • FDIC expanded deposit insurance temporarily to $250,000 (from $100,000)

These interventions prevented the 2008 cascade from becoming a Great Depression-level disaster. Banking system stabilized by late 2008, though the recession persisted through 2009.

The 2023 SVB Crisis: Swift Resolution

The Fed responded immediately:

  • Announced the "Bank Term Funding Program" (BTFP), allowing banks to borrow against Treasury bonds at par (face value)
  • FDIC guaranteed all SVB deposits (even uninsured amounts) to prevent contagion
  • Within days, SVB was acquired and the crisis was contained

This swift response prevented the spread to other regional banks that faced similar (but less acute) problems.

Common Mistakes About Bank Failures​

Mistake 1: Assuming Bank Failures Are Unpredictable

Bank failures often appear sudden, but precursors are visible. 2008: subprime lending to unqualified borrowers was obvious. 2023: SVB's interest rate risk was knowable.

Failures are surprising to markets (which maintain irrational confidence until they don't), but predictable to careful analysts studying leverage and credit risk.

Mistake 2: Believing All Banks Are Equal Risk

Some banks take more risk than others. Pre-2008, highly leveraged investment banks (Lehman, Bear Stearns) took more risk than conservative, well-capitalized banks.

Pre-2023, SVB's concentrated deposits and long-duration assets were riskier than diversified banks with shorter-duration assets.

Risk varies; failures concentrate among higher-risk institutions.

Mistake 3: Assuming Bank Failures Are Isolated Events

Bank failures trigger contagion:

  • Depositors panic at similar institutions
  • Wholesale funding dries up for all banks
  • Fire-sale asset liquidations accelerate
  • Real economy credit contracts

A "regional bank failure" can cascade into systemic instability if the Fed doesn't intervene.

FAQ: Bank Failures Explained​

Q: Why didn't regulators see the 2008 crisis coming?

A: Some did. However, many ignored warning signs because:

  • Models suggested housing couldn't fail nationally
  • Regulators were politically pressured to not restrict growth
  • The financial industry has regulatory capture (influence over regulators)
  • Risk management models were flawed

In retrospect, the warning signs were obvious (subprime to unqualified borrowers). Regulators chose (or were unable to) restrict the risky activity.

Q: Could the 2008 crisis have been prevented?

A: Yes. Tighter lending standards (no NINJA loans, higher FICO requirements), higher capital requirements (preventing leverage to 30:1 ratios), and oversight would have prevented the crisis.

The costs: slower housing appreciation, lower homeownership rates, smaller profits for lenders. Politically unpalatable, but economically preventive.

Q: Will banking crises happen again?

A: Almost certainly. As long as banks operate with leverage (deposits backing loans), credit standards relax during good times, and asset bubbles form, crises will recur.

However, post-2008 reforms (higher capital, stress testing, prompt resolution) should reduce severity. Complete prevention requires eliminating banking leverage, impractical for modern economies.

Q: Could a bank failure cause a depression?

A: Yes. If a large bank fails and contagion spreads (deposit runs at other banks, wholesale funding seizures), credit markets freeze and the real economy contracts severely. The Great Depression illustrated this scenario.

However, post-2008 reforms (deposit insurance, emergency central bank lending, quick resolution) reduce but don't eliminate this risk.

Q: Why do depositors panic and run on banks?

A: Rational information asymmetry: depositors know banks take risks but don't know exact exposures. When one bank fails, depositors assume others face similar risks and withdraw preemptively. Even safe banks face runs if panic is severe.

Deposit insurance reduces panic by protecting deposits.

Q: Could SVB's failure have been prevented?

A: Yes. Better interest rate risk management (shorter-duration bond portfolio), higher capital buffers, and regulatory forbearance removal would have prevented failure.

The costs: lower yields in the near-zero rate environment, lower profitability. SVB chose to take interest rate risk to boost returns. When rates rose, the risk materialized.

Summary​

Banks fail when liabilities exceed assets (insolvency) or when depositors lose confidence simultaneously and withdraw funds faster than the bank can provide (illiquidity). The 2008 financial crisis saw over 500 bank failures caused by subprime mortgage losses and subsequent defaults. Lehman Brothers ($619 billion assets) and Washington Mutual ($307 billion assets) were among the largest failures.

Silicon Valley Bank's March 2023 failure demonstrates how digital banking accelerates runs: SVB faced $42 billion in withdrawals in a single day, far exceeding its liquid assets. The bank's unrealized bond losses ($15 billion) exceeded its capital buffer ($10 billion), rendering it insolvent.

Both crises involved regulatory failures preceding the crisis: 2008 involved tolerating reckless subprime lending, and 2023 involved allowing interest rate risk to accumulate without adequate capital buffers.

Preventing failures entirely is impractical without eliminating banking's credit creation function. Modern economies accept failure risk while providing deposit insurance, emergency central bank lending, and quick resolution mechanisms to contain damage. These post-2008 reforms reduced crisis severity in 2023.

Bank failures are largely predictable to careful observers of leverage and credit cycles. The surprise is not whether crises will occur, but when and how severely central banks and regulators will respond.

Next chapter​

→ Interest rates 101